class RunStarDist(ImageSegmentation):
category = "Object Processing"
module_name = "RunStarDist"
variable_revision_number = 1
def create_settings(self):
super(RunStarDist, self).create_settings()
self.model = Choice(
text="Model",
choices=MODEL_OPTIONS,
value=GREY_1,
doc="""\
StarDist comes with models for detecting nuclei. Alternatively, you can supply a custom-trained model
generated outside of CellProfiler within Python. Custom models can be useful if working with unusual cell types.
The inbuilt fluorescent and DSB models expect greyscale images. The H&E model expects a color image as input (from
brightfield). Custom models will require images of the type they were trained with. It should be noted that the
models supplied with StarDist do not support 3D images, but it's possible to train and use your own.
""",
)
self.tile_image = Binary(
text="Tile input image?",
value=False,
doc="""\
If enabled, the input image will be broken down into overlapping tiles.
This can help to conserve memory when working with large images.
The image is split into a set number of vertical and horizontal tiles.
The total number of tiles will be the result of multiplying the horizontal
and vertical tile number.""",
)
self.n_tiles_x = Integer(text="Horizontal tiles",
value=1,
minval=1,
doc="""\
Specify the number of tiles to break the image down into along the x-axis (horizontal)."""
)
self.n_tiles_y = Integer(text="Vertical tiles",
value=1,
minval=1,
doc="""\
Specify the number of tiles to break the image down into along the y-axis (vertical)."""
)
self.save_probabilities = Binary(
text="Save probability image?",
value=False,
doc="""
If enabled, the probability scores from the model will be recorded as a new image.
Probability scales from 0-1, with 1 representing absolute certainty of a pixel being in a cell.
You may want to use a custom threshold to manually generate objects.""",
)
self.probabilities_name = ImageName(
"Name the probability image",
"Probabilities",
doc=
"Enter the name you want to call the probability image produced by this module.",
)
self.model_directory = Directory("Model folder",
doc=f"""\
*(Used only when using a custom pre-trained model)*
Select the folder containing your StarDist model. This should have the config, threshold and weights files
exported after training.""")
self.gpu_test = DoSomething(
"",
"Test GPU",
self.do_check_gpu,
doc=f"""\
Press this button to check whether a GPU is correctly configured.
If you have a dedicated GPU, a failed test usually means that either your GPU does not support deep learning or the
required dependencies are not installed.
Make sure you followed the setup instructions here: https://www.tensorflow.org/install/gpu
If you don't have a GPU or it's not configured, StarDist will instead run on the CPU.
This will be slower but should work on any system.
""",
)
def settings(self):
return [
self.x_name,
self.model,
self.y_name,
self.tile_image,
self.n_tiles_x,
self.n_tiles_y,
self.save_probabilities,
self.probabilities_name,
self.model_directory,
]
def visible_settings(self):
vis_settings = [self.x_name, self.model]
if self.model.value == MODEL_CUSTOM:
vis_settings += [self.model_directory]
vis_settings += [self.y_name, self.save_probabilities]
if self.save_probabilities.value:
vis_settings += [self.probabilities_name]
vis_settings += [self.tile_image]
if self.tile_image.value:
vis_settings += [self.n_tiles_x, self.n_tiles_y]
vis_settings += [self.gpu_test]
return vis_settings
def run(self, workspace):
images = workspace.image_set
x = images.get_image(self.x_name.value)
dimensions = x.dimensions
x_data = x.pixel_data
# Validate some settings
if self.model.value in (GREY_1, GREY_2) and x.multichannel:
raise ValueError(
"Color images are not supported by this model. Please provide greyscale images."
)
elif self.model.value == COLOR_1 and not x.multichannel:
raise ValueError(
"Greyscale images are not supported by this model. Please provide a color overlay."
)
if self.model.value != MODEL_CUSTOM:
if x.volumetric:
raise NotImplementedError(
"StarDist's inbuilt models do not currently support 3D images"
)
model = StarDist2D.from_pretrained(self.model.value)
else:
model_directory, model_name = os.path.split(
self.model_directory.get_absolute_path())
if x.volumetric:
from stardist.models import StarDist3D
model = StarDist3D(config=None,
basedir=model_directory,
name=model_name)
else:
model = StarDist2D(config=None,
basedir=model_directory,
name=model_name)
tiles = None
if self.tile_image.value:
tiles = []
if x.volumetric:
tiles += [1]
tiles += [self.n_tiles_x.value, self.n_tiles_y.value]
# Handle colour channels
tiles += [1] * max(0, x.pixel_data.ndim - len(tiles))
print(x.pixel_data.shape, x.pixel_data.ndim, tiles)
if not self.save_probabilities.value:
# Probabilities aren't wanted, things are simple
data = model.predict_instances(normalize(x.pixel_data),
return_predict=False,
n_tiles=tiles)
y_data = data[0]
else:
data, probs = model.predict_instances(normalize(x.pixel_data),
return_predict=True,
sparse=False,
n_tiles=tiles)
y_data = data[0]
# Scores aren't at the same resolution as the input image.
# We need to slightly resize to match the original image.
size_corrected = resize(probs[0], y_data.shape)
prob_image = Image(
size_corrected,
parent_image=x.parent_image,
convert=False,
dimensions=len(size_corrected.shape),
)
workspace.image_set.add(self.probabilities_name.value, prob_image)
if self.show_window:
workspace.display_data.probabilities = size_corrected
y = Objects()
y.segmented = y_data
y.parent_image = x.parent_image
objects = workspace.object_set
objects.add_objects(y, self.y_name.value)
self.add_measurements(workspace)
if self.show_window:
workspace.display_data.x_data = x_data
workspace.display_data.y_data = y_data
workspace.display_data.dimensions = dimensions
def display(self, workspace, figure):
if self.save_probabilities.value:
layout = (2, 2)
else:
layout = (2, 1)
figure.set_subplots(dimensions=workspace.display_data.dimensions,
subplots=layout)
figure.subplot_imshow(
colormap="gray",
image=workspace.display_data.x_data,
title="Input Image",
x=0,
y=0,
)
figure.subplot_imshow_labels(
image=workspace.display_data.y_data,
sharexy=figure.subplot(0, 0),
title=self.y_name.value,
x=1,
y=0,
)
if self.save_probabilities.value:
figure.subplot_imshow(
colormap="gray",
image=workspace.display_data.probabilities,
sharexy=figure.subplot(0, 0),
title=self.probabilities_name.value,
x=0,
y=1,
)
def do_check_gpu(self):
import tensorflow
if len(tensorflow.config.list_physical_devices('GPU')) > 0:
message = "GPU appears to be working correctly!"
print("GPUs:", tensorflow.config.list_physical_devices('GPU'))
else:
message = "GPU test failed. There may be something wrong with your configuration."
import wx
wx.MessageBox(message, caption="GPU Test")
class ExportToACC(Module):
module_name = 'ExportToACC'
category = ["File Processing", "Data Tools"]
variable_revision_number = 2
def create_settings(self):
self.directory = Directory("Output file location",
dir_choices=[
ABSOLUTE_FOLDER_NAME,
DEFAULT_OUTPUT_FOLDER_NAME,
DEFAULT_OUTPUT_SUBFOLDER_NAME,
DEFAULT_INPUT_FOLDER_NAME,
DEFAULT_INPUT_SUBFOLDER_NAME
],
doc="""
This setting lets you choose the folder for the output
files. %(IO_FOLDER_CHOICE_HELP_TEXT)s
%(IO_WITH_METADATA_HELP_TEXT)s
""" % globals())
self.directory.dir_choice = DEFAULT_OUTPUT_FOLDER_NAME
self.wants_file_name_suffix = Binary(
"Append a suffix to the file name?",
False,
doc="""
Select *"YES"* to add a suffix to the file name.
Select *"NO"* to use the file name as-is.
""" % globals())
self.file_name_suffix = Text("Text to append to the file name",
"",
metadata=True,
doc="""
"*(Used only when constructing the filename from the image filename)*"
Enter the text that should be appended to the filename specified above.
""")
self.wants_overwrite_without_warning = Binary(
"Overwrite without warning?",
False,
doc="""This setting either prevents or allows overwriting of
old .txt files by **ExportToACC** without confirmation.
Select "*YES*" to overwrite without warning any .txt file
that already exists. Select "*NO*" to prompt before overwriting
when running CellProfiler in the GUI and to fail when running
headless.
""" % globals())
self.nan_representation = Choice("Representation of Nan/Inf",
[NANS_AS_NANS, NANS_AS_NULLS],
doc="""
This setting controls the output for numeric fields
if the calculated value is infinite (*"Inf"*) or undefined (*"NaN*").
CellProfiler will produce Inf or NaN values under certain rare
circumstances, for instance when calculating the mean intensity
of an object within a masked region of an image.
- "*%(NANS_AS_NULLS)s:*" Output these values as empty fields.
- "*%(NANS_AS_NANS)s:*" Output them as the strings "NaN", "Inf" or "-Inf".
""" % globals())
self.pick_columns = Binary("Select the measurements to export",
False,
doc="""
Select *"YES"* to provide a button that allows you to select which measurements you want to export.
This is useful if you know exactly what measurements you want included in the final spreadheet(s). """
% globals())
self.columns = MeasurementMultiChoice(
"Press button to select measurements to export",
doc="""
"*(Used only when selecting the columns of measurements to export)*"
This setting controls the columns to be exported. Press
the button and check the measurements or categories to export.""")
self.file_image_name = FileImageSubscriber(
"Select image name for file prefix",
"None",
doc="""
Select an image loaded using **NamesAndTypes**. The original filename will be
used as the prefix for the output filename.""" % globals())
def settings(self):
"""Return the settings in the order used when storing """
result = [
self.pick_columns, self.directory, self.columns,
self.nan_representation, self.wants_file_name_suffix,
self.file_name_suffix, self.wants_overwrite_without_warning,
self.file_image_name
]
return result
def visible_settings(self):
"""Return the settings as seen by the user"""
result = [
self.directory, self.file_image_name, self.wants_file_name_suffix
]
if self.wants_file_name_suffix:
result += [self.file_name_suffix]
result += [
self.wants_overwrite_without_warning, self.nan_representation,
self.pick_columns
]
if self.pick_columns:
result += [self.columns]
return result
def validate_module(self, pipeline):
'''Test the module settings to make sure they are internally consistent'''
'''Make sure metadata tags exist'''
if self.wants_file_name_suffix.value:
text_str = self.file_name_suffix.value
undefined_tags = pipeline.get_undefined_metadata_tags(text_str)
if len(undefined_tags) > 0:
raise ValidationError(
"%s is not a defined metadata tag. Check the metadata specifications in your load modules"
% undefined_tags[0], self.file_name_suffix)
def validate_module_warnings(self, pipeline):
'''Warn user re: Test mode '''
if pipeline.test_mode:
raise ValidationError(
"ExportToACC will not produce output in Test Mode",
self.directory)
def prepare_run(self, workspace):
'''Prepare an image set to be run
workspace - workspace with image set populated (at this point)
returns False if analysis can't be done
'''
return self.check_overwrite(workspace)
def run(self, workspace):
# all of the work is done in post_run()
if self.show_window:
image_set_number = workspace.measurements.image_set_number
header = ["Filename"]
columns = []
path = self.make_image_file_name(workspace, image_set_number)
columns.append((path, ))
workspace.display_data.header = header
workspace.display_data.columns = columns
def display(self, workspace, figure):
figure.set_subplots((
1,
1,
))
if workspace.display_data.columns is None:
figure.subplot_table(0, 0, [["Data written to acc files"]])
elif workspace.pipeline.test_mode:
figure.subplot_table(
0, 0, [["Data not written to acc files in test mode"]])
else:
figure.subplot_table(0,
0,
workspace.display_data.columns,
col_labels=workspace.display_data.header)
def run_as_data_tool(self, workspace):
'''Run the module as a data tool
For ExportToACC, we do the "post_run" method in order to write
out the .txt files as if the experiment had just finished.
'''
#
# Set the measurements to the end of the list to mimic the state
# at the end of the run.
#
m = workspace.measurements
m.image_set_number = m.image_set_count
self.post_run(workspace)
def post_run(self, workspace):
'''Save measurements at end of run'''
#
# Don't export in test mode
#
#if workspace.pipeline.test_mode:
# return
object_names = self.filter_object_names(
workspace.measurements.get_object_names())
self.run_objects(object_names, workspace)
def should_stop_writing_measurements(self):
'''All subsequent modules should not write measurements'''
return True
def get_metadata_groups(self, workspace, settings_group=None):
'''Find the metadata groups that are relevant for creating the file name
workspace - the workspace with the image set metadata elements and
grouping measurements populated.
settings_group - if saving individual objects, this is the settings
group that controls naming the files.
'''
if settings_group is None or settings_group.wants_automatic_file_name:
tags = []
else:
tags = find_metadata_tokens(settings_group.file_name.value)
if self.directory.is_custom_choice:
tags += find_metadata_tokens(self.directory.custom_path)
metadata_groups = workspace.measurements.group_by_metadata(tags)
return metadata_groups
def run_objects(self, object_names, workspace, settings_group=None):
"""Create a file based on the object names
object_names - a sequence of object names (or Image or Experiment)
which tell us which objects get piled into each file
workspace - get the images from here.
settings_group - if present, use the settings group for naming.
"""
metadata_groups = self.get_metadata_groups(workspace, settings_group)
for metadata_group in metadata_groups:
self.make_object_file(object_names, metadata_group.image_numbers,
workspace, settings_group)
def make_full_filename(self,
file_name,
workspace=None,
image_set_number=None):
"""Convert a file name into an absolute path
We do a few things here:
* apply metadata from an image set to the file name if an
image set is specified
* change the relative path into an absolute one using the "." and "&"
convention
* Create any directories along the path
"""
if image_set_number is not None and workspace is not None:
file_name = workspace.measurements.apply_metadata(
file_name, image_set_number)
measurements = None if workspace is None else workspace.measurements
path_name = self.directory.get_absolute_path(measurements,
image_set_number)
file_name = os.path.join(path_name, file_name)
path, fname = os.path.split(file_name)
if not os.path.isdir(path):
os.makedirs(path)
return os.path.join(path, fname)
def extension(self):
'''Return the appropriate extension for the txt file name
The appropriate extension is "txt"
'''
return "txt"
def make_image_file_name(self,
workspace,
image_set_number,
settings_group=None):
'''Make file name for objects measured from an image
:param workspace: the current workspace
:param image_set_number: the current image set number
:param settings_group: the settings group used to name the file
'''
imagename = workspace.measurements.get_measurement(
IMAGE, "FileName_" + self.file_image_name.value, image_set_number)
filename = "%s" % os.path.splitext(imagename)[0]
if self.wants_file_name_suffix:
suffix = self.file_name_suffix.value
suffix = workspace.measurements.apply_metadata(
suffix, image_set_number)
filename += suffix
filename = "%s.%s" % (filename, self.extension())
return self.make_full_filename(filename, workspace, image_set_number)
def check_overwrite(self, workspace):
"""Make sure it's ok to overwrite any existing files before starting run
workspace - workspace with all image sets already populated
returns True if ok to proceed, False if user cancels
"""
if self.wants_overwrite_without_warning:
return True
files_to_check = []
metadata_groups = self.get_metadata_groups(workspace)
for metadata_group in metadata_groups:
image_number = metadata_group.image_numbers[0]
files_to_check.append(
self.make_image_file_name(workspace, image_number))
files_to_overwrite = filter(os.path.isfile, files_to_check)
if len(files_to_overwrite) > 0:
if get_headless():
logger.error(
"ExportToACC is configured to refrain from overwriting files and the following file(s) already exist: %s"
% ", ".join(files_to_overwrite))
return False
msg = "Overwrite the following file(s)?\n" +\
"\n".join(files_to_overwrite)
import wx
result = wx.MessageBox(
msg,
caption="ExportToACC: Overwrite existing files",
style=wx.YES_NO | wx.NO_DEFAULT | wx.ICON_QUESTION)
if result != wx.YES:
return False
return True
def filter_columns(self, features, object_name):
if self.pick_columns:
columns = [
self.columns.get_measurement_feature(x)
for x in self.columns.selections
if self.columns.get_measurement_object(x) == object_name
]
columns = set(columns)
features = [x for x in features if x in columns]
return features
def filter_object_names(self, object_names):
object_names.remove('Image')
object_names.remove('Experiment')
return object_names
def make_object_file(self,
object_names,
image_set_numbers,
workspace,
settings_group=None):
"""Make a file containing object measurements
object_names - sequence of names of the objects whose measurements
will be included
image_set_numbers - the image sets whose data gets extracted
workspace - workspace containing the measurements
settings_group - the settings group used to choose to make the file
"""
m = workspace.measurements
acc_file_name = os.path.join(
os.path.dirname(
self.make_image_file_name(workspace, image_set_numbers[0],
settings_group)), ACC_FILE_NAME)
features = []
objects_with_selected_features = []
center_x = ("", "")
center_y = ("", "")
for object_name in object_names:
if not object_name in m.get_object_names():
continue
rfeatures = m.get_feature_names(object_name)
rfeatures = self.filter_columns(rfeatures, object_name)
ofeatures = [
x for x in rfeatures if not [y for y in REMOVE_FEAT if y in x]
]
ofeatures = [(object_name, feature_name)
for feature_name in ofeatures]
ofeatures.sort()
features += ofeatures
# Haggish way to find feature to use as object coordinates
if ofeatures:
objects_with_selected_features.append(object_name)
coord = [feat for feat in rfeatures if "Location_Center_" in feat]
for feat in coord:
if (not center_x
or "Nuclei" == object_name) and "Center_X" in feat:
center_x = (object_name, feat)
if (not center_y
or "Nuclei" == object_name) and "Center_Y" in feat:
center_y = (object_name, feat)
features.insert(0, center_y)
features.insert(0, center_x)
# Write ACC file
try:
fd = open(acc_file_name, "w")
for feat in features:
fd.write(feat[0] + "_" + feat[1] + "\n")
fd.close()
except:
pass
for img_number in image_set_numbers:
try:
file_name = self.make_image_file_name(workspace, img_number,
settings_group)
fd = open(file_name, "w")
writer = csv.writer(fd, delimiter=DELIMITER)
object_count =\
np.max([m.get_measurement(IMAGE, "Count_%s"%name, img_number)
for name in objects_with_selected_features])
object_count = int(object_count) if object_count else 0
columns = [
np.repeat(img_number, object_count)
if feature_name == IMAGE_NUMBER else
np.arange(1, object_count +
1) if feature_name == OBJECT_NUMBER else np.
repeat(m.get_measurement(IMAGE, feature_name, img_number),
object_count) if object_name == IMAGE else
m.get_measurement(object_name, feature_name, img_number)
for object_name, feature_name in features
]
for obj_index in range(object_count):
row = [
column[obj_index] if
(column is not None
and obj_index < column.shape[0]) else np.NAN
for column in columns
]
if self.nan_representation == NANS_AS_NULLS:
row = [
"" if (field is None) or
(np.isreal(field) and not np.isfinite(field)) else
field for field in row
]
writer.writerow(row)
fd.close()
except:
pass
def prepare_to_create_batch(self, workspace, fn_alter_path):
'''Prepare to create a batch file
This function is called when CellProfiler is about to create a
file for batch processing. It will pickle the image set list's
"legacy_fields" dictionary. This callback lets a module prepare for
saving.
pipeline - the pipeline to be saved
image_set_list - the image set list to be saved
fn_alter_path - this is a function that takes a pathname on the local
host and returns a pathname on the remote host. It
handles issues such as replacing backslashes and
mapping mountpoints. It should be called for every
pathname stored in the settings or legacy fields.
ExportToACC has to convert the path to file names to
something that can be used on the cluster.
'''
self.directory.alter_for_create_batch_files(fn_alter_path)
return True
def upgrade_settings(self, setting_values, variable_revision_number,
module_name):
"""Adjust the setting values based on the version that saved them
"""
if variable_revision_number == 1:
setting_values = (setting_values[:4] + [NO, ""] +
setting_values[4:])
variable_revision_number = 2
# Standardize input/output directory name references
SLOT_DIRCHOICE = 1
directory = setting_values[SLOT_DIRCHOICE]
directory = Directory.upgrade_setting(directory)
setting_values = (setting_values[:SLOT_DIRCHOICE] + [directory] +
setting_values[SLOT_DIRCHOICE + 1:])
return setting_values, variable_revision_number
class RunCellpose(ImageSegmentation):
category = "Object Processing"
module_name = "RunCellpose"
variable_revision_number = 2
def create_settings(self):
super(RunCellpose, self).create_settings()
self.expected_diameter = Integer(
text="Expected object diameter",
value=15,
minval=0,
doc="""\
The average diameter of the objects to be detected. Setting this to 0 will attempt to automatically detect object size.
Note that automatic diameter mode does not work when running on 3D images.
Cellpose models come with a pre-defined object diameter. Your image will be resized during detection to attempt to
match the diameter expected by the model. The default models have an expected diameter of ~16 pixels, if trying to
detect much smaller objects it may be more efficient to resize the image first using the Resize module.
""",
)
self.mode = Choice(
text="Detection mode",
choices=[MODE_NUCLEI, MODE_CELLS, MODE_CUSTOM],
value=MODE_NUCLEI,
doc="""\
CellPose comes with models for detecting nuclei or cells. Alternatively, you can supply a custom-trained model
generated using the command line or Cellpose GUI. Custom models can be useful if working with unusual cell types.
""",
)
self.use_gpu = Binary(text="Use GPU",
value=False,
doc=f"""\
If enabled, Cellpose will attempt to run detection on your system's graphics card (GPU).
Note that you will need a CUDA-compatible GPU and correctly configured PyTorch version, see this link for details:
{CUDA_LINK}
If disabled or incorrectly configured, Cellpose will run on your CPU instead. This is much slower but more compatible
with different hardware setups.
Note that, particularly when in 3D mode, lack of GPU memory can become a limitation. If a model crashes you may need to
re-start CellProfiler to release GPU memory. Resizing large images prior to running them through the model can free up
GPU memory.
""")
self.use_averaging = Binary(text="Use averaging",
value=True,
doc="""\
If enabled, CellPose will run it's 4 inbuilt models and take a consensus to determine the results. If disabled, only a
single model will be called to produce results. Disabling averaging is faster to run but less accurate."""
)
self.supply_nuclei = Binary(text="Supply nuclei image as well?",
value=False,
doc="""
When detecting whole cells, you can provide a second image featuring a nuclear stain to assist
the model with segmentation. This can help to split touching cells.""")
self.nuclei_image = ImageSubscriber(
"Select the nuclei image",
doc="Select the image you want to use as the nuclear stain.")
self.save_probabilities = Binary(
text="Save probability image?",
value=False,
doc="""
If enabled, the probability scores from the model will be recorded as a new image.
Probability >0 is considered as being part of a cell.
You may want to use a higher threshold to manually generate objects.""",
)
self.probabilities_name = ImageName(
"Name the probability image",
"Probabilities",
doc=
"Enter the name you want to call the probability image produced by this module.",
)
self.model_directory = Directory(
"Location of the pre-trained model file",
doc=f"""\
*(Used only when using a custom pre-trained model)*
Select the location of the pre-trained CellPose model file that will be used for detection."""
)
def get_directory_fn():
"""Get the directory for the rules file name"""
return self.model_directory.get_absolute_path()
def set_directory_fn(path):
dir_choice, custom_path = self.model_directory.get_parts_from_path(
path)
self.model_directory.join_parts(dir_choice, custom_path)
self.model_file_name = Filename("Pre-trained model file name",
"cyto_0",
get_directory_fn=get_directory_fn,
set_directory_fn=set_directory_fn,
doc=f"""\
*(Used only when using a custom pre-trained model)*
This file can be generated by training a custom model withing the CellPose GUI or command line applications."""
)
self.gpu_test = DoSomething(
"",
"Test GPU",
self.do_check_gpu,
doc=f"""\
Press this button to check whether a GPU is correctly configured.
If you have a dedicated GPU, a failed test usually means that either your GPU does not support deep learning or the
required dependencies are not installed.
If you have multiple GPUs on your system, this button will only test the first one.
""",
)
self.flow_threshold = Float(
text="Flow threshold",
value=0.4,
minval=0,
doc=
"""Flow error threshold. All cells with errors below this threshold are kept. Recommended default is 0.4""",
)
self.dist_threshold = Float(
text="Cell probability threshold",
value=0.0,
minval=0,
doc=f"""\
Cell probability threshold (all pixels with probability above threshold kept for masks). Recommended default is 0.0. """,
)
def settings(self):
return [
self.x_name, self.expected_diameter, self.mode, self.y_name,
self.use_gpu, self.use_averaging, self.supply_nuclei,
self.nuclei_image, self.save_probabilities,
self.probabilities_name, self.model_directory,
self.model_file_name, self.flow_threshold, self.dist_threshold
]
def visible_settings(self):
vis_settings = [self.mode, self.x_name]
if self.mode.value != MODE_NUCLEI:
vis_settings += [self.supply_nuclei]
if self.supply_nuclei.value:
vis_settings += [self.nuclei_image]
if self.mode.value == MODE_CUSTOM:
vis_settings += [self.model_directory, self.model_file_name]
vis_settings += [
self.expected_diameter, self.flow_threshold, self.dist_threshold,
self.y_name, self.save_probabilities
]
if self.save_probabilities.value:
vis_settings += [self.probabilities_name]
vis_settings += [self.use_averaging, self.use_gpu]
if self.use_gpu.value:
vis_settings += [self.gpu_test]
return vis_settings
def run(self, workspace):
if self.mode.value != MODE_CUSTOM:
model = models.Cellpose(model_type='cyto' if self.mode.value
== MODE_CELLS else 'nuclei',
gpu=self.use_gpu.value)
else:
model_file = self.model_file_name.value
model_directory = self.model_directory.get_absolute_path()
model_path = os.path.join(model_directory, model_file)
model = models.CellposeModel(pretrained_model=model_path,
gpu=self.use_gpu.value)
x_name = self.x_name.value
y_name = self.y_name.value
images = workspace.image_set
x = images.get_image(x_name)
dimensions = x.dimensions
x_data = x.pixel_data
if x.multichannel:
raise ValueError(
"Color images are not currently supported. Please provide greyscale images."
)
if self.mode.value != "Nuclei" and self.supply_nuclei.value:
nuc_image = images.get_image(self.nuclei_image.value)
# CellPose expects RGB, we'll have a blank red channel, cells in green and nuclei in blue.
if x.volumetric:
x_data = numpy.stack(
(numpy.zeros_like(x_data), x_data, nuc_image.pixel_data),
axis=1)
else:
x_data = numpy.stack(
(numpy.zeros_like(x_data), x_data, nuc_image.pixel_data),
axis=-1)
channels = [2, 3]
else:
channels = [0, 0]
diam = self.expected_diameter.value if self.expected_diameter.value > 0 else None
try:
y_data, flows, *_ = model.eval(
x_data,
channels=channels,
diameter=diam,
net_avg=self.use_averaging.value,
do_3D=x.volumetric,
flow_threshold=self.flow_threshold.value,
cellprob_threshold=self.dist_threshold.value)
finally:
if self.use_gpu.value and model.torch:
# Try to clear some GPU memory for other worker processes.
try:
from torch import cuda
cuda.empty_cache()
except Exception as e:
print(
f"Unable to clear GPU memory. You may need to restart CellProfiler to change models. {e}"
)
y = Objects()
y.segmented = y_data
y.parent_image = x.parent_image
objects = workspace.object_set
objects.add_objects(y, y_name)
if self.save_probabilities.value:
# Flows come out sized relative to CellPose's inbuilt model size.
# We need to slightly resize to match the original image.
size_corrected = resize(flows[2], y_data.shape)
prob_image = Image(
size_corrected,
parent_image=x.parent_image,
convert=False,
dimensions=len(size_corrected.shape),
)
workspace.image_set.add(self.probabilities_name.value, prob_image)
if self.show_window:
workspace.display_data.probabilities = size_corrected
self.add_measurements(workspace)
if self.show_window:
if x.volumetric:
# Can't show CellPose-accepted colour images in 3D
workspace.display_data.x_data = x.pixel_data
else:
workspace.display_data.x_data = x_data
workspace.display_data.y_data = y_data
workspace.display_data.dimensions = dimensions
def display(self, workspace, figure):
if self.save_probabilities.value:
layout = (2, 2)
else:
layout = (2, 1)
figure.set_subplots(dimensions=workspace.display_data.dimensions,
subplots=layout)
figure.subplot_imshow(
colormap="gray",
image=workspace.display_data.x_data,
title="Input Image",
x=0,
y=0,
)
figure.subplot_imshow_labels(
image=workspace.display_data.y_data,
sharexy=figure.subplot(0, 0),
title=self.y_name.value,
x=1,
y=0,
)
if self.save_probabilities.value:
figure.subplot_imshow(
colormap="gray",
image=workspace.display_data.probabilities,
sharexy=figure.subplot(0, 0),
title=self.probabilities_name.value,
x=0,
y=1,
)
def do_check_gpu(self):
import importlib.util
torch_installed = importlib.util.find_spec('torch') is not None
if models.use_gpu(istorch=torch_installed):
message = "GPU appears to be working correctly!"
else:
message = "GPU test failed. There may be something wrong with your configuration."
import wx
wx.MessageBox(message, caption="GPU Test")
def upgrade_settings(self, setting_values, variable_revision_number,
module_name):
if variable_revision_number == 1:
setting_values = setting_values + ["0.4", "0.0"]
variable_revision_number = 2
return setting_values, variable_revision_number
class SaveCroppedObjects(Module):
category = "File Processing"
module_name = "SaveCroppedObjects"
variable_revision_number = 2
def create_settings(self):
self.export_option = Choice(
"Do you want to save cropped images or object masks?",
[SAVE_PER_OBJECT, SAVE_MASK],
doc="""\
Choose the way you want the per-object crops to be exported.
The choices are:
- *{SAVE_PER_OBJECT}*: Save a per-object crop from the original image
based on the object's bounding box.
- *{SAVE_MASK}*: Export a per-object mask.""".format(
SAVE_PER_OBJECT=SAVE_PER_OBJECT, SAVE_MASK=SAVE_MASK),
)
self.objects_name = LabelSubscriber(
"Objects",
doc="Select the objects you want to export as per-object crops.")
self.image_name = ImageSubscriber("Image",
doc="Select the image to crop")
self.directory = Directory(
"Directory",
doc="Enter the directory where object crops are saved.",
value=DEFAULT_OUTPUT_FOLDER_NAME,
)
self.file_format = Choice(
"Saved file format",
[O_PNG, O_TIFF_8, O_TIFF_16],
value=O_TIFF_8,
doc="""\
**{O_PNG}** files do not support 3D. **{O_TIFF_8}** files use zlib compression level 6."""
.format(O_PNG=O_PNG, O_TIFF_8=O_TIFF_8, O_TIFF_16=O_TIFF_16),
)
def display(self, workspace, figure):
figure.set_subplots((1, 1))
figure.subplot_table(0, 0,
[["\n".join(workspace.display_data.filenames)]])
def run(self, workspace):
objects = workspace.object_set.get_objects(self.objects_name.value)
directory = self.directory.get_absolute_path(workspace.measurements)
if not os.path.exists(directory):
os.makedirs(directory)
labels = objects.segmented
unique_labels = numpy.unique(labels)
if unique_labels[0] == 0:
unique_labels = unique_labels[1:]
filenames = []
if self.export_option == SAVE_PER_OBJECT:
images = workspace.image_set
x = images.get_image(self.image_name.value)
if len(x.pixel_data.shape) == len(
labels.shape) + 1 and not x.volumetric:
# Color 2D image, repeat mask for all channels
labels = numpy.repeat(labels[:, :, numpy.newaxis],
x.pixel_data.shape[-1],
axis=2)
for label in unique_labels:
if self.export_option == SAVE_MASK:
mask = labels == label
elif self.export_option == SAVE_PER_OBJECT:
mask_in = labels == label
properties = skimage.measure.regionprops(
mask_in.astype(int), intensity_image=x.pixel_data)
mask = properties[0].intensity_image
if self.file_format.value == O_PNG:
filename = os.path.join(
directory, "{}_{}.{}".format(self.objects_name.value,
label, O_PNG))
skimage.io.imsave(filename,
skimage.img_as_ubyte(mask),
check_contrast=False)
elif self.file_format.value == O_TIFF_8:
filename = os.path.join(
directory, "{}_{}.{}".format(self.objects_name.value,
label, "tiff"))
skimage.io.imsave(
filename,
skimage.img_as_ubyte(mask),
compress=6,
check_contrast=False,
)
elif self.file_format.value == O_TIFF_16:
filename = os.path.join(
directory, "{}_{}.{}".format(self.objects_name.value,
label, "tiff"))
skimage.io.imsave(
filename,
skimage.img_as_uint(mask),
compress=6,
check_contrast=False,
)
filenames.append(filename)
if self.show_window:
workspace.display_data.filenames = filenames
def settings(self):
settings = [
self.objects_name,
self.directory,
self.file_format,
self.export_option,
self.image_name,
]
return settings
def visible_settings(self):
result = [
self.export_option,
self.objects_name,
self.directory,
self.file_format,
]
if self.export_option.value == SAVE_PER_OBJECT:
result += [self.image_name]
return result
def volumetric(self):
return True
class SaveImages(Module):
module_name = "SaveImages"
variable_revision_number = 16
category = "File Processing"
def create_settings(self):
self.save_image_or_figure = Choice(
"Select the type of image to save",
IF_ALL,
IF_IMAGE,
doc="""\
The following types of images can be saved as a file on the hard drive:
- *{IF_IMAGE}:* Any of the images produced upstream of **SaveImages**
can be selected for saving. Outlines of objects created by other
modules such as **Identify** modules, **Watershed**, and various object
processing modules can also be saved with this option, but you must
use the **OverlayOutlines** module to create them prior to saving images.
Likewise, if you wish to save the objects themselves, you must use the
**ConvertObjectsToImage** module to create a savable image.
- *{IF_MASK}:* Relevant only if a module that produces masks has been used
such as **Crop**, **MaskImage**, or **MaskObjects**. These
modules create a mask of the pixels of interest in the
image. Saving the mask will produce a binary image in which the
pixels of interest are set to 1; all other pixels are set to 0.
- *{IF_CROPPING}:* Relevant only if the **Crop** module is used. The
**Crop** module also creates a cropping image which is typically the
same size as the original image. However, since **Crop** permits
removal of the rows and columns that are left blank, the cropping can
be of a different size than the mask.
- *{IF_MOVIE}:* A sequence of images can be saved as a TIFF stack.
""".format(
**{
"IF_CROPPING": IF_CROPPING,
"IF_IMAGE": IF_IMAGE,
"IF_MASK": IF_MASK,
"IF_MOVIE": IF_MOVIE,
}
),
)
self.image_name = ImageSubscriber(
"Select the image to save", doc="Select the image you want to save."
)
self.file_name_method = Choice(
"Select method for constructing file names",
[FN_FROM_IMAGE, FN_SEQUENTIAL, FN_SINGLE_NAME],
FN_FROM_IMAGE,
doc="""\
*(Used only if saving non-movie files)*
Several choices are available for constructing the image file name:
- *{FN_FROM_IMAGE}:* The filename will be constructed based on the
original filename of an input image specified in **NamesAndTypes**.
You will have the opportunity to prefix or append additional text.
If you have metadata associated with your images, you can append
text to the image filename using a metadata tag. This is especially
useful if you want your output given a unique label according to the
metadata corresponding to an image group. The name of the metadata to
substitute can be provided for each image for each cycle using the
**Metadata** module.
- *{FN_SEQUENTIAL}:* Same as above, but in addition, each filename
will have a number appended to the end that corresponds to the image
cycle number (starting at 1).
- *{FN_SINGLE_NAME}:* A single name will be given to the file. Since
the filename is fixed, this file will be overwritten with each cycle.
In this case, you would probably want to save the image on the last
cycle (see the *Select how often to save* setting). The exception to
this is to use a metadata tag to provide a unique label, as mentioned
in the *{FN_FROM_IMAGE}* option.
{USING_METADATA_TAGS_REF}
{USING_METADATA_HELP_REF}
""".format(
**{
"FN_FROM_IMAGE": FN_FROM_IMAGE,
"FN_SEQUENTIAL": FN_SEQUENTIAL,
"FN_SINGLE_NAME": FN_SINGLE_NAME,
"USING_METADATA_HELP_REF": _help.USING_METADATA_HELP_REF,
"USING_METADATA_TAGS_REF": _help.USING_METADATA_TAGS_REF,
}
),
)
self.file_image_name = FileImageSubscriber(
"Select image name for file prefix",
"None",
doc="""\
*(Used only when “{FN_FROM_IMAGE}” is selected for constructing the filename)*
Select an image loaded using **NamesAndTypes**. The original filename
will be used as the prefix for the output filename.""".format(
**{"FN_FROM_IMAGE": FN_FROM_IMAGE}
),
)
self.single_file_name = Text(
SINGLE_NAME_TEXT,
"OrigBlue",
metadata=True,
doc="""\
*(Used only when “{FN_SEQUENTIAL}” or “{FN_SINGLE_NAME}” are selected
for constructing the filename)*
Specify the filename text here. If you have metadata associated with
your images, enter the filename text with the metadata tags.
{USING_METADATA_TAGS_REF}
Do not enter the file extension in this setting; it will be appended
automatically.""".format(
**{
"FN_SEQUENTIAL": FN_SEQUENTIAL,
"FN_SINGLE_NAME": FN_SINGLE_NAME,
"USING_METADATA_TAGS_REF": _help.USING_METADATA_TAGS_REF,
}
),
)
self.number_of_digits = Integer(
"Number of digits",
4,
doc="""\
*(Used only when “{FN_SEQUENTIAL}” is selected for constructing the filename)*
Specify the number of digits to be used for the sequential numbering.
Zeros will be used to left-pad the digits. If the number specified here
is less than that needed to contain the number of image sets, the latter
will override the value entered.""".format(
**{"FN_SEQUENTIAL": FN_SEQUENTIAL}
),
)
self.wants_file_name_suffix = Binary(
"Append a suffix to the image file name?",
False,
doc="""\
Select "*{YES}*" to add a suffix to the image’s file name. Select "*{NO}*"
to use the image name as-is.
""".format(
**{"NO": "No", "YES": "Yes"}
),
)
self.file_name_suffix = Text(
"Text to append to the image name",
"",
metadata=True,
doc="""\
*(Used only when constructing the filename from the image filename)*
Enter the text that should be appended to the filename specified above.
If you have metadata associated with your images, you may use metadata tags.
{USING_METADATA_TAGS_REF}
Do not enter the file extension in this setting; it will be appended
automatically.
""".format(
**{"USING_METADATA_TAGS_REF": _help.USING_METADATA_TAGS_REF}
),
)
self.file_format = Choice(
"Saved file format",
[FF_JPEG, FF_NPY, FF_PNG, FF_TIFF, FF_H5],
value=FF_TIFF,
doc="""\
*(Used only when saving non-movie files)*
Select the format to save the image(s).
Only *{FF_TIFF}* supports saving as 16-bit or 32-bit. *{FF_TIFF}* is a
"lossless" file format.
*{FF_PNG}* is also a "lossless" file format and it tends to produce
smaller files without losing any image data.
*{FF_JPEG}* is also small but is a "lossy" file format and should not be
used for any images that will undergo further quantitative analysis.
Select *{FF_NPY}* to save an illumination correction image generated by
**CorrectIlluminationCalculate**.
Select *{FF_H5}* to save files to be used for Ilastik pixel classificaiton.
The images should be correctly recognized as yxcz images.""".format(
**{
"FF_NPY": FF_NPY,
"FF_TIFF": FF_TIFF,
"FF_PNG": FF_PNG,
"FF_JPEG": FF_JPEG,
"FF_H5": FF_H5,
}
),
)
self.pathname = SaveImagesDirectoryPath(
"Output file location",
self.file_image_name,
doc="""\
This setting lets you choose the folder for the output files.
{IO_FOLDER_CHOICE_HELP_TEXT}
An additional option is the following:
- *Same folder as image*: Place the output file in the same folder that
the source image is located.
{IO_WITH_METADATA_HELP_TEXT}
If the subfolder does not exist when the pipeline is run, CellProfiler
will create it.
If you are creating nested subfolders using the sub-folder options, you
can specify the additional folders separated with slashes. For example,
“Outlines/Plate1” will create a “Plate1” folder in the “Outlines”
folder, which in turn is under the Default Input/Output Folder. The use
of a forward slash (“/”) as a folder separator will avoid ambiguity
between the various operating systems.
""".format(
**{
"IO_FOLDER_CHOICE_HELP_TEXT": _help.IO_FOLDER_CHOICE_HELP_TEXT,
"IO_WITH_METADATA_HELP_TEXT": _help.IO_WITH_METADATA_HELP_TEXT,
}
),
)
self.bit_depth = Choice(
"Image bit depth",
[BIT_DEPTH_8, BIT_DEPTH_16, BIT_DEPTH_FLOAT, BIT_DEPTH_RAW],
doc=f"""\
Select the bit-depth at which you want to save the images.
*{BIT_DEPTH_FLOAT}* saves the image as floating-point decimals with
32-bit precision. When the input data is integer or binary type, pixel
values are scaled within the range (0, 1). Floating point data is not
rescaled.
*{BIT_DEPTH_16}* and *{BIT_DEPTH_FLOAT}* images are supported only for
TIFF formats.
Data is normally checked and transformed to ensure that it matches the
selected format's requirements. Selecting *{BIT_DEPTH_RAW}* will attempt
to automatically save to a compatible format without applying any
transformations to the data. This could be used to save integer labels
in 32-bit float format if you had more labels than the 16-bit format can
handle (without rescaling to the 0-1 range of *{BIT_DEPTH_FLOAT}*).
Note that because the data validation step is skipped some images may
fail to save if they contain unusable data.
Note: Opening exported multichannel 16-bit TIFF stacks in ImageJ may require
the BioFormats Importer plugin due to the compression method used by
CellProfiler.""",
)
self.tiff_compress = Binary(
"Save with lossless compression?",
value=True,
doc="""\
*(Used only when saving 2D images as file type tiff)*
Choose whether or not to use lossless compression when saving
images. This will lead to smaller file sizes, but somewhat longer
module execution time. Note that the value of this setting will
be ignored when saving 3D tiff images, which have been saved by
default with compression since CellProfiler 3.1. Do not use for
multichannel tiff images created as Stacks in GrayToColor."""
)
self.stack_axis = Choice(
"How to save the series",
[AXIS_T, AXIS_Z],
value=AXIS_T,
doc="""\
*(Used only when saving movie/stack files)*
This setting determines how planes are saved into a movie/stack.
Selecting "T" will save planes as a time series. Selecting "Z"
will save planes as slices in a 3D z-axis.
""",
)
self.overwrite = Binary(
"Overwrite existing files without warning?",
False,
doc="""\
Select "*{YES}*" to automatically overwrite a file if it already exists.
Select "*{NO}*" to be prompted for confirmation first.
If you are running the pipeline on a computing cluster, select "*{YES}*"
since you will not be able to intervene and answer the confirmation
prompt.""".format(
**{"NO": "No", "YES": "Yes"}
),
)
self.when_to_save = Choice(
"When to save",
[WS_EVERY_CYCLE, WS_FIRST_CYCLE, WS_LAST_CYCLE],
WS_EVERY_CYCLE,
doc="""\
*(Used only when saving non-movie files)*
Specify at what point during pipeline execution to save file(s).
- *{WS_EVERY_CYCLE}:* Useful for when the image of interest is
created every cycle and is not dependent on results from a prior
cycle.
- *{WS_FIRST_CYCLE}:* Useful for when you are saving an aggregate
image created on the first cycle, e.g.,
**CorrectIlluminationCalculate** with the *All* setting used on
images obtained directly from **NamesAndTypes**.
- *{WS_LAST_CYCLE}:* Useful for when you are saving an aggregate image
completed on the last cycle, e.g., **CorrectIlluminationCalculate**
with the *All* setting used on intermediate images generated during
each cycle.""".format(
**{
"WS_EVERY_CYCLE": WS_EVERY_CYCLE,
"WS_FIRST_CYCLE": WS_FIRST_CYCLE,
"WS_LAST_CYCLE": WS_LAST_CYCLE,
}
),
)
self.update_file_names = Binary(
"Record the file and path information to the saved image?",
False,
doc="""\
Select "*{YES}*" to store filename and pathname data for each of the new
files created via this module as a per-image measurement.
Instances in which this information may be useful include:
- Exporting measurements to a database, allowing access to the saved
image. If you are using the machine-learning tools or image viewer in
CellProfiler Analyst, for example, you will want to enable this
setting if you want the saved images to be displayed along with the
original images.""".format(
**{"YES": "Yes"}
),
)
self.create_subdirectories = Binary(
"Create subfolders in the output folder?",
False,
doc="""
Select "*{YES}*" to create subfolders to match the input image folder structure.
For example, if your input images are organized into subfolders (e.g., for each plate, well, animal, etc.),
this option allows you to mirror some or all of that nested folder structure in the output folder.""".format(
**{"YES": "Yes"}
),
)
self.root_dir = Directory(
"Base image folder",
doc="""\
*Used only if creating subfolders in the output folder*
In subfolder mode, **SaveImages** determines the folder for an output image file by
examining the path of the matching input file.
You should choose as **Base image folder** the input folder that has the structure you'd like
to mirror in the output folder.
Consider an example where your input images are stored in a nested folder structure of
"images\/experiment-name\/plate-name" (i.e., your files are in folders for each plate, nested
inside of folders for each experiment, nested in a parent folder called "images").
If you select the base image folder to be **images**, **SaveImages** will go to your "Output file
location" and save images in subfolders "experiment-name\/plate-name" that corresponds to each
input image. If the base image folder chosen is one level deeper at "images\/experiment-name",
**SaveImages** will store images in subfolders for each "plate-name" they belong to.
**Warning**: Do not select the same folder you selected for "Output file location" as this can lead
to unexpected behavior like saving in the original input file directory. For safety, ensure
"Overwrite existing files without warning?" is set to "No" while testing this option. """,
)
def settings(self):
"""Return the settings in the order to use when saving"""
return [
self.save_image_or_figure,
self.image_name,
self.file_name_method,
self.file_image_name,
self.single_file_name,
self.number_of_digits,
self.wants_file_name_suffix,
self.file_name_suffix,
self.file_format,
self.pathname,
self.bit_depth,
self.overwrite,
self.when_to_save,
self.update_file_names,
self.create_subdirectories,
self.root_dir,
self.stack_axis,
self.tiff_compress,
]
def visible_settings(self):
"""Return only the settings that should be shown"""
result = [self.save_image_or_figure, self.image_name, self.file_name_method]
if self.file_name_method == FN_FROM_IMAGE:
result += [self.file_image_name, self.wants_file_name_suffix]
if self.wants_file_name_suffix:
result.append(self.file_name_suffix)
elif self.file_name_method == FN_SEQUENTIAL:
self.single_file_name.text = SEQUENTIAL_NUMBER_TEXT
# XXX - Change doc, as well!
result.append(self.single_file_name)
result.append(self.number_of_digits)
elif self.file_name_method == FN_SINGLE_NAME:
self.single_file_name.text = SINGLE_NAME_TEXT
result.append(self.single_file_name)
else:
raise NotImplementedError(
"Unhandled file name method: %s" % self.file_name_method
)
if self.save_image_or_figure != IF_MOVIE:
result.append(self.file_format)
supports_16_bit = (
self.file_format in (FF_TIFF, FF_H5)
and self.save_image_or_figure == IF_IMAGE
) or self.save_image_or_figure == IF_MOVIE
if supports_16_bit:
# TIFF supports 8 & 16-bit, all others are written 8-bit
result.append(self.bit_depth)
if self.file_format == FF_TIFF:
result.append(self.tiff_compress)
if self.save_image_or_figure == IF_MOVIE:
result.append(self.stack_axis)
result.append(self.pathname)
result.append(self.overwrite)
if self.save_image_or_figure != IF_MOVIE:
result.append(self.when_to_save)
result.append(self.update_file_names)
if self.file_name_method == FN_FROM_IMAGE:
result.append(self.create_subdirectories)
if self.create_subdirectories:
result.append(self.root_dir)
return result
@property
def module_key(self):
return "%s_%d" % (self.module_name, self.module_num)
def prepare_group(self, workspace, grouping, image_numbers):
d = self.get_dictionary(workspace.image_set_list)
if self.save_image_or_figure == IF_MOVIE:
d["N_FRAMES"] = len(image_numbers)
d["CURRENT_FRAME"] = 0
return True
def prepare_to_create_batch(self, workspace, fn_alter_path):
self.pathname.alter_for_create_batch_files(fn_alter_path)
if self.create_subdirectories:
self.root_dir.alter_for_create_batch_files(fn_alter_path)
def run(self, workspace):
"""Run the module
pipeline - instance of cellprofiler_core.pipeline for this run
workspace - the workspace contains:
image_set - the images in the image set being processed
object_set - the objects (labeled masks) in this image set
measurements - the measurements for this run
frame - display within this frame (or None to not display)
"""
if self.save_image_or_figure.value in (IF_IMAGE, IF_MASK, IF_CROPPING):
should_save = self.run_image(workspace)
elif self.save_image_or_figure == IF_MOVIE:
self.run_movie(workspace)
else:
raise NotImplementedError(
("Saving a %s is not yet supported" % self.save_image_or_figure)
)
workspace.display_data.filename = self.get_filename(
workspace, make_dirs=False, check_overwrite=False
)
def is_aggregation_module(self):
"""SaveImages is an aggregation module when it writes movies"""
return (
self.save_image_or_figure == IF_MOVIE or self.when_to_save == WS_LAST_CYCLE
)
def display(self, workspace, figure):
if self.show_window:
if self.save_image_or_figure == IF_MOVIE:
return
figure.set_subplots((1, 1))
outcome = (
"Wrote %s" if workspace.display_data.wrote_image else "Did not write %s"
)
figure.subplot_table(0, 0, [[outcome % workspace.display_data.filename]])
def run_image(self, workspace):
"""Handle saving an image"""
#
# First, check to see if we should save this image
#
if self.when_to_save == WS_FIRST_CYCLE:
d = self.get_dictionary(workspace.image_set_list)
if workspace.measurements["Image", "Group_Index",] > 1:
workspace.display_data.wrote_image = False
self.save_filename_measurements(workspace)
return
d["FIRST_IMAGE"] = False
elif self.when_to_save == WS_LAST_CYCLE:
workspace.display_data.wrote_image = False
self.save_filename_measurements(workspace)
return
self.save_image(workspace)
return True
def run_movie(self, workspace):
out_file = self.get_filename(workspace, check_overwrite=False)
# overwrite checks are made only for first frame.
d = self.get_dictionary(workspace.image_set_list)
if d["CURRENT_FRAME"] == 0 and os.path.exists(out_file):
if not self.check_overwrite(out_file, workspace):
d["CURRENT_FRAME"] = "Ignore"
return
else:
# Have to delete the old movie before making the new one
os.remove(out_file)
elif d["CURRENT_FRAME"] == "Ignore":
return
image = workspace.image_set.get_image(self.image_name.value)
pixels = image.pixel_data
if self.get_bit_depth() == BIT_DEPTH_8:
pixels = skimage.util.img_as_ubyte(pixels)
pixel_type = bioformats.omexml.PT_UINT8
elif self.get_bit_depth() == BIT_DEPTH_16:
pixels = skimage.util.img_as_uint(pixels)
pixel_type = bioformats.omexml.PT_UINT16
elif self.get_bit_depth() == BIT_DEPTH_FLOAT:
pixels = skimage.util.img_as_float32(pixels)
pixel_type = bioformats.omexml.PT_FLOAT
else:
raise ValueError("Bit depth unsupported in movie mode")
frames = d["N_FRAMES"]
current_frame = d["CURRENT_FRAME"]
d["CURRENT_FRAME"] += 1
if self.stack_axis == AXIS_T:
self.do_save_image(
workspace, out_file, pixels, pixel_type, t=current_frame, size_t=frames,
)
else:
self.do_save_image(
workspace, out_file, pixels, pixel_type, z=current_frame, size_z=frames,
)
def post_group(self, workspace, *args):
if self.when_to_save == WS_LAST_CYCLE and self.save_image_or_figure != IF_MOVIE:
try:
self.save_image(workspace)
except ValueError:
raise ValueError(
"You have tried to save %s on the last cycle but that cycle failed FlagImages. Please adjust the FlagImages settings and rerun"
% (self.image_name.value)
)
def do_save_image(
self,
workspace,
filename,
pixels,
pixel_type,
c=0,
z=0,
t=0,
size_c=1,
size_z=1,
size_t=1,
channel_names=None,
):
"""Save image using bioformats
workspace - the current workspace
filename - save to this filename
pixels - the image to save
pixel_type - save using this pixel type
c - the image's channel index
z - the image's z index
t - the image's t index
sizeC - # of channels in the stack
sizeZ - # of z stacks
sizeT - # of timepoints in the stack
channel_names - names of the channels (make up names if not present
"""
bioformats.formatwriter.write_image(
filename,
pixels,
pixel_type,
c=c,
z=z,
t=t,
size_c=size_c,
size_z=size_z,
size_t=size_t,
channel_names=channel_names,
)
def save_image(self, workspace):
if self.show_window:
workspace.display_data.wrote_image = False
filename = self.get_filename(workspace)
if filename is None: # failed overwrite check
return
image = workspace.image_set.get_image(self.image_name.value)
volumetric_extensions = [FF_NPY, FF_TIFF, FF_H5]
if image.volumetric and self.file_format.value not in volumetric_extensions:
raise RuntimeError(
"Unsupported file format {} for 3D pipeline. Use {} format when processing images as 3D.".format(
self.file_format.value, ", or ".join(volumetric_extensions)
)
)
if self.save_image_or_figure.value == IF_IMAGE:
pixels = image.pixel_data
elif self.save_image_or_figure.value == IF_MASK:
pixels = image.mask
elif self.save_image_or_figure.value == IF_CROPPING:
pixels = image.crop_mask
if self.file_format == FF_NPY:
numpy.save(filename, pixels)
else:
save_kwargs = {}
if self.get_bit_depth() == BIT_DEPTH_8:
pixels = skimage.util.img_as_ubyte(pixels)
elif self.get_bit_depth() == BIT_DEPTH_16:
pixels = skimage.util.img_as_uint(pixels)
elif self.get_bit_depth() == BIT_DEPTH_FLOAT:
pixels = skimage.util.img_as_float32(pixels)
elif self.get_bit_depth() == BIT_DEPTH_RAW:
# No bit depth transformation
pass
# skimage will save out color images (M,N,3) or (M,N,4) appropriately
# but any more than that will need to be transposed so they conform to the
# CYX convention rather than YXC
# http://scikit-image.org/docs/dev/api/skimage.io.html#skimage.io.imsave
if (
not image.volumetric
and len(pixels.shape) > 2
and image.channelstack
and self.file_format.value == FF_TIFF
):
pixels = numpy.transpose(pixels, (2, 0, 1))
save_kwargs.update({'imagej':True})
if (image.volumetric or self.tiff_compress.value) and self.file_format.value == FF_TIFF:
save_kwargs.update({"compress": 6})
if self.file_format.value == FF_H5:
save_h5(filename, pixels, volumetric=image.volumetric)
else:
skimage.io.imsave(filename, pixels, **save_kwargs)
if self.show_window:
workspace.display_data.wrote_image = True
if self.when_to_save != WS_LAST_CYCLE:
self.save_filename_measurements(workspace)
def check_overwrite(self, filename, workspace):
"""Check to see if it's legal to overwrite a file
Throws an exception if can't overwrite and no interaction available.
Returns False if can't overwrite, otherwise True.
"""
if not self.overwrite.value and os.path.isfile(filename):
try:
return (
workspace.interaction_request(
self, workspace.measurements.image_set_number, filename
)
== "Yes"
)
except workspace.NoInteractionException:
raise ValueError(
'SaveImages: trying to overwrite %s in headless mode, but Overwrite files is set to "No"'
% filename
)
return True
def handle_interaction(self, image_set_number, filename):
"""handle an interaction request from check_overwrite()"""
import wx
dlg = wx.MessageDialog(
wx.GetApp().TopWindow,
"%s #%d, set #%d - Do you want to overwrite %s?"
% (self.module_name, self.module_num, image_set_number, filename),
"Warning: overwriting file",
wx.YES_NO | wx.ICON_QUESTION,
)
result = dlg.ShowModal() == wx.ID_YES
return "Yes" if result else "No"
def save_filename_measurements(self, workspace):
if self.update_file_names.value:
filename = self.get_filename(
workspace, make_dirs=False, check_overwrite=False
)
pn, fn = os.path.split(filename)
url = cellprofiler_core.utilities.pathname.pathname2url(filename)
workspace.measurements.add_measurement(
"Image", self.file_name_feature, fn,
)
workspace.measurements.add_measurement(
"Image", self.path_name_feature, pn,
)
workspace.measurements.add_measurement(
"Image", self.url_feature, url,
)
@property
def file_name_feature(self):
return "_".join((C_FILE_NAME, self.image_name.value))
@property
def path_name_feature(self):
return "_".join((C_PATH_NAME, self.image_name.value))
@property
def url_feature(self):
return "_".join((C_URL, self.image_name.value))
@property
def source_file_name_feature(self):
"""The file name measurement for the exemplar disk image"""
return "_".join((C_FILE_NAME, self.file_image_name.value))
def source_path(self, workspace):
"""The path for the image data, or its first parent with a path"""
if self.file_name_method.value == FN_FROM_IMAGE:
path_feature = "%s_%s" % (C_PATH_NAME, self.file_image_name.value,)
assert workspace.measurements.has_feature("Image", path_feature), (
"Image %s does not have a path!" % self.file_image_name.value
)
return workspace.measurements.get_current_image_measurement(path_feature)
# ... otherwise, chase the cpimage hierarchy looking for an image with a path
cur_image = workspace.image_set.get_image(self.image_name.value)
while cur_image.path_name is None:
cur_image = cur_image.parent_image
assert (
cur_image is not None
), "Could not determine source path for image %s' % (self.image_name.value)"
return cur_image.path_name
def get_measurement_columns(self, pipeline):
if self.update_file_names.value:
return [
("Image", self.file_name_feature, COLTYPE_VARCHAR_FILE_NAME,),
("Image", self.path_name_feature, COLTYPE_VARCHAR_PATH_NAME,),
]
else:
return []
def get_filename(self, workspace, make_dirs=True, check_overwrite=True):
"""Concoct a filename for the current image based on the user settings"""
measurements = workspace.measurements
if self.file_name_method == FN_SINGLE_NAME:
filename = self.single_file_name.value
filename = workspace.measurements.apply_metadata(filename)
elif self.file_name_method == FN_SEQUENTIAL:
filename = self.single_file_name.value
filename = workspace.measurements.apply_metadata(filename)
n_image_sets = workspace.measurements.image_set_count
ndigits = int(numpy.ceil(numpy.log10(n_image_sets + 1)))
ndigits = max((ndigits, self.number_of_digits.value))
padded_num_string = str(measurements.image_set_number).zfill(ndigits)
filename = "%s%s" % (filename, padded_num_string)
else:
file_name_feature = self.source_file_name_feature
filename = measurements.get_current_measurement("Image", file_name_feature)
filename = os.path.splitext(filename)[0]
if self.wants_file_name_suffix:
suffix = self.file_name_suffix.value
suffix = workspace.measurements.apply_metadata(suffix)
filename += suffix
filename = "%s.%s" % (filename, self.get_file_format())
pathname = self.pathname.get_absolute_path(measurements)
if self.create_subdirectories:
image_path = self.source_path(workspace)
subdir = os.path.relpath(image_path, self.root_dir.get_absolute_path())
pathname = os.path.join(pathname, subdir)
if len(pathname) and not os.path.isdir(pathname) and make_dirs:
try:
os.makedirs(pathname)
except:
#
# On cluster, this can fail if the path was created by
# another process after this process found it did not exist.
#
if not os.path.isdir(pathname):
raise
result = os.path.join(pathname, filename)
if check_overwrite and not self.check_overwrite(result, workspace):
return
if check_overwrite and os.path.isfile(result):
try:
os.remove(result)
except:
import bioformats
bioformats.clear_image_reader_cache()
os.remove(result)
return result
def get_file_format(self):
"""Return the file format associated with the extension in self.file_format
"""
if self.save_image_or_figure == IF_MOVIE:
return FF_TIFF
return self.file_format.value
def get_bit_depth(self):
if self.save_image_or_figure in (
IF_IMAGE,
IF_MOVIE,
) and self.get_file_format() in (FF_TIFF, FF_H5):
return self.bit_depth.value
else:
return BIT_DEPTH_8
def upgrade_settings(self, setting_values, variable_revision_number, module_name):
if variable_revision_number == 11:
if setting_values[0] == "Objects":
raise NotImplementedError(
"Unsupported image type: Objects. Use <i>ConvertObjectsToImage</i> to create an image."
)
if setting_values[10] in ("bmp", "mat"):
raise NotImplementedError(
"Unsupported file format: {}".format(setting_values[10])
)
elif setting_values[10] == "tif":
setting_values[10] = FF_TIFF
elif setting_values[10] == "jpg":
setting_values[10] = FF_JPEG
new_setting_values = setting_values[:2]
new_setting_values += setting_values[4:15]
new_setting_values += setting_values[18:-1]
setting_values = new_setting_values
if setting_values[10] == "8":
setting_values[10] = BIT_DEPTH_8
elif setting_values[10] == "16":
setting_values[10] = BIT_DEPTH_16
variable_revision_number = 12
if variable_revision_number == 12:
if setting_values[10] == "64-bit floating point":
setting_values[10] = BIT_DEPTH_FLOAT
variable_revision_number = 13
if variable_revision_number == 13:
variable_revision_number = 14
if variable_revision_number == 14:
# Renamed "Movie" to "Movie/Stack"
if setting_values[0] == "Movie":
setting_values[0] = IF_MOVIE
# Added movie save axis
setting_values.append(AXIS_T)
variable_revision_number = 15
if variable_revision_number == 15:
setting_values.append(False)
variable_revision_number == 16
return setting_values, variable_revision_number
def validate_module(self, pipeline):
if self.save_image_or_figure in (
IF_IMAGE,
IF_MASK,
IF_CROPPING,
) and self.when_to_save in (WS_FIRST_CYCLE, WS_EVERY_CYCLE):
#
# Make sure that the image name is available on every cycle
#
for setting in get_name_providers(pipeline, self.image_name):
if setting.provided_attributes.get("available_on_last"):
#
# If we fell through, then you can only save on the last cycle
#
raise ValidationError(
"%s is only available after processing all images in an image group"
% self.image_name.value,
self.when_to_save,
)
# XXX - should check that if file_name_method is
# FN_FROM_IMAGE, that the named image actually has the
# required path measurement
# Make sure metadata tags exist
if self.file_name_method == FN_SINGLE_NAME or (
self.file_name_method == FN_FROM_IMAGE and self.wants_file_name_suffix.value
):
text_str = (
self.single_file_name.value
if self.file_name_method == FN_SINGLE_NAME
else self.file_name_suffix.value
)
undefined_tags = pipeline.get_undefined_metadata_tags(text_str)
if len(undefined_tags) > 0:
raise ValidationError(
"%s is not a defined metadata tag. Check the metadata specifications in your load modules"
% undefined_tags[0],
self.single_file_name
if self.file_name_method == FN_SINGLE_NAME
else self.file_name_suffix,
)
def volumetric(self):
return True
class ExportToCellH5(cpm.Module):
#
# TODO: model z and t. Currently, CellProfiler would analyze each
# stack plane independently. I think the easiest way to handle
# z and t would be to add them to the site path if they are
# used in the experiment (e.g. a time series would have a
# path of "/plate/well/site/time")
#
# I can add two more optional metadata keys that would let
# users capture this.
#
# The more-complicated choice would be to store the data in a
# stack which would mean reworking the indices in every segmentation
# after the first. There are some legacy measurements that are
# really object numbers, so these would require a lot of work
# to get right. Also, the resulting segmentations are a little
# artificial since they seem to say that every object is one
# pixel thick in the T or Z direction.
#
module_name = "ExportToCellH5"
variable_revision_number = 1
category = ["File Processing"]
SUBFILE_KEY = "subfile"
IGNORE_METADATA = "None"
def create_settings(self):
'''Create the settings for the ExportToCellH5 module'''
self.directory = Directory("Output file location",
doc="""
This setting lets you choose the folder for the output files.
%(IO_FOLDER_CHOICE_HELP_TEXT)s
""" % globals())
def get_directory_fn():
'''Get the directory for the CellH5 file'''
return self.directory.get_absolute_path()
def set_directory_fn(path):
dir_choice, custom_path = self.directory.get_parts_from_path(path)
self.directory.join_parts(dir_choice, custom_path)
self.file_name = cps.text.Filename("Output file name",
"DefaultOut.ch5",
get_directory_fn=get_directory_fn,
set_directory_fn=set_directory_fn,
metadata=True,
browse_msg="Choose CellH5 file",
mode=cps.text.Filename.MODE_APPEND,
exts=[("CellH5 file (*.cellh5)",
"*.ch5"),
("HDF5 file (*.h5)", "*.h5"),
("All files (*.*", "*.*")],
doc="""
This setting lets you name your CellH5 file. If you choose an
existing file, CellProfiler will add new data to the file
or overwrite existing locations.
<p>%(IO_WITH_METADATA_HELP_TEXT)s %(USING_METADATA_TAGS_REF)s.
For instance, if you have a metadata tag named
"Plate", you can create a per-plate folder by selecting one the subfolder options
and then specifying the subfolder name as "\g<Plate>". The module will
substitute the metadata values for the current image set for any metadata tags in the
folder name.%(USING_METADATA_HELP_REF)s.</p>
""" % globals())
self.overwrite_ok = cps.Binary(
"Overwrite existing data without warning?",
False,
doc="""
Select <i>"Yes"</i> to automatically overwrite any existing data
for a site. Select <i>"No"</i> to be prompted first.
If you are running the pipeline on a computing cluster,
select <i>"Yes"</i> unless you want execution to stop because you
will not be prompted to intervene. Also note that two instances
of CellProfiler cannot write to the same file at the same time,
so you must ensure that separate names are used on a cluster.
""" % globals())
self.repack = cps.Binary("Repack after analysis",
True,
doc="""
This setting determines whether CellProfiler in multiprocessing mode
repacks the data at the end of analysis. If you select <i>"Yes"</i>,
CellProfiler will combine all of the satellite files into a single
file upon completion. This option requires some extra temporary disk
space and takes some time at the end of analysis, but results in
a single file which may occupy less disk space. If you select
<i>"No"</i>, CellProfiler will create a master file using the
name that you give and this file will have links to individual
data files that contain the actual data. Using the data generated by
this option requires that you keep the master file and the linked
files together when copying them to a new folder.
""" % globals())
self.plate_metadata = Choice("Plate metadata", [],
value="Plate",
choices_fn=self.get_metadata_choices,
doc="""
This is the metadata tag that identifies the plate name of
the images for the current cycle. Choose <i>None</i> if
your assay does not have metadata for plate name. If your
assay is slide-based, you can use a metadata item that identifies
the slide as the choice for this setting and set the well
and site metadata items to <i>None</i>.""")
self.well_metadata = Choice(
"Well metadata", [],
value="Well",
choices_fn=self.get_metadata_choices,
doc="""This is the metadata tag that identifies the well name
for the images in the current cycle. Choose <i>None</i> if
your assay does not have metadata for the well.""")
self.site_metadata = Choice(
"Site metadata", [],
value="Site",
choices_fn=self.get_metadata_choices,
doc="""This is the metadata tag that identifies the site name
for the images in the current cycle. Choose <i>None</i> if
your assay doesn't divide wells up into sites or if this
tag is not required for other reasons.""")
self.divider = cps.Divider()
self.wants_to_choose_measurements = cps.Binary("Choose measurements?",
False,
doc="""
This setting lets you choose between exporting all measurements or
just the ones that you choose. Select <i>"Yes"</i> to pick the
measurements to be exported. Select <i>"No"</i> to automatically
export all measurements available at this stage of the pipeline.
""" % globals())
self.measurements = MeasurementMultiChoice("Measurements to export",
doc="""
<i>(Used only if choosing measurements.)</i>
<br>
This setting lets you choose individual measurements to be exported.
Check the measurements you want to export.
""")
self.objects_to_export = []
self.add_objects_button = DoSomething("Add objects to export",
"Add objects", self.add_objects)
self.images_to_export = []
self.add_image_button = DoSomething("Add an image to export",
"Add image", self.add_image)
self.objects_count = cps.HiddenCount(self.objects_to_export)
self.images_count = cps.HiddenCount(self.images_to_export)
def add_objects(self, can_delete=True):
group = cps.SettingsGroup()
self.objects_to_export.append(group)
group.append(
"objects_name",
LabelSubscriber("Objects name",
value="Nuclei",
doc="""
This setting lets you choose the objects you want to export.
<b>ExportToCellH5</b> will write the segmentation of the objects
to your CellH5 file so that they can be saved and used by other
applications that support the format.
"""))
group.append(
"Remover",
cps.do_something.RemoveSettingButton("Remove the objects above",
"Remove",
self.objects_to_export,
group))
def add_image(self, can_delete=True):
group = cps.SettingsGroup()
self.images_to_export.append(group)
group.append(
"image_name",
ImageSubscriber("Image name",
value="DNA",
doc="""
This setting lets you choose the images you want to export.
<b>ExportToCellH5</b> will write the image
to your CellH5 file so that it can be used by other
applications that support the format.
"""))
group.append(
"remover",
cps.do_something.RemoveSettingButton("Remove the image above",
"Remove",
self.objects_to_export,
group))
def get_metadata_choices(self, pipeline):
columns = pipeline.get_measurement_columns(self)
choices = [self.IGNORE_METADATA]
for column in columns:
object_name, feature_name, column_type = column[:3]
if object_name == "Image" and \
column_type.startswith(COLTYPE_VARCHAR) and \
feature_name.startswith(C_METADATA + "_"):
choices.append(feature_name.split("_", 1)[1])
return choices
def settings(self):
result = [
self.objects_count, self.images_count, self.directory,
self.file_name, self.overwrite_ok, self.repack,
self.plate_metadata, self.well_metadata, self.site_metadata,
self.wants_to_choose_measurements, self.measurements
]
for objects_group in self.objects_to_export:
result += objects_group.pipeline_settings()
for images_group in self.images_to_export:
result += images_group.pipeline_settings()
return result
def visible_settings(self):
result = [
self.directory, self.file_name, self.overwrite_ok, self.repack,
self.plate_metadata, self.well_metadata, self.site_metadata,
self.divider, self.wants_to_choose_measurements
]
if self.wants_to_choose_measurements:
result.append(self.measurements)
for group in self.objects_to_export:
result += group.visible_settings()
result.append(self.add_objects_button)
for group in self.images_to_export:
result += group.visible_settings()
result.append(self.add_image_button)
return result
def get_path_to_master_file(self, measurements):
return os.path.join(self.directory.get_absolute_path(measurements),
self.file_name.value)
def get_site_path(self, workspace, image_number):
'''Get the plate / well / site tuple that identifies a field of view
workspace - workspace for the analysis containing the metadata
measurements to be mined.
image_number - the image number for the field of view
returns a tuple which can be used for the hierarchical path
to the group for a particular field of view
'''
m = workspace.measurements
path = []
for setting in self.plate_metadata, self.well_metadata, self.site_metadata:
if setting.value == self.IGNORE_METADATA:
path.append("NA")
else:
feature = "_".join((C_METADATA, setting.value))
path.append(m["Image", feature, image_number])
return tuple(path)
def get_subfile_name(self, workspace):
'''Contact the UI to find the cellh5 file to use to store results
Internally, this tells the UI to create a link from the master file
to the plate / well / site group that will be used to store results.
Then, the worker writes into that file.
'''
master_file_name = self.get_path_to_master_file(workspace.measurements)
path = self.get_site_path(workspace,
workspace.measurements.image_set_number)
return workspace.interaction_request(self,
master_file_name,
os.getpid(),
path,
headless_ok=True)
def handle_interaction(self, master_file, pid, path):
'''Handle an analysis worker / UI interaction
This function is used to coordinate linking a group in the master file
with a group in a subfile that is reserved for a particular
analysis worker. Upon entry, the worker should be sure to have
flushed and closed its subfile.
master_file - the master cellh5 file which has links to groups
for each field of view
pid - the process ID or other unique identifier of the worker
talking to the master
path - The combination of (Plate, Well, Site) that should be used
as the folder path to the data.
returns the name of the subfile to be used. After return, the
subfile has been closed by the UI and a link has been established
to the group named by the path.
'''
master_dict = self.get_dictionary().setdefault(master_file, {})
if pid not in master_dict:
md_head, md_tail = os.path.splitext(master_file)
subfile = "%s_%s%s" % (md_head, str(pid), md_tail)
master_dict[pid] = subfile
else:
subfile = master_dict[pid]
ch5_master = cellh5.cellh5write.CH5MasterFile(master_file, "a")
try:
ch5_master.add_link_to_coord(self._to_ch5_coord(*path), subfile)
finally:
ch5_master.close()
return subfile
def _to_ch5_coord(self, plate, well, site):
return cellh5.CH5PositionCoordinate(plate, well, site)
def run(self, workspace):
m = workspace.measurements
object_set = workspace.object_set
#
# get plate / well / site as tuple
#
path = self.get_site_path(workspace, m.image_set_number)
subfile_name = self.get_subfile_name(workspace)
### create CellH5 file
with cellh5.cellh5write.CH5FileWriter(subfile_name,
mode="a") as c5_file:
### add Postion (==plate, well, site) triple
c5_pos = c5_file.add_position(self._to_ch5_coord(*path))
for ch_idx, object_group in enumerate(self.objects_to_export):
objects_name = object_group.objects_name.value
objects = object_set.get_objects(objects_name)
labels = objects.segmented
if ch_idx == 0:
### get shape of 5D cube
shape5D = (len(self.objects_to_export), 1, 1,
labels.shape[0], labels.shape[1])
dtype5D = np.uint16
### create lablel writer for incremental writing
c5_label_writer = c5_pos.add_label_image(shape=shape5D,
dtype=dtype5D)
c5_label_def = cellh5.cellh5write.CH5ImageRegionDefinition(
)
c5_label_writer.write(labels, c=ch_idx, t=0, z=0)
c5_label_def.add_row(region_name=objects_name,
channel_idx=ch_idx)
if len(self.objects_to_export) > 0:
### finalize the writer
c5_label_writer.write_definition(c5_label_def)
c5_label_writer.finalize()
n_channels = 0
max_scale = 1
max_i = 1
max_j = 1
for image_group in self.images_to_export:
image = m.get_image(image_group.image_name.value)
pixel_data = image.pixel_data
if pixel_data.ndim == 3:
n_channels += min(pixel_data.shape[2], 3)
else:
n_channels += 1
max_scale = max(image.scale, max_scale)
max_i = max(pixel_data.shape[0], max_i)
max_j = max(pixel_data.shape[1], max_j)
### get shape of 5D cube
shape5D = (n_channels, 1, 1, max_i, max_j)
for dtype in (np.uint8, np.uint16, np.uint32, np.uint64):
if max_scale <= np.iinfo(dtype).max:
dtype5D = dtype
break
### create image writer for incremental writing
c5_image_writer = c5_pos.add_image(shape=shape5D, dtype=dtype5D)
c5_image_def = cellh5.cellh5write.CH5ImageChannelDefinition()
ch_idx = 0
for image_group in self.images_to_export:
image_name = image_group.image_name.value
image = m.get_image(image_name).pixel_data
scale = m.get_image(image_name).scale
if not np.issubdtype(image.dtype, np.dtype(bool).type):
if scale == 1:
scale = max_scale
image = image * scale
if image.ndim == 3:
for c in range(min(image.shape[2], 3)):
color_name, html_color = COLORS[c]
c5_image_writer.write(image[:, :, c].astype(dtype5D),
c=ch_idx,
t=0,
z=0)
c5_image_def.add_row(channel_name="_".join(
(image_name, color_name)),
description="%s %s intensity" %
(image_name, color_name),
is_physical=True,
voxel_size=(1, 1, 1),
color=html_color)
ch_idx += 1
else:
c5_image_writer.write(image.astype(dtype5D),
c=ch_idx,
t=0,
z=0)
c5_image_def.add_row(channel_name=image_name,
description=image_name,
is_physical=True,
voxel_size=(1, 1, 1),
color="0xFFFFFF")
ch_idx += 1
c5_image_writer.write_definition(c5_image_def)
c5_image_writer.finalize()
columns = workspace.pipeline.get_measurement_columns(self)
if self.wants_to_choose_measurements:
to_keep = set([(self.measurements.get_measurement_object(s),
self.measurements.get_measurement_feature(s))
for s in self.measurements.selections])
def keep(column):
return (column[0], column[1]) in to_keep
columns = filter(keep, columns)
#
# I'm breaking the data up into the most granular form so that
# it's clearer how it's organized. I'm expecting that you would
# organize it differently when actually storing.
#
### 0) extract object information (i.e. object_label_no)
### 1) extract all single cell features and write it as feature matrix (for e.g. classification)
### 2) extract Center
### 3) create artifical Bounding box... usefull for displaying it in fiji lateron
### 4) Don't see the point of features extracted on "Image" the only real and useful feature there is "Count" which can be deduced from single cell information
### 0) and 1) filter columns for cellular features
feature_cols = filter(
lambda xxx:
(xxx[0] not in
(EXPERIMENT, "Image")) and m.has_feature(xxx[0], xxx[1]),
columns)
### iterate over objects to export
for ch_idx, object_group in enumerate(self.objects_to_export):
objects_name = object_group.objects_name.value
objects = object_set.get_objects(objects_name)
### find features for that object
feature_cols_per_object = filter(
lambda xxx: xxx[0] == objects_name, feature_cols)
c5_object_writer = c5_pos.add_region_object(objects_name)
object_labels = objects.indices
c5_object_writer.write(t=0,
object_labels=np.array(object_labels))
c5_object_writer.write_definition()
c5_object_writer.finalize()
### iterate over all cellular feature to get feature matrix
n_features = len(feature_cols_per_object)
if n_features > 0:
feature_names = []
feature_matrix = []
for column in feature_cols_per_object:
object_name, feature_name = column[:2]
values = m[object_name, feature_name]
feature_names.append(feature_name)
feature_matrix.append(values[:, np.newaxis])
feature_matrix = np.concatenate(feature_matrix, axis=1)
c5_feature_writer = c5_pos.add_object_feature_matrix(
object_name=object_name,
feature_name="object_features",
n_features=n_features,
dtype=np.float32)
c5_feature_writer.write(feature_matrix)
c5_feature_writer.write_definition(feature_names)
c5_feature_writer.finalize()
### iterate over Location to create bounding_box and center
c5_bbox = c5_pos.add_object_bounding_box(
object_name=objects_name)
if objects.count > 0:
ijv = objects.ijv
min_x = scipy.ndimage.minimum(ijv[:, 1], ijv[:, 2],
objects.indices)
max_x = scipy.ndimage.maximum(ijv[:, 1], ijv[:, 2],
objects.indices)
min_y = scipy.ndimage.minimum(ijv[:, 0], ijv[:, 2],
objects.indices)
max_y = scipy.ndimage.maximum(ijv[:, 0], ijv[:, 2],
objects.indices)
location_x = scipy.ndimage.mean(ijv[:, 1], ijv[:, 2],
objects.indices)
location_y = scipy.ndimage.mean(ijv[:, 0], ijv[:, 2],
objects.indices)
bb = np.c_[min_x, max_x, min_y, max_y]
else:
bb = np.zeros((0, 4))
location_x = np.zeros(0)
location_y = np.zeros(0)
c5_bbox.write(bb.astype(np.int32))
c5_bbox.write_definition()
c5_bbox.finalize()
c5_center = c5_pos.add_object_center(object_name=objects_name)
locations = {'x': location_x, 'y': location_y}
cent = np.column_stack(
[locations[axis] for axis in c5_center.dtype.names])
c5_center.write(cent.astype(np.int32))
c5_center.write_definition()
c5_center.finalize()
#
# The last part deals with relationships between segmentations.
# The most typical relationship is "Parent" which is explained below,
# but you can also have things like first nearest and second nearest
# neighbor or in tracking, the relationship between the segmentation
# of the previous and next frames.
#
for key in m.get_relationship_groups():
relationships = m.get_relationships(key.module_number,
key.relationship,
key.object_name1,
key.object_name2,
[m.image_set_number])
for image_number1, image_number2, \
object_number1, object_number2 in relationships:
if image_number1 == image_number2 and \
key.relationship == R_PARENT:
#
# Object 1 is the parent to object 2 - this is the
# most common relationship, so if you can only record
# one, this is it. "Parent" usually means that
# the child's segmentation was seeded by the parent
# segmentation (e.g. Parent = nucleus, child = cell),
# but can also be something like Parent = cell,
# child = all organelles within the cell
#
# object_name1 is the name of the parent segmentation
# object_name2 is the name of the child segmentation
# object_number1 is the index used to label the
# parent in the parent segmentation
# object_number2 is the index used to label the
# child in the child segmentation
continue
if image_number1 != m.image_set_number:
path1 = self.get_site_path(workspace, image_number1)
else:
path1 = path
if image_number2 != m.image_set_number:
path2 = self.get_site_path(workspace, image_number2)
else:
path2 = path
#
# TODO: this is sort of extra credit, but the relationships
# relate an object in one segmentation to another.
# For tracking, these can be in different image
# sets, (e.g. the cell at time T and at time T+1).
# So, given object 1 and object 2, path1 and path2
# tell you how the objects are related between planes.
pass
def post_run(self, workspace):
if self.repack:
### to be implemented with
### ch5_master.repack()
return
measurements = workspace.measurements
fd, temp_name = tempfile.mkstemp(
suffix=".ch5", dir=self.directory.get_absolute_path())
master_name = self.get_path_to_master_file(workspace.measurements)
src = h5py.File(master_name, "r")
dest = h5py.File(temp_name)
os.close(fd)
for key in src:
dest.copy(src[key], dest, expand_external=True)
src.close()
dest.close()
os.unlink(master_name)
os.rename(temp_name, master_name)
def prepare_settings(self, setting_values):
objects_count, images_count = [int(x) for x in setting_values[:2]]
del self.objects_to_export[:]
while len(self.objects_to_export) < objects_count:
self.add_objects()
del self.images_to_export[:]
while len(self.images_to_export) < images_count:
self.add_image()
class RunImageJScript(Module):
"""
Module to run ImageJ scripts via pyimagej
"""
module_name = "RunImageJScript"
variable_revision_number = 2
category = "Advanced"
def __init__(self):
super().__init__()
self.parsed_params = False # Used for validation
self.initialization_failed = False # Used for validation
def create_settings(self):
module_explanation = [
"The" + self.module_name + "module allows you to run any supported ImageJ script as part of your workflow.",
"First, select your desired initialization method and specify the app directory or endpoint(s) if needed.",
"Then select a script file to be executed by this module.",
"Click the \"Get parameters from script\" button to detect required inputs for your script:",
"each input will have its own setting created, allowing you to pass data from CellProfiler to ImageJ.",
"After filling in any required inputs you can run the module normally.",
"Note: ImageJ will only be initialized once per CellProfiler session.",
"Note: only numeric, text and image parameters are currently supported.",
"See also ImageJ Scripting: https://imagej.net/Scripting."
]
self.set_notes([" ".join(module_explanation)])
self.init_choice = Choice(
"Initialization type", [INIT_LOCAL, INIT_ENDPOINT, INIT_LATEST],
tooltips={INIT_LOCAL: "Use a local ImageJ/Fiji installation", INIT_ENDPOINT: "Specify a particular endpoint",
INIT_LATEST: "Use the latest Fiji, downloading if needed."},
doc="""\
Note that initialization will only occur once per CellProfiler session! After initialization, these options will be
locked for the remainder of the session.
Select the mechanism for initializing ImageJ:
* {init_local}: Use a local Fiji or ImageJ installation
* {init_endpoint}: Precisely specify the version of one or more components
* {init_latest}: Use the latest Fiji version
Note that any option besides {init_local} may result in a download of the requested components.
""".format(
init_local=INIT_LOCAL,
init_endpoint=INIT_ENDPOINT,
init_latest=INIT_LATEST,
),
)
self.endpoint_string = Text(
"Initialization endpoint", "sc.fiji:fiji:2.1.0", doc="""\
Specify an initialization string as described in https://github.com/imagej/pyimagej/blob/master/doc/Initialization.md
""",
)
self.initialized_method = Text("Initialization type", value="Do not use", doc="""\
Indicates the method that was used to initialized ImageJ in this CellProfiler session.
""",
)
self.convert_types = Binary("Adjust image type?", True, doc="""\
If enabled, ensures images are always converted to unsigned integer types when sent to ImageJ, and back to signed float types when returned to CellProfiler.
This can help common display issues by providing each application a best guess at its "expected" data type.
If you choose to disable this function, your ImageJ script will need to account for images coming in as signed float types.
""",
)
global init_display_string
if init_display_string:
# ImageJ thread is already running
self.initialized_method.set_value(init_display_string)
self.app_directory = Directory(
"ImageJ directory", allow_metadata=False, doc="""\
Select the folder containing the desired ImageJ/Fiji application.
{fcht}
""".format(
fcht=IO_FOLDER_CHOICE_HELP_TEXT
),
)
if platform != 'darwin':
self.app_directory.join_parts(ABSOLUTE_FOLDER_NAME, "Fiji.app")
def set_directory_fn_app(path):
dir_choice, custom_path = self.app_directory.get_parts_from_path(path)
self.app_directory.join_parts(dir_choice, custom_path)
self.app_file = Filename(
"Local App", "Fiji.app", doc="Select the desired app, such as Fiji.app",
get_directory_fn=self.app_directory.get_absolute_path,
set_directory_fn=set_directory_fn_app,
browse_msg="Choose local application"
)
self.script_directory = Directory(
"Script directory",
allow_metadata=False,
doc="""\
Select the folder containing the script.
{fcht}
""".format(
fcht=IO_FOLDER_CHOICE_HELP_TEXT
),
)
def set_directory_fn_script(script_path):
dir_choice, custom_path = self.script_directory.get_parts_from_path(script_path)
self.script_directory.join_parts(dir_choice, custom_path)
self.clear_script_parameters()
self.script_file = Filename(
"ImageJ Script", "script.py", doc="Select a script file written in any ImageJ-supported scripting language.",
get_directory_fn=self.script_directory.get_absolute_path,
set_directory_fn=set_directory_fn_script,
browse_msg="Choose ImageJ script file",
)
self.get_parameters_button = DoSomething("", 'Get parameters from script',
self.get_parameters_helper,
doc="""\
Parse parameters from the currently selected script and add the appropriate settings to this CellProfiler module.
Note: this must be done each time you change the script, before running the CellProfiler pipeline!
"""
)
self.script_parameter_list = []
self.script_input_settings = {} # Map of input parameter names to CellProfiler settings objects
self.script_output_settings = {} # Map of output parameter names to CellProfiler settings objects
self.script_parameter_count = HiddenCount(self.script_parameter_list)
def get_init_string(self):
"""
Determine if a particular initialization method has been specified. This could be a path to a local installation
or a version string.
"""
choice = self.init_choice.get_value()
if choice == INIT_LATEST:
return None
if choice == INIT_LOCAL:
init_string = self.app_directory.get_absolute_path()
if platform == 'darwin':
init_string = path.join(init_string, self.app_file.value)
elif choice == INIT_ENDPOINT:
init_string = self.endpoint_string.get_value()
return init_string
def close_pyimagej(self):
"""
Close the pyimagej daemon thread
"""
global imagej_process, to_imagej
if imagej_process is not None:
to_imagej.put({PYIMAGEJ_KEY_COMMAND: PYIMAGEJ_CMD_EXIT})
def init_pyimagej(self):
"""
Start the pyimagej daemon thread if it isn't already running.
"""
self.initialization_failed = False
init_string = self.get_init_string()
global imagej_process, to_imagej, from_imagej, init_display_string
if imagej_process is None:
to_imagej = mp.Queue()
from_imagej = mp.Queue()
# TODO if needed we could set daemon=True
imagej_process = Process(target=start_imagej_process, name="PyImageJ Daemon",
args=(to_imagej, from_imagej, init_string,))
imagej_process.start()
result = from_imagej.get()
if result == PYIMAGEJ_STATUS_STARTUP_FAILED:
imagej_process = None
self.initialization_failed = True
else:
atexit.register(self.close_pyimagej) # TODO is there a more CP-ish way to do this?
init_display_string = self.init_choice.get_value()
if init_display_string != INIT_LATEST:
init_display_string += ": " + init_string
self.initialized_method.set_value(init_display_string)
def clear_script_parameters(self):
"""
Remove any existing settings added by scripts
"""
self.script_parameter_list.clear()
self.script_input_settings.clear()
self.script_output_settings.clear()
self.parsed_params = False
self.initialization_failed = False
def get_parameters_helper(self):
"""
Helper method to launch get_parameters_from_script on a thread so that it isn't run on the GUI thread, since
it may be slow (when initializing pyimagej).
"""
# Reset previously parsed parameters
self.clear_script_parameters()
global stop_progress_thread
stop_progress_thread = False
progress_gauge = Gauge(Window.FindFocus(), -1, size=(100, -1))
progress_gauge.Show(True)
parse_param_thread = Thread(target=self.get_parameters_from_script, name="Parse Parameters Thread", daemon=True)
parse_param_thread.start()
while True:
# Wait for get_parameters_from_script to finish
progress_gauge.Pulse()
time.sleep(0.025)
if stop_progress_thread:
progress_gauge.Show(False)
break
if not self.initialization_failed:
self.parsed_params = True
def get_parameters_from_script(self):
"""
Use PyImageJ to read header text from an ImageJ script and extract inputs/outputs, which are then converted to
CellProfiler settings for this module
"""
global stop_progress_thread, imagej_process, to_imagej, from_imagej
script_filepath = path.join(self.script_directory.get_absolute_path(), self.script_file.value)
if not self.script_file.value or not path.exists(script_filepath):
# nothing to do
stop_progress_thread = True
return
# Start pyimagej if needed
self.init_pyimagej()
if not imagej_process:
stop_progress_thread = True
return
# Tell pyimagej to parse the script parameters
to_imagej.put({PYIMAGEJ_KEY_COMMAND: PYIMAGEJ_CMD_SCRIPT_PARSE, PYIMAGEJ_KEY_INPUT: script_filepath})
ij_return = from_imagej.get()
# Process pyimagej's output, converting script parameters to settings
if ij_return != PYIMAGEJ_STATUS_CMD_UNKNOWN:
input_params = ij_return[PYIMAGEJ_SCRIPT_PARSE_INPUTS]
output_params = ij_return[PYIMAGEJ_SCRIPT_PARSE_OUTPUTS]
for param_dict, settings_dict, io_class in ((input_params, self.script_input_settings, INPUT_CLASS),
(output_params, self.script_output_settings, OUTPUT_CLASS)):
for param_name in param_dict:
param_type = param_dict[param_name]
next_setting = convert_java_type_to_setting(param_name, param_type, io_class)
if next_setting is not None:
settings_dict[param_name] = next_setting
group = SettingsGroup()
group.append("setting", next_setting)
group.append("remover", RemoveSettingButton("", "Remove this variable",
self.script_parameter_list, group))
add_param_info_settings(group, param_name, param_type, io_class)
# Each setting gets a group containing:
# 0 - the setting
# 1 - its remover
# 2 - (hidden) parameter name
# 3 - (hidden) parameter type
# 4 - (hidden) parameter i/o class
self.script_parameter_list.append(group)
stop_progress_thread = True
def settings(self):
result = [self.script_parameter_count, self.init_choice, self.app_directory, self.app_file, self.endpoint_string, self.script_directory, self.script_file, self.get_parameters_button, self.convert_types]
if len(self.script_parameter_list) > 0:
result += [Divider(line=True)]
for script_parameter_group in self.script_parameter_list:
result += [script_parameter_group.setting]
result += [script_parameter_group.remover]
result += [script_parameter_group.name]
result += [script_parameter_group.type]
result += [script_parameter_group.io_class]
return result
def visible_settings(self):
visible_settings = []
global imagej_process
# Update the visible settings based on the selected initialization method
# If ImageJ is already initialized we just want to report how it was initialized
# Otherwise we show: a string entry for "endpoint", a directory chooser for "local" (and file chooser if on mac),
# and nothing if "latest"
if not imagej_process:
visible_settings += [self.init_choice]
input_type = self.init_choice.get_value()
# ImageJ is not initialized yet
if input_type == INIT_ENDPOINT:
visible_settings += [self.endpoint_string]
elif input_type == INIT_LOCAL:
visible_settings += [self.app_directory]
if platform == 'darwin':
visible_settings += [self.app_file]
else:
# ImageJ is initialized
visible_settings += [self.initialized_method]
visible_settings += [Divider(line=True)]
visible_settings += [self.script_directory, self.script_file, self.get_parameters_button, self.convert_types]
if len(self.script_parameter_list) > 0:
visible_settings += [Divider(line=True)]
for script_parameter in self.script_parameter_list:
visible_settings += [script_parameter.setting]
visible_settings += [script_parameter.remover]
return visible_settings
def prepare_settings(self, setting_values):
settings_count = int(setting_values[0])
if settings_count == 0:
# No params were saved
return
# Params were parsed previously and saved
self.parsed_params = True
# Looking at the last 5N elements will give the us (value, remover, name, type, io_class) for the N settings
# We care about the name and type information, since this goes in one of our settings
settings_info = setting_values[-settings_count * 5:]
for i in range(0, len(settings_info), 5):
group = SettingsGroup()
param_name = settings_info[i + 2]
param_type = settings_info[i + 3]
io_class = settings_info[i + 4]
setting = convert_java_type_to_setting(param_name, param_type, io_class)
group.append("setting", setting)
group.append("remover", RemoveSettingButton("", "Remove this variable", self.script_parameter_list, group))
add_param_info_settings(group, param_name, param_type, io_class)
self.script_parameter_list.append(group)
if INPUT_CLASS == io_class:
self.script_input_settings[param_name] = setting
elif OUTPUT_CLASS == io_class:
self.script_output_settings[param_name] = setting
def validate_module(self, pipeline):
if self.initialization_failed:
raise ValidationError(
"Error starting ImageJ. Please check your initialization settings and try again.",
self.init_choice
)
no_script_msg = "Please select a valid ImageJ script and use the \"Get parameters from script\" button."
if not self.parsed_params or not self.script_directory or not self.script_file.value:
raise ValidationError(
no_script_msg,
self.script_file
)
script_filepath = path.join(self.script_directory.get_absolute_path(), self.script_file.value)
if not path.exists(script_filepath):
raise ValidationError(
"The script you have selected is not a valid path. " + no_script_msg,
self.script_file
)
if self.init_choice.get_value() == INIT_LOCAL:
app_path = self.get_init_string()
if not path.exists(app_path):
raise ValidationError(
"The local application you have selected is not a valid path.",
self.app_directory
)
def validate_module_warnings(self, pipeline):
global imagej_process
"""Warn user if the specified FIJI executable directory is not found, and warn that a copy of FIJI will be downloaded"""
warn_msg = "Please note: any initialization method except \"Local\", a new Fiji may be downloaded"
" to your machine if cached dependencies not found."
init_type = self.init_choice.get_value()
if init_type != INIT_LOCAL:
# The component we attach the error to depends on if initialization has happened or not
if not imagej_process:
raise ValidationError(warn_msg, self.init_choice)
else:
raise ValidationError(warn_msg + " If re-initialization is required, please restart CellProfiler.",
self.initialized_method)
def run(self, workspace):
self.init_pyimagej()
if self.show_window:
workspace.display_data.script_input_pixels = {}
workspace.display_data.script_input_dimensions = {}
workspace.display_data.script_output_pixels = {}
workspace.display_data.script_output_dimensions = {}
script_filepath = path.join(self.script_directory.get_absolute_path(), self.script_file.value)
# convert the CP settings to script parameters for pyimagej
script_inputs = {}
for name in self.script_input_settings:
setting = self.script_input_settings[name]
if isinstance(setting, ImageSubscriber):
# Images need to be pulled from the workspace
script_inputs[name] = workspace.image_set.get_image(setting.get_value())
if self.show_window:
workspace.display_data.script_input_pixels[name] = script_inputs[name].pixel_data
workspace.display_data.script_input_dimensions[name] = script_inputs[name].dimensions
else:
# Other settings can be read directly
script_inputs[name] = setting.get_value()
# Start the script
to_imagej.put({PYIMAGEJ_KEY_COMMAND: PYIMAGEJ_CMD_SCRIPT_RUN, PYIMAGEJ_KEY_INPUT:
{PYIMAGEJ_SCRIPT_RUN_FILE_KEY: script_filepath,
PYIMAGEJ_SCRIPT_RUN_INPUT_KEY: script_inputs,
PYIMAGEJ_SCRIPT_RUN_CONVERT_IMAGES: self.convert_types.value}
})
# Retrieve script output
ij_return = from_imagej.get()
if ij_return != PYIMAGEJ_STATUS_CMD_UNKNOWN:
script_outputs = ij_return[PYIMAGEJ_KEY_OUTPUT]
for name in self.script_output_settings:
output_key = self.script_output_settings[name].get_value()
output_value = script_outputs[name]
# convert back to floats for CellProfiler
if self.convert_types.value:
output_value = skimage.img_as_float(output_value)
output_image = Image(image=output_value, convert=False)
workspace.image_set.add(output_key, output_image)
if self.show_window:
workspace.display_data.script_output_pixels[name] = output_image.pixel_data
workspace.display_data.dimensions = output_image.dimensions
def display(self, workspace, figure):
# TODO how do we handle differences in dimensionality between input/output images?
figure.set_subplots((2, max(len(workspace.display_data.script_input_pixels),
len(workspace.display_data.script_output_pixels))), dimensions=2)
i = 0
for name in workspace.display_data.script_input_pixels:
figure.subplot_imshow_grayscale(
0,
i,
workspace.display_data.script_input_pixels[name],
title="Input image: {}".format(name),
)
i += 1
i = 0
for name in workspace.display_data.script_output_pixels:
figure.subplot_imshow_grayscale(
1,
i,
workspace.display_data.script_output_pixels[name],
title="Output image: {}".format(name),
)
i += 1
def upgrade_settings(self, setting_values, variable_revision_number, module_name):
if variable_revision_number == 1:
# Added convert_types Binary setting
setting_values = setting_values[:8]+[True]+setting_values[8:]
variable_revision_number = 2
return setting_values, variable_revision_number
class FilterObjects(ObjectProcessing):
module_name = "FilterObjects"
variable_revision_number = 8
def create_settings(self):
super(FilterObjects, self).create_settings()
self.x_name.text = """Select the objects to filter"""
self.x_name.doc = """\
Select the set of objects that you want to filter. This setting also
controls which measurement choices appear for filtering: you can only
filter based on measurements made on the object you select. Be sure
the **FilterObjects** module is downstream of the necessary **Measure**
modules. If you
intend to use a measurement calculated by the **CalculateMath** module
to to filter objects, select the first operand’s object here, because
**CalculateMath** measurements are stored with the first operand’s
object."""
self.y_name.text = """Name the output objects"""
self.y_name.doc = "Enter a name for the collection of objects that are retained after applying the filter(s)."
self.spacer_1 = Divider(line=False)
self.mode = Choice(
"Select the filtering mode",
[MODE_MEASUREMENTS, MODE_RULES, MODE_BORDER, MODE_CLASSIFIERS],
doc="""\
You can choose from the following options:
- *{MODE_MEASUREMENTS}*: Specify a per-object measurement made by an
upstream module in the pipeline.
- *{MODE_BORDER}*: Remove objects touching the border of the image
and/or the edges of an image mask.
- *{MODE_RULES}*: Use a file containing rules generated by
CellProfiler Analyst. You will need to ensure that the measurements
specified by the rules file are produced by upstream modules in the
pipeline. This setting is not compatible with data processed as 3D.
- *{MODE_CLASSIFIERS}*: Use a file containing a trained classifier from
CellProfiler Analyst. You will need to ensure that the measurements
specified by the file are produced by upstream modules in the
pipeline. This setting is not compatible with data processed as 3D.""".format(
**{
"MODE_MEASUREMENTS": MODE_MEASUREMENTS,
"MODE_RULES": MODE_RULES,
"MODE_BORDER": MODE_BORDER,
"MODE_CLASSIFIERS": MODE_CLASSIFIERS,
}
),
)
self.spacer_2 = Divider(line=False)
self.measurements = []
self.measurement_count = HiddenCount(self.measurements, "Measurement count")
self.add_measurement(False)
self.add_measurement_button = DoSomething(
"", "Add another measurement", self.add_measurement
)
self.filter_choice = Choice(
"Select the filtering method",
FI_ALL,
FI_LIMITS,
doc="""\
*(Used only if filtering using measurements)*
There are five different ways to filter objects:
- *{FI_LIMITS}:* Keep an object if its measurement value falls within
a range you specify.
- *{FI_MAXIMAL}:* Keep the object with the maximum value for the
measurement of interest. If multiple objects share a maximal value,
retain one object selected arbitrarily per image.
- *{FI_MINIMAL}:* Keep the object with the minimum value for the
measurement of interest. If multiple objects share a minimal value,
retain one object selected arbitrarily per image.
- *{FI_MAXIMAL_PER_OBJECT}:* This option requires you to choose a
parent object. The parent object might contain several child objects
of choice (for instance, mitotic spindles within a cell or FISH probe
spots within a nucleus). Only the child object whose measurements
equal the maximum child-measurement value among that set of child
objects will be kept (for example, the longest spindle in each cell).
You do not have to explicitly relate objects before using this
module.
- *{FI_MINIMAL_PER_OBJECT}:* Same as *Maximal per object*, except
filtering is based on the minimum value.""".format(
**{
"FI_LIMITS": FI_LIMITS,
"FI_MAXIMAL": FI_MAXIMAL,
"FI_MINIMAL": FI_MINIMAL,
"FI_MAXIMAL_PER_OBJECT": FI_MAXIMAL_PER_OBJECT,
"FI_MINIMAL_PER_OBJECT": FI_MINIMAL_PER_OBJECT,
}
),
)
self.per_object_assignment = Choice(
"Assign overlapping child to",
PO_ALL,
doc="""\
*(Used only if filtering per object)*
A child object can overlap two parent objects and can have the
maximal/minimal measurement of all child objects in both parents. This
option controls how an overlapping maximal/minimal child affects
filtering of other children of its parents and to which parent the
maximal child is assigned. The choices are:
- *{PO_BOTH}*: The child will be assigned to both parents and all
other children of both parents will be filtered. Only the maximal
child per parent will be left, but if **RelateObjects** is used to
relate the maximal child to its parent, one or the other of the
overlapping parents will not have a child even though the excluded
parent may have other child objects. The maximal child can still be
assigned to both parents using a database join via the relationships
table if you are using **ExportToDatabase** and separate object
tables.
- *{PO_PARENT_WITH_MOST_OVERLAP}*: The child will be assigned to
the parent with the most overlap and a child with a less
maximal/minimal measurement, if available, will be assigned to other
parents. Use this option to ensure that parents with an alternate
non-overlapping child object are assigned some child object by a
subsequent **RelateObjects** module.""".format(
**{
"PO_BOTH": PO_BOTH,
"PO_PARENT_WITH_MOST_OVERLAP": PO_PARENT_WITH_MOST_OVERLAP,
}
),
)
self.enclosing_object_name = LabelSubscriber(
"Select the objects that contain the filtered objects",
"None",
doc="""\
*(Used only if a per-object filtering method is selected)*
This setting selects the container (i.e., parent) objects for the
*{FI_MAXIMAL_PER_OBJECT}* and *{FI_MINIMAL_PER_OBJECT}* filtering
choices.""".format(
**{
"FI_MAXIMAL_PER_OBJECT": FI_MAXIMAL_PER_OBJECT,
"FI_MINIMAL_PER_OBJECT": FI_MINIMAL_PER_OBJECT,
}
),
)
self.rules_directory = Directory(
"Select the location of the rules or classifier file",
doc="""\
*(Used only when filtering using {MODE_RULES} or {MODE_CLASSIFIERS})*
Select the location of the rules or classifier file that will be used for
filtering.
{IO_FOLDER_CHOICE_HELP_TEXT}
""".format(
**{
"MODE_CLASSIFIERS": MODE_CLASSIFIERS,
"MODE_RULES": MODE_RULES,
"IO_FOLDER_CHOICE_HELP_TEXT": _help.IO_FOLDER_CHOICE_HELP_TEXT,
}
),
)
self.rules_class = Choice(
"Class number",
choices=["1", "2"],
choices_fn=self.get_class_choices,
doc="""\
*(Used only when filtering using {MODE_RULES} or {MODE_CLASSIFIERS})*
Select which of the classes to keep when filtering. The CellProfiler
Analyst classifier user interface lists the names of the classes in
left-to-right order. **FilterObjects** uses the first class from
CellProfiler Analyst if you choose “1”, etc.
Please note the following:
- The object is retained if the object falls into the selected class.
- You can make multiple class selections. If you do so, the module will
retain the object if the object falls into any of the selected
classes.""".format(
**{"MODE_CLASSIFIERS": MODE_CLASSIFIERS, "MODE_RULES": MODE_RULES}
),
)
def get_directory_fn():
"""Get the directory for the rules file name"""
return self.rules_directory.get_absolute_path()
def set_directory_fn(path):
dir_choice, custom_path = self.rules_directory.get_parts_from_path(path)
self.rules_directory.join_parts(dir_choice, custom_path)
self.rules_file_name = Filename(
"Rules or classifier file name",
"rules.txt",
get_directory_fn=get_directory_fn,
set_directory_fn=set_directory_fn,
doc="""\
*(Used only when filtering using {MODE_RULES} or {MODE_CLASSIFIERS})*
The name of the rules or classifier file.
A rules file is a plain text file containing the complete set of rules.
Each line of the rules file should be a rule naming a measurement to be made
on the object you selected, for instance:
IF (Nuclei_AreaShape_Area < 351.3, [0.79, -0.79], [-0.94, 0.94])
The above rule will score +0.79 for the positive category and -0.94
for the negative category for nuclei whose area is less than 351.3
pixels and will score the opposite for nuclei whose area is larger.
The filter adds positive and negative and keeps only objects whose
positive score is higher than the negative score.
A classifier file is a trained classifier exported from CellProfiler Analyst.
You will need to ensure that the measurements specified by the file are
produced by upstream modules in the pipeline. This setting is not compatible
with data processed as 3D.
""".format(
**{"MODE_CLASSIFIERS": MODE_CLASSIFIERS, "MODE_RULES": MODE_RULES}
),
)
self.additional_objects = []
self.additional_object_count = HiddenCount(
self.additional_objects, "Additional object count"
)
self.spacer_3 = Divider(line=False)
self.additional_object_button = DoSomething(
"Relabel additional objects to match the filtered object?",
"Add an additional object",
self.add_additional_object,
doc="""\
Click this button to add an object to receive the same post-filtering labels as
the filtered object. This is useful in making sure that labeling is maintained
between related objects (e.g., primary and secondary objects) after filtering.""",
)
def get_class_choices(self, pipeline):
if self.mode == MODE_CLASSIFIERS:
return self.get_bin_labels()
elif self.mode == MODE_RULES:
rules = self.get_rules()
nclasses = len(rules.rules[0].weights[0])
return [str(i) for i in range(1, nclasses + 1)]
def get_rules_class_choices(self, pipeline):
try:
rules = self.get_rules()
nclasses = len(rules.rules[0].weights[0])
return [str(i) for i in range(1, nclasses + 1)]
except:
return [str(i) for i in range(1, 3)]
def add_measurement(self, can_delete=True):
"""Add another measurement to the filter list"""
group = SettingsGroup()
group.append(
"measurement",
Measurement(
"Select the measurement to filter by",
self.x_name.get_value,
"AreaShape_Area",
doc="""\
*(Used only if filtering using {MODE_MEASUREMENTS})*
See the **Measurements** modules help pages for more information on the
features measured.""".format(
**{"MODE_MEASUREMENTS": MODE_MEASUREMENTS}
),
),
)
group.append(
"wants_minimum",
Binary(
"Filter using a minimum measurement value?",
True,
doc="""\
*(Used only if {FI_LIMITS} is selected for filtering method)*
Select "*{YES}*" to filter the objects based on a minimum acceptable
object measurement value. Objects which are greater than or equal to
this value will be retained.""".format(
**{"FI_LIMITS": FI_LIMITS, "YES": "Yes"}
),
),
)
group.append("min_limit", Float("Minimum value", 0))
group.append(
"wants_maximum",
Binary(
"Filter using a maximum measurement value?",
True,
doc="""\
*(Used only if {FI_LIMITS} is selected for filtering method)*
Select "*{YES}*" to filter the objects based on a maximum acceptable
object measurement value. Objects which are less than or equal to this
value will be retained.""".format(
**{"FI_LIMITS": FI_LIMITS, "YES": "Yes"}
),
),
)
group.append("max_limit", Float("Maximum value", 1))
group.append("divider", Divider())
self.measurements.append(group)
if can_delete:
group.append(
"remover",
RemoveSettingButton(
"", "Remove this measurement", self.measurements, group
),
)
def add_additional_object(self):
group = SettingsGroup()
group.append(
"object_name",
LabelSubscriber("Select additional object to relabel", "None"),
)
group.append(
"target_name", LabelName("Name the relabeled objects", "FilteredGreen"),
)
group.append(
"remover",
RemoveSettingButton(
"", "Remove this additional object", self.additional_objects, group
),
)
group.append("divider", Divider(line=False))
self.additional_objects.append(group)
def prepare_settings(self, setting_values):
"""Make sure the # of slots for additional objects matches
the anticipated number of additional objects"""
additional_object_count = int(setting_values[ADDITIONAL_OBJECT_SETTING_INDEX])
while len(self.additional_objects) > additional_object_count:
self.remove_additional_object(self.additional_objects[-1].key)
while len(self.additional_objects) < additional_object_count:
self.add_additional_object()
measurement_count = int(setting_values[MEASUREMENT_COUNT_SETTING_INDEX])
while len(self.measurements) > measurement_count:
del self.measurements[-1]
while len(self.measurements) < measurement_count:
self.add_measurement()
def settings(self):
settings = super(FilterObjects, self).settings()
settings += [
self.mode,
self.filter_choice,
self.enclosing_object_name,
self.rules_directory,
self.rules_file_name,
self.rules_class,
self.measurement_count,
self.additional_object_count,
self.per_object_assignment,
]
for x in self.measurements:
settings += x.pipeline_settings()
for x in self.additional_objects:
settings += [x.object_name, x.target_name]
return settings
def help_settings(self):
return [
self.x_name,
self.y_name,
self.mode,
self.filter_choice,
self.per_object_assignment,
self.rules_directory,
self.rules_file_name,
self.rules_class,
self.enclosing_object_name,
self.additional_object_button,
]
def visible_settings(self):
visible_settings = super(FilterObjects, self).visible_settings()
visible_settings += [self.spacer_2, self.mode]
if self.mode == MODE_RULES or self.mode == MODE_CLASSIFIERS:
visible_settings += [
self.rules_file_name,
self.rules_directory,
self.rules_class,
]
self.rules_class.text = (
"Class number" if self.mode == MODE_RULES else "Class name"
)
try:
self.rules_class.test_valid(None)
except:
pass
elif self.mode == MODE_MEASUREMENTS:
visible_settings += [self.spacer_1, self.filter_choice]
if self.filter_choice in (FI_MINIMAL, FI_MAXIMAL):
visible_settings += [
self.measurements[0].measurement,
self.measurements[0].divider,
]
elif self.filter_choice in (FI_MINIMAL_PER_OBJECT, FI_MAXIMAL_PER_OBJECT):
visible_settings += [
self.per_object_assignment,
self.measurements[0].measurement,
self.enclosing_object_name,
self.measurements[0].divider,
]
elif self.filter_choice == FI_LIMITS:
for i, group in enumerate(self.measurements):
visible_settings += [group.measurement, group.wants_minimum]
if group.wants_minimum:
visible_settings.append(group.min_limit)
visible_settings.append(group.wants_maximum)
if group.wants_maximum.value:
visible_settings.append(group.max_limit)
if i > 0:
visible_settings += [group.remover]
visible_settings += [group.divider]
visible_settings += [self.add_measurement_button]
visible_settings.append(self.spacer_3)
for x in self.additional_objects:
visible_settings += x.visible_settings()
visible_settings += [self.additional_object_button]
return visible_settings
def validate_module(self, pipeline):
"""Make sure that the user has selected some limits when filtering"""
if self.mode == MODE_MEASUREMENTS and self.filter_choice == FI_LIMITS:
for group in self.measurements:
if not (group.wants_minimum.value or group.wants_maximum.value):
raise ValidationError(
"Please enter a minimum and/or maximum limit for your measurement",
group.wants_minimum,
)
if self.mode == MODE_RULES:
try:
rules = self.get_rules()
except Exception as instance:
logging.warning(
"Failed to load rules: %s", str(instance), exc_info=True
)
raise ValidationError(str(instance), self.rules_file_name)
measurement_columns = pipeline.get_measurement_columns(self)
for r in rules.rules:
if not any(
[
mc[0] == r.object_name and mc[1] == r.feature
for mc in measurement_columns
]
):
raise ValidationError(
(
"The rules file, %s, uses the measurement, %s "
"for object %s, but that measurement is not available "
"at this stage of the pipeline. Consider editing the "
"rules to match the available measurements or adding "
"measurement modules to produce the measurement."
)
% (self.rules_file_name, r.feature, r.object_name),
self.rules_file_name,
)
elif self.mode == MODE_CLASSIFIERS:
try:
self.get_classifier()
self.get_bin_labels()
self.get_classifier_features()
except IOError:
raise ValidationError(
"Failed to load classifier file %s" % self.rules_file_name.value,
self.rules_file_name,
)
except:
raise ValidationError(
"Unable to load %s as a classifier file"
% self.rules_file_name.value,
self.rules_file_name,
)
def run(self, workspace):
"""Filter objects for this image set, display results"""
src_objects = workspace.get_objects(self.x_name.value)
if self.mode == MODE_RULES:
indexes = self.keep_by_rules(workspace, src_objects)
elif self.mode == MODE_MEASUREMENTS:
if self.filter_choice in (FI_MINIMAL, FI_MAXIMAL):
indexes = self.keep_one(workspace, src_objects)
if self.filter_choice in (FI_MINIMAL_PER_OBJECT, FI_MAXIMAL_PER_OBJECT):
indexes = self.keep_per_object(workspace, src_objects)
if self.filter_choice == FI_LIMITS:
indexes = self.keep_within_limits(workspace, src_objects)
elif self.mode == MODE_BORDER:
indexes = self.discard_border_objects(src_objects)
elif self.mode == MODE_CLASSIFIERS:
indexes = self.keep_by_class(workspace, src_objects)
else:
raise ValueError("Unknown filter choice: %s" % self.mode.value)
#
# Create an array that maps label indexes to their new values
# All labels to be deleted have a value in this array of zero
#
new_object_count = len(indexes)
max_label = numpy.max(src_objects.segmented)
label_indexes = numpy.zeros((max_label + 1,), int)
label_indexes[indexes] = numpy.arange(1, new_object_count + 1)
#
# Loop over both the primary and additional objects
#
object_list = [(self.x_name.value, self.y_name.value)] + [
(x.object_name.value, x.target_name.value) for x in self.additional_objects
]
m = workspace.measurements
for src_name, target_name in object_list:
src_objects = workspace.get_objects(src_name)
target_labels = src_objects.segmented.copy()
#
# Reindex the labels of the old source image
#
target_labels[target_labels > max_label] = 0
target_labels = label_indexes[target_labels]
#
# Make a new set of objects - retain the old set's unedited
# segmentation for the new and generally try to copy stuff
# from the old to the new.
#
target_objects = cellprofiler_core.object.Objects()
target_objects.segmented = target_labels
target_objects.unedited_segmented = src_objects.unedited_segmented
#
# Remove the filtered objects from the small_removed_segmented
# if present. "small_removed_segmented" should really be
# "filtered_removed_segmented".
#
small_removed = src_objects.small_removed_segmented.copy()
small_removed[(target_labels == 0) & (src_objects.segmented != 0)] = 0
target_objects.small_removed_segmented = small_removed
if src_objects.has_parent_image:
target_objects.parent_image = src_objects.parent_image
workspace.object_set.add_objects(target_objects, target_name)
self.add_measurements(workspace, src_name, target_name)
if self.show_window:
workspace.display_data.src_objects_segmented = src_objects.segmented
workspace.display_data.target_objects_segmented = target_objects.segmented
workspace.display_data.dimensions = src_objects.dimensions
def display(self, workspace, figure):
"""Display what was filtered"""
src_name = self.x_name.value
src_objects_segmented = workspace.display_data.src_objects_segmented
target_objects_segmented = workspace.display_data.target_objects_segmented
dimensions = workspace.display_data.dimensions
target_name = self.y_name.value
figure.set_subplots((2, 2), dimensions=dimensions)
figure.subplot_imshow_labels(
0, 0, src_objects_segmented, title="Original: %s" % src_name
)
figure.subplot_imshow_labels(
1,
0,
target_objects_segmented,
title="Filtered: %s" % target_name,
sharexy=figure.subplot(0, 0),
)
statistics = [
[numpy.max(src_objects_segmented)],
[numpy.max(target_objects_segmented)],
]
figure.subplot_table(
0,
1,
statistics,
row_labels=(
"Number of objects pre-filtering",
"Number of objects post-filtering",
),
)
def keep_one(self, workspace, src_objects):
"""Return an array containing the single object to keep
workspace - workspace passed into Run
src_objects - the Objects instance to be filtered
"""
measurement = self.measurements[0].measurement.value
src_name = self.x_name.value
values = workspace.measurements.get_current_measurement(src_name, measurement)
if len(values) == 0:
return numpy.array([], int)
best_idx = (
numpy.argmax(values)
if self.filter_choice == FI_MAXIMAL
else numpy.argmin(values)
) + 1
return numpy.array([best_idx], int)
def keep_per_object(self, workspace, src_objects):
"""Return an array containing the best object per enclosing object
workspace - workspace passed into Run
src_objects - the Objects instance to be filtered
"""
measurement = self.measurements[0].measurement.value
src_name = self.x_name.value
enclosing_name = self.enclosing_object_name.value
src_objects = workspace.get_objects(src_name)
enclosing_objects = workspace.get_objects(enclosing_name)
enclosing_labels = enclosing_objects.segmented
enclosing_max = enclosing_objects.count
if enclosing_max == 0:
return numpy.array([], int)
enclosing_range = numpy.arange(1, enclosing_max + 1)
#
# Make a vector of the value of the measurement per label index.
# We can then label each pixel in the image with the measurement
# value for the object at that pixel.
# For unlabeled pixels, put the minimum value if looking for the
# maximum value and vice-versa
#
values = workspace.measurements.get_current_measurement(src_name, measurement)
wants_max = self.filter_choice == FI_MAXIMAL_PER_OBJECT
src_labels = src_objects.segmented
src_count = src_objects.count
if self.per_object_assignment == PO_PARENT_WITH_MOST_OVERLAP:
#
# Find the number of overlapping pixels in enclosing
# and source objects
#
mask = enclosing_labels * src_labels != 0
enclosing_labels = enclosing_labels[mask]
src_labels = src_labels[mask]
order = numpy.lexsort((enclosing_labels, src_labels))
src_labels = src_labels[order]
enclosing_labels = enclosing_labels[order]
firsts = numpy.hstack(
(
[0],
numpy.where(
(src_labels[:-1] != src_labels[1:])
| (enclosing_labels[:-1] != enclosing_labels[1:])
)[0]
+ 1,
[len(src_labels)],
)
)
areas = firsts[1:] - firsts[:-1]
enclosing_labels = enclosing_labels[firsts[:-1]]
src_labels = src_labels[firsts[:-1]]
#
# Re-sort by source label value and area descending
#
if wants_max:
svalues = -values
else:
svalues = values
order = numpy.lexsort((-areas, svalues[src_labels - 1]))
src_labels, enclosing_labels, areas = [
x[order] for x in (src_labels, enclosing_labels, areas)
]
firsts = numpy.hstack(
(
[0],
numpy.where(src_labels[:-1] != src_labels[1:])[0] + 1,
src_labels.shape[:1],
)
)
counts = firsts[1:] - firsts[:-1]
#
# Process them in order. The maximal or minimal child
# will be assigned to the most overlapping parent and that
# parent will be excluded.
#
best_src_label = numpy.zeros(enclosing_max + 1, int)
for idx, count in zip(firsts[:-1], counts):
for i in range(count):
enclosing_object_number = enclosing_labels[idx + i]
if best_src_label[enclosing_object_number] == 0:
best_src_label[enclosing_object_number] = src_labels[idx]
break
#
# Remove best source labels = 0 and sort to get the list
#
best_src_label = best_src_label[best_src_label != 0]
best_src_label.sort()
return best_src_label
else:
tricky_values = numpy.zeros((len(values) + 1,))
tricky_values[1:] = values
if wants_max:
tricky_values[0] = -numpy.Inf
else:
tricky_values[0] = numpy.Inf
src_values = tricky_values[src_labels]
#
# Now find the location of the best for each of the enclosing objects
#
fn = (
scipy.ndimage.maximum_position
if wants_max
else scipy.ndimage.minimum_position
)
best_pos = fn(src_values, enclosing_labels, enclosing_range)
best_pos = numpy.array(
(best_pos,) if isinstance(best_pos, tuple) else best_pos
)
best_pos = best_pos.astype(numpy.uint32)
#
# Get the label of the pixel at each location
#
indexes = src_labels[best_pos.transpose().tolist()]
indexes = set(indexes)
indexes = list(indexes)
indexes.sort()
return indexes[1:] if len(indexes) > 0 and indexes[0] == 0 else indexes
def keep_within_limits(self, workspace, src_objects):
"""Return an array containing the indices of objects to keep
workspace - workspace passed into Run
src_objects - the Objects instance to be filtered
"""
src_name = self.x_name.value
hits = None
m = workspace.measurements
for group in self.measurements:
measurement = group.measurement.value
values = m.get_current_measurement(src_name, measurement)
if hits is None:
hits = numpy.ones(len(values), bool)
elif len(hits) < len(values):
temp = numpy.ones(len(values), bool)
temp[~hits] = False
hits = temp
low_limit = group.min_limit.value
high_limit = group.max_limit.value
if group.wants_minimum.value:
hits[values < low_limit] = False
if group.wants_maximum.value:
hits[values > high_limit] = False
indexes = numpy.argwhere(hits)[:, 0]
indexes = indexes + 1
return indexes
def discard_border_objects(self, src_objects):
"""Return an array containing the indices of objects to keep
workspace - workspace passed into Run
src_objects - the Objects instance to be filtered
"""
labels = src_objects.segmented
if src_objects.has_parent_image and src_objects.parent_image.has_mask:
mask = src_objects.parent_image.mask
interior_pixels = scipy.ndimage.binary_erosion(mask)
else:
interior_pixels = scipy.ndimage.binary_erosion(numpy.ones_like(labels))
border_pixels = numpy.logical_not(interior_pixels)
border_labels = set(labels[border_pixels])
if (
border_labels == {0}
and src_objects.has_parent_image
and src_objects.parent_image.has_mask
):
# The assumption here is that, if nothing touches the border,
# the mask is a large, elliptical mask that tells you where the
# well is. That's the way the old Matlab code works and it's duplicated here
#
# The operation below gets the mask pixels that are on the border of the mask
# The erosion turns all pixels touching an edge to zero. The not of this
# is the border + formerly masked-out pixels.
mask = src_objects.parent_image.mask
interior_pixels = scipy.ndimage.binary_erosion(mask)
border_pixels = numpy.logical_not(interior_pixels)
border_labels = set(labels[border_pixels])
return list(set(labels.ravel()).difference(border_labels))
def get_rules(self):
"""Read the rules from a file"""
rules_file = self.rules_file_name.value
rules_directory = self.rules_directory.get_absolute_path()
path = os.path.join(rules_directory, rules_file)
if not os.path.isfile(path):
raise ValidationError("No such rules file: %s" % path, self.rules_file_name)
else:
rules = cellprofiler.utilities.rules.Rules()
rules.parse(path)
return rules
def load_classifier(self):
"""Load the classifier pickle if not cached
returns classifier, bin_labels, name and features
"""
d = self.get_dictionary()
file_ = self.rules_file_name.value
directory_ = self.rules_directory.get_absolute_path()
path_ = os.path.join(directory_, file_)
if path_ not in d:
if not os.path.isfile(path_):
raise ValidationError(
"No such classifier file: %s" % path_, self.rules_file_name
)
else:
import joblib
d[path_] = joblib.load(path_)
return d[path_]
def get_classifier(self):
return self.load_classifier()[0]
def get_bin_labels(self):
return self.load_classifier()[1]
def get_classifier_features(self):
return self.load_classifier()[3]
def keep_by_rules(self, workspace, src_objects):
"""Keep objects according to rules
workspace - workspace holding the measurements for the rules
src_objects - filter these objects (uses measurement indexes instead)
Open the rules file indicated by the settings and score the
objects by the rules. Return the indexes of the objects that pass.
"""
rules = self.get_rules()
rules_class = int(self.rules_class.value) - 1
scores = rules.score(workspace.measurements)
if len(scores) > 0:
is_not_nan = numpy.any(~numpy.isnan(scores), 1)
best_class = numpy.argmax(scores[is_not_nan], 1).flatten()
hits = numpy.zeros(scores.shape[0], bool)
hits[is_not_nan] = best_class == rules_class
indexes = numpy.argwhere(hits).flatten() + 1
else:
indexes = numpy.array([], int)
return indexes
def keep_by_class(self, workspace, src_objects):
""" Keep objects according to their predicted class
:param workspace: workspace holding the measurements for the rules
:param src_objects: filter these objects (uses measurement indexes instead)
:return: indexes (base 1) of the objects that pass
"""
classifier = self.get_classifier()
target_idx = self.get_bin_labels().index(self.rules_class.value)
target_class = classifier.classes_[target_idx]
features = []
for feature_name in self.get_classifier_features():
feature_name = feature_name.split("_", 1)[1]
if feature_name == "x_loc":
feature_name = M_LOCATION_CENTER_X
elif feature_name == "y_loc":
feature_name = M_LOCATION_CENTER_Y
features.append(feature_name)
feature_vector = numpy.column_stack(
[
workspace.measurements[self.x_name.value, feature_name]
for feature_name in features
]
)
predicted_classes = classifier.predict(feature_vector)
hits = predicted_classes == target_class
indexes = numpy.argwhere(hits) + 1
return indexes.flatten()
def get_measurement_columns(self, pipeline):
return super(FilterObjects, self).get_measurement_columns(
pipeline,
additional_objects=[
(x.object_name.value, x.target_name.value)
for x in self.additional_objects
],
)
def prepare_to_create_batch(self, workspace, fn_alter_path):
"""Prepare to create a batch file
This function is called when CellProfiler is about to create a
file for batch processing. It will pickle the image set list's
"legacy_fields" dictionary. This callback lets a module prepare for
saving.
pipeline - the pipeline to be saved
image_set_list - the image set list to be saved
fn_alter_path - this is a function that takes a pathname on the local
host and returns a pathname on the remote host. It
handles issues such as replacing backslashes and
mapping mountpoints. It should be called for every
pathname stored in the settings or legacy fields.
"""
self.rules_directory.alter_for_create_batch_files(fn_alter_path)
return True
def upgrade_settings(self, setting_values, variable_revision_number, module_name):
if variable_revision_number == 1:
#
# Added CPA rules
#
setting_values = (
setting_values[:11]
+ [MODE_MEASUREMENTS, DEFAULT_INPUT_FOLDER_NAME, ".",]
+ setting_values[11:]
)
variable_revision_number = 2
if variable_revision_number == 2:
#
# Forgot file name (???!!!)
#
setting_values = setting_values[:14] + ["rules.txt"] + setting_values[14:]
variable_revision_number = 3
if variable_revision_number == 3:
#
# Allowed multiple measurements
# Structure changed substantially.
#
(
target_name,
object_name,
measurement,
filter_choice,
enclosing_objects,
wants_minimum,
minimum_value,
wants_maximum,
maximum_value,
wants_outlines,
outlines_name,
rules_or_measurements,
rules_directory_choice,
rules_path_name,
rules_file_name,
) = setting_values[:15]
additional_object_settings = setting_values[15:]
additional_object_count = len(additional_object_settings) // 4
setting_values = [
target_name,
object_name,
rules_or_measurements,
filter_choice,
enclosing_objects,
wants_outlines,
outlines_name,
rules_directory_choice,
rules_path_name,
rules_file_name,
"1",
str(additional_object_count),
measurement,
wants_minimum,
minimum_value,
wants_maximum,
maximum_value,
] + additional_object_settings
variable_revision_number = 4
if variable_revision_number == 4:
#
# Used Directory to combine directory choice & custom path
#
rules_directory_choice = setting_values[7]
rules_path_name = setting_values[8]
if rules_directory_choice == DIR_CUSTOM:
rules_directory_choice = ABSOLUTE_FOLDER_NAME
if rules_path_name.startswith("."):
rules_directory_choice = DEFAULT_INPUT_SUBFOLDER_NAME
elif rules_path_name.startswith("&"):
rules_directory_choice = DEFAULT_OUTPUT_SUBFOLDER_NAME
rules_path_name = "." + rules_path_name[1:]
rules_directory = Directory.static_join_string(
rules_directory_choice, rules_path_name
)
setting_values = setting_values[:7] + [rules_directory] + setting_values[9:]
variable_revision_number = 5
if variable_revision_number == 5:
#
# added rules class
#
setting_values = setting_values[:9] + ["1"] + setting_values[9:]
variable_revision_number = 6
if variable_revision_number == 6:
#
# Added per-object assignment
#
setting_values = (
setting_values[:FIXED_SETTING_COUNT_V6]
+ [PO_BOTH]
+ setting_values[FIXED_SETTING_COUNT_V6:]
)
variable_revision_number = 7
if variable_revision_number == 7:
x_name = setting_values[1]
y_name = setting_values[0]
measurement_count = int(setting_values[10])
additional_object_count = int(setting_values[11])
n_measurement_settings = measurement_count * 5
additional_object_settings = setting_values[13 + n_measurement_settings :]
additional_object_names = additional_object_settings[::4]
additional_target_names = additional_object_settings[1::4]
new_additional_object_settings = sum(
[
[object_name, target_name]
for object_name, target_name in zip(
additional_object_names, additional_target_names
)
],
[],
)
setting_values = (
[x_name, y_name]
+ setting_values[2:5]
+ setting_values[7 : 13 + n_measurement_settings]
+ new_additional_object_settings
)
variable_revision_number = 8
slot_directory = 5
setting_values[slot_directory] = Directory.upgrade_setting(
setting_values[slot_directory]
)
return setting_values, variable_revision_number
class Noise2Void(ImageProcessing):
module_name = "Noise2Void"
variable_revision_number = 1
def create_settings(self):
super(Noise2Void, self).create_settings()
self.ml_model = Directory("Path to ML Model",
doc="""\
Select the folder containing the machine learning model to be used.
This model has to be generated via the noise2void training. See
https://github.com/juglab/n2v/blob/master/examples/2D/denoising2D_RGB/01_training.ipynb
for an example of training.
"""
)
self.color = Binary("Process as color image?",
value=False,
doc="""\
Select whether your image should be processed as a color image or not.
""")
self.manual_slicing = Binary("Slice Image manually?",
value=False, doc="""\
If necessary, **Noise2Void** will slice your image into tiles automatically for a better memory fit.
If you want to manually determine the size of the said tiles, check this setting.
Colored images **do not** support custom slicing as of right now!
""")
self.slicing_configuration = Text("Tile size", value="(2,2,2)", doc="""\
You can provide an image slicing configuration for Noise2Void for a better memory fit.
Specify your custom slicing configuration as follows:
- (x,y) for 2D Images
- (x,y,z) for 3D Images, whereas x,y and z are positive integers.
If your input cannot be parsed, no slicing configuration will be provided to n2v.
""")
self.axes_configuration = Text(text="N2V Axes", value=N2V_AXES_3D, doc="""\
For internal use only.
Communicates axes configuration (2D or 3D, color or not) to n2v.
""")
self.x_name.doc = """\
This is the image that the module operates on. You can choose any image
that is made available by a prior module.
**Noise2Void** will denoise this image using a tensorflow based neural network.
"""
def settings(self):
settings = super(Noise2Void, self).settings()
return settings + [self.ml_model, self.slicing_configuration, self.color, self.axes_configuration]
def visible_settings(self):
visible_settings = super(Noise2Void, self).visible_settings()
visible_settings += [self.ml_model, self.color]
if not self.color:
visible_settings += [self.manual_slicing]
if self.manual_slicing:
visible_settings += [self.slicing_configuration]
return visible_settings
#
# This is the function that gets called during "run" to create the output image.
#
def denoise(self, pixels, ml_model, final_tile_choice, color, axes):
path = self.ml_model.get_absolute_path()
(basedir, model_name) = split(path)
try:
model = N2V(config=None, name=model_name, basedir=basedir)
except FileNotFoundError as e:
raise FileNotFoundError(
'Path ' + path + ' doesn\'t lead to valid model') from e
if self.manual_slicing:
tile_tuple = self.convert_string_to_tuple(final_tile_choice)
if color or not self.manual_slicing or tile_tuple == None:
axes = self.adjust_for_color(axes)
pred = model.predict(pixels, axes=axes)
else:
pred = model.predict(pixels, axes=axes, n_tiles=tile_tuple)
return pred
def run(self, workspace):
image = workspace.image_set.get_image(self.x_name.value)
self.adjust_settings_for_dimensionality(image.volumetric)
self.function = self.denoise
super(Noise2Void, self).run(workspace)
def volumetric(self):
return True
def adjust_settings_for_dimensionality(self, image_is_3d_in_workspace):
if image_is_3d_in_workspace:
self.axes_configuration.value = N2V_AXES_3D
else:
self.axes_configuration.value = N2V_AXES_2D
def adjust_for_color(self, axes):
axes.replace(N2V_AXES_COLOR, '')
if self.color:
axes += N2V_AXES_COLOR
return axes
def convert_string_to_tuple(self, text):
try:
text = text.strip()
text = text.replace('(', '')
text = text.replace(')', '')
text = text.replace(' ', '')
return tuple(map(int, text.split(',')))
except ValueError:
return None
class MeasureObjectSkeleton(Module):
module_name = "MeasureObjectSkeleton"
category = "Measurement"
variable_revision_number = 3
def create_settings(self):
"""Create the UI settings for the module"""
self.seed_objects_name = LabelSubscriber(
"Select the seed objects",
"None",
doc="""\
Select the previously identified objects that you want to use as the
seeds for measuring branches and distances. Branches and trunks are assigned
per seed object. Seed objects are typically not single points/pixels but
instead are usually objects of varying sizes.""",
)
self.image_name = ImageSubscriber(
"Select the skeletonized image",
"None",
doc="""\
Select the skeletonized image of the dendrites and/or axons as produced
by the **Morph** module’s *Skel* operation.""",
)
self.wants_branchpoint_image = Binary(
"Retain the branchpoint image?",
False,
doc="""\
Select "*Yes*" if you want to save the color image of branchpoints and
trunks. This is the image that is displayed in the output window for
this module."""
% globals(),
)
self.branchpoint_image_name = ImageName(
"Name the branchpoint image",
"BranchpointImage",
doc="""\
*(Used only if a branchpoint image is to be retained)*
Enter a name for the branchpoint image here. You can then use this image
in a later module, such as **SaveImages**.""",
)
self.wants_to_fill_holes = Binary(
"Fill small holes?",
True,
doc="""\
The algorithm reskeletonizes the image and this can leave artifacts
caused by small holes in the image prior to skeletonizing. These holes
result in false trunks and branchpoints. Select "*Yes*" to fill in
these small holes prior to skeletonizing."""
% globals(),
)
self.maximum_hole_size = Integer(
"Maximum hole size",
10,
minval=1,
doc="""\
*(Used only when filling small holes)*
This is the area of the largest hole to fill, measured in pixels. The
algorithm will fill in any hole whose area is this size or smaller.""",
)
self.wants_objskeleton_graph = Binary(
"Export the skeleton graph relationships?",
False,
doc="""\
Select "*Yes*" to produce an edge file and a vertex file that gives the
relationships between vertices (trunks, branchpoints and endpoints)."""
% globals(),
)
self.intensity_image_name = ImageSubscriber(
"Intensity image",
"None",
doc="""\
Select the image to be used to calculate
the total intensity along the edges between the vertices (trunks, branchpoints, and endpoints).""",
)
self.directory = Directory(
"File output directory",
doc="Select the directory you want to save the graph relationships to.",
dir_choices=[
DEFAULT_OUTPUT_FOLDER_NAME,
DEFAULT_INPUT_FOLDER_NAME,
ABSOLUTE_FOLDER_NAME,
DEFAULT_OUTPUT_SUBFOLDER_NAME,
DEFAULT_INPUT_SUBFOLDER_NAME,
],
)
self.directory.dir_choice = DEFAULT_OUTPUT_FOLDER_NAME
self.vertex_file_name = Text(
"Vertex file name",
"vertices.csv",
doc="""\
*(Used only when exporting graph relationships)*
Enter the name of the file that will hold the edge information. You can
use metadata tags in the file name.
Each line of the file is a row of comma-separated values. The first
row is the header; this names the file’s columns. Each subsequent row
represents a vertex in the skeleton graph: either a trunk, a
branchpoint or an endpoint. The file has the following columns:
- *image\_number:* The image number of the associated image.
- *vertex\_number:* The number of the vertex within the image.
- *i:* The I coordinate of the vertex.
- *j:* The J coordinate of the vertex.
- *label:* The label of the seed object associated with the vertex.
- *kind:* The vertex type, with the following choices:
- **T:** Trunk
- **B:** Branchpoint
- **E:** Endpoint
""",
)
self.edge_file_name = Text(
"Edge file name",
"edges.csv",
doc="""\
*(Used only when exporting graph relationships)*
Enter the name of the file that will hold the edge information. You can
use metadata tags in the file name. Each line of the file is a row of
comma-separated values. The first row is the header; this names the
file’s columns. Each subsequent row represents an edge or connection
between two vertices (including between a vertex and itself for certain
loops). Note that vertices include trunks, branchpoints, and endpoints.
The file has the following columns:
- *image\_number:* The image number of the associated image.
- *v1:* The zero-based index into the vertex table of the first vertex
in the edge.
- *v2:* The zero-based index into the vertex table of the second vertex
in the edge.
- *length:* The number of pixels in the path connecting the two
vertices, including both vertex pixels.
- *total\_intensity:* The sum of the intensities of the pixels in the
edge, including both vertex pixel intensities.
""",
)
def settings(self):
"""The settings, in the order that they are saved in the pipeline"""
return [
self.seed_objects_name,
self.image_name,
self.wants_branchpoint_image,
self.branchpoint_image_name,
self.wants_to_fill_holes,
self.maximum_hole_size,
self.wants_objskeleton_graph,
self.intensity_image_name,
self.directory,
self.vertex_file_name,
self.edge_file_name,
]
def visible_settings(self):
"""The settings that are displayed in the GUI"""
result = [self.seed_objects_name, self.image_name, self.wants_branchpoint_image]
if self.wants_branchpoint_image:
result += [self.branchpoint_image_name]
result += [self.wants_to_fill_holes]
if self.wants_to_fill_holes:
result += [self.maximum_hole_size]
result += [self.wants_objskeleton_graph]
if self.wants_objskeleton_graph:
result += [
self.intensity_image_name,
self.directory,
self.vertex_file_name,
self.edge_file_name,
]
return result
def get_graph_file_paths(self, m, image_number):
"""Get the paths to the graph files for the given image set
Apply metadata tokens to the graph file names to get the graph files
for the given image set.
m - measurements for the run
image_number - the image # for the current image set
Returns the edge file's path and vertex file's path
"""
path = self.directory.get_absolute_path(m)
edge_file = m.apply_metadata(self.edge_file_name.value, image_number)
edge_path = os.path.abspath(os.path.join(path, edge_file))
vertex_file = m.apply_metadata(self.vertex_file_name.value, image_number)
vertex_path = os.path.abspath(os.path.join(path, vertex_file))
return edge_path, vertex_path
VF_IMAGE_NUMBER = "image_number"
VF_VERTEX_NUMBER = "vertex_number"
VF_I = "i"
VF_J = "j"
VF_LABELS = "labels"
VF_KIND = "kind"
vertex_file_columns = (
VF_IMAGE_NUMBER,
VF_VERTEX_NUMBER,
VF_I,
VF_J,
VF_LABELS,
VF_KIND,
)
EF_IMAGE_NUMBER = "image_number"
EF_V1 = "v1"
EF_V2 = "v2"
EF_LENGTH = "length"
EF_TOTAL_INTENSITY = "total_intensity"
edge_file_columns = (EF_IMAGE_NUMBER, EF_V1, EF_V2, EF_LENGTH, EF_TOTAL_INTENSITY)
def prepare_run(self, workspace):
"""Initialize graph files"""
if not self.wants_objskeleton_graph:
return True
edge_files = set()
vertex_files = set()
m = workspace.measurements
assert isinstance(m, Measurements)
for image_number in m.get_image_numbers():
edge_path, vertex_path = self.get_graph_file_paths(m, image_number)
edge_files.add(edge_path)
vertex_files.add(vertex_path)
for file_path, header in (
(edge_path, self.edge_file_columns),
(vertex_path, self.vertex_file_columns),
):
if os.path.exists(file_path):
import wx
if (
wx.MessageBox(
"%s already exists. Do you want to overwrite it?" % file_path,
"Warning: overwriting file",
style=wx.YES_NO,
parent=workspace.frame,
)
!= wx.YES
):
return False
os.remove(file_path)
fd = open(file_path, "wt")
header = ",".join(header)
fd.write(header + "\n")
fd.close()
return True
def run(self, workspace):
"""Run the module on the image set"""
seed_objects_name = self.seed_objects_name.value
skeleton_name = self.image_name.value
seed_objects = workspace.object_set.get_objects(seed_objects_name)
labels = seed_objects.segmented
labels_count = numpy.max(labels)
label_range = numpy.arange(labels_count, dtype=numpy.int32) + 1
skeleton_image = workspace.image_set.get_image(
skeleton_name, must_be_binary=True
)
skeleton = skeleton_image.pixel_data
if skeleton_image.has_mask:
skeleton = skeleton & skeleton_image.mask
try:
labels = skeleton_image.crop_image_similarly(labels)
except:
labels, m1 = size_similarly(skeleton, labels)
labels[~m1] = 0
#
# The following code makes a ring around the seed objects with
# the skeleton trunks sticking out of it.
#
# Create a new skeleton with holes at the seed objects
# First combine the seed objects with the skeleton so
# that the skeleton trunks come out of the seed objects.
#
# Erode the labels once so that all of the trunk branchpoints
# will be within the labels
#
#
# Dilate the objects, then subtract them to make a ring
#
my_disk = centrosome.cpmorphology.strel_disk(1.5).astype(int)
dilated_labels = grey_dilation(labels, footprint=my_disk)
seed_mask = dilated_labels > 0
combined_skel = skeleton | seed_mask
closed_labels = grey_erosion(dilated_labels, footprint=my_disk)
seed_center = closed_labels > 0
combined_skel = combined_skel & (~seed_center)
#
# Fill in single holes (but not a one-pixel hole made by
# a one-pixel image)
#
if self.wants_to_fill_holes:
def size_fn(area, is_object):
return (~is_object) and (area <= self.maximum_hole_size.value)
combined_skel = centrosome.cpmorphology.fill_labeled_holes(
combined_skel, ~seed_center, size_fn
)
#
# Reskeletonize to make true branchpoints at the ring boundaries
#
combined_skel = centrosome.cpmorphology.skeletonize(combined_skel)
#
# The skeleton outside of the labels
#
outside_skel = combined_skel & (dilated_labels == 0)
#
# Associate all skeleton points with seed objects
#
dlabels, distance_map = propagate.propagate(
numpy.zeros(labels.shape), dilated_labels, combined_skel, 1
)
#
# Get rid of any branchpoints not connected to seeds
#
combined_skel[dlabels == 0] = False
#
# Find the branchpoints
#
branch_points = centrosome.cpmorphology.branchpoints(combined_skel)
#
# Odd case: when four branches meet like this, branchpoints are not
# assigned because they are arbitrary. So assign them.
#
# . .
# B.
# .B
# . .
#
odd_case = (
combined_skel[:-1, :-1]
& combined_skel[1:, :-1]
& combined_skel[:-1, 1:]
& combined_skel[1, 1]
)
branch_points[:-1, :-1][odd_case] = True
branch_points[1:, 1:][odd_case] = True
#
# Find the branching counts for the trunks (# of extra branches
# emanating from a point other than the line it might be on).
#
branching_counts = centrosome.cpmorphology.branchings(combined_skel)
branching_counts = numpy.array([0, 0, 0, 1, 2])[branching_counts]
#
# Only take branches within 1 of the outside skeleton
#
dilated_skel = scipy.ndimage.binary_dilation(
outside_skel, centrosome.cpmorphology.eight_connect
)
branching_counts[~dilated_skel] = 0
#
# Find the endpoints
#
end_points = centrosome.cpmorphology.endpoints(combined_skel)
#
# We use two ranges for classification here:
# * anything within one pixel of the dilated image is a trunk
# * anything outside of that range is a branch
#
nearby_labels = dlabels.copy()
nearby_labels[distance_map > 1.5] = 0
outside_labels = dlabels.copy()
outside_labels[nearby_labels > 0] = 0
#
# The trunks are the branchpoints that lie within one pixel of
# the dilated image.
#
if labels_count > 0:
trunk_counts = fix(
scipy.ndimage.sum(branching_counts, nearby_labels, label_range)
).astype(int)
else:
trunk_counts = numpy.zeros((0,), int)
#
# The branches are the branchpoints that lie outside the seed objects
#
if labels_count > 0:
branch_counts = fix(
scipy.ndimage.sum(branch_points, outside_labels, label_range)
)
else:
branch_counts = numpy.zeros((0,), int)
#
# Save the endpoints
#
if labels_count > 0:
end_counts = fix(scipy.ndimage.sum(end_points, outside_labels, label_range))
else:
end_counts = numpy.zeros((0,), int)
#
# Calculate the distances
#
total_distance = centrosome.cpmorphology.skeleton_length(
dlabels * outside_skel, label_range
)
#
# Save measurements
#
m = workspace.measurements
assert isinstance(m, Measurements)
feature = "_".join((C_OBJSKELETON, F_NUMBER_TRUNKS, skeleton_name))
m.add_measurement(seed_objects_name, feature, trunk_counts)
feature = "_".join((C_OBJSKELETON, F_NUMBER_NON_TRUNK_BRANCHES, skeleton_name))
m.add_measurement(seed_objects_name, feature, branch_counts)
feature = "_".join((C_OBJSKELETON, F_NUMBER_BRANCH_ENDS, skeleton_name))
m.add_measurement(seed_objects_name, feature, end_counts)
feature = "_".join((C_OBJSKELETON, F_TOTAL_OBJSKELETON_LENGTH, skeleton_name))
m[seed_objects_name, feature] = total_distance
#
# Collect the graph information
#
if self.wants_objskeleton_graph:
trunk_mask = (branching_counts > 0) & (nearby_labels != 0)
intensity_image = workspace.image_set.get_image(
self.intensity_image_name.value
)
edge_graph, vertex_graph = self.make_objskeleton_graph(
combined_skel,
dlabels,
trunk_mask,
branch_points & ~trunk_mask,
end_points,
intensity_image.pixel_data,
)
image_number = workspace.measurements.image_set_number
edge_path, vertex_path = self.get_graph_file_paths(m, m.image_number)
workspace.interaction_request(
self,
m.image_number,
edge_path,
edge_graph,
vertex_path,
vertex_graph,
headless_ok=True,
)
if self.show_window:
workspace.display_data.edge_graph = edge_graph
workspace.display_data.vertex_graph = vertex_graph
workspace.display_data.intensity_image = intensity_image.pixel_data
#
# Make the display image
#
if self.show_window or self.wants_branchpoint_image:
branchpoint_image = numpy.zeros((skeleton.shape[0], skeleton.shape[1], 3))
trunk_mask = (branching_counts > 0) & (nearby_labels != 0)
branch_mask = branch_points & (outside_labels != 0)
end_mask = end_points & (outside_labels != 0)
branchpoint_image[outside_skel, :] = 1
branchpoint_image[trunk_mask | branch_mask | end_mask, :] = 0
branchpoint_image[trunk_mask, 0] = 1
branchpoint_image[branch_mask, 1] = 1
branchpoint_image[end_mask, 2] = 1
branchpoint_image[dilated_labels != 0, :] *= 0.875
branchpoint_image[dilated_labels != 0, :] += 0.1
if self.show_window:
workspace.display_data.branchpoint_image = branchpoint_image
if self.wants_branchpoint_image:
bi = Image(branchpoint_image, parent_image=skeleton_image)
workspace.image_set.add(self.branchpoint_image_name.value, bi)
def handle_interaction(
self, image_number, edge_path, edge_graph, vertex_path, vertex_graph
):
columns = tuple(
[vertex_graph[f].tolist() for f in self.vertex_file_columns[2:]]
)
with open(vertex_path, "at") as fd:
for vertex_number, fields in enumerate(zip(*columns)):
fd.write(
("%d,%d," % (image_number, vertex_number + 1))
+ ("%d,%d,%d,%s\n" % fields)
)
columns = tuple([edge_graph[f].tolist() for f in self.edge_file_columns[1:]])
with open(edge_path, "at") as fd:
line_format = "%d,%%d,%%d,%%d,%%.4f\n" % image_number
for fields in zip(*columns):
fd.write(line_format % fields)
def display(self, workspace, figure):
"""Display a visualization of the results"""
from matplotlib.axes import Axes
from matplotlib.lines import Line2D
import matplotlib.cm
if self.wants_objskeleton_graph:
figure.set_subplots((2, 1))
else:
figure.set_subplots((1, 1))
title = (
"Branchpoints of %s and %s\nTrunks are red\nBranches are green\nEndpoints are blue"
% (self.seed_objects_name.value, self.image_name.value)
)
figure.subplot_imshow(0, 0, workspace.display_data.branchpoint_image, title)
if self.wants_objskeleton_graph:
image = workspace.display_data.intensity_image
figure.subplot_imshow_grayscale(
1, 0, image, title="ObjectSkeleton graph", sharexy=figure.subplot(0, 0)
)
axes = figure.subplot(1, 0)
assert isinstance(axes, Axes)
edge_graph = workspace.display_data.edge_graph
vertex_graph = workspace.display_data.vertex_graph
i = vertex_graph["i"]
j = vertex_graph["j"]
kind = vertex_graph["kind"]
brightness = edge_graph["total_intensity"] / edge_graph["length"]
brightness = (brightness - numpy.min(brightness)) / (
numpy.max(brightness) - numpy.min(brightness) + 0.000001
)
cm = matplotlib.cm.get_cmap(get_default_colormap())
cmap = matplotlib.cm.ScalarMappable(cmap=cm)
edge_color = cmap.to_rgba(brightness)
for idx in range(len(edge_graph["v1"])):
v = numpy.array([edge_graph["v1"][idx] - 1, edge_graph["v2"][idx] - 1])
line = Line2D(j[v], i[v], color=edge_color[idx])
axes.add_line(line)
def get_measurement_columns(self, pipeline):
"""Return database column definitions for measurements made here"""
return [
(
self.seed_objects_name.value,
"_".join((C_OBJSKELETON, feature, self.image_name.value)),
COLTYPE_FLOAT
if feature == F_TOTAL_OBJSKELETON_LENGTH
else COLTYPE_INTEGER,
)
for feature in F_ALL
]
def get_categories(self, pipeline, object_name):
"""Get the measurement categories generated by this module
pipeline - pipeline being run
object_name - name of seed object
"""
if object_name == self.seed_objects_name:
return [C_OBJSKELETON]
else:
return []
def get_measurements(self, pipeline, object_name, category):
"""Return the measurement features generated by this module
pipeline - pipeline being run
object_name - object being measured (must be the seed object)
category - category of measurement (must be C_OBJSKELETON)
"""
if category == C_OBJSKELETON and object_name == self.seed_objects_name:
return F_ALL
else:
return []
def get_measurement_images(self, pipeline, object_name, category, measurement):
"""Return the images measured by this module
pipeline - pipeline being run
object_name - object being measured (must be the seed object)
category - category of measurement (must be C_OBJSKELETON)
measurement - one of the object skeleton measurements
"""
if measurement in self.get_measurements(pipeline, object_name, category):
return [self.image_name.value]
else:
return []
def upgrade_settings(self, setting_values, variable_revision_number, module_name):
"""Provide backwards compatibility for old pipelines
setting_values - the strings to be fed to settings
variable_revision_number - the version number at time of saving
module_name - name of original module
"""
if variable_revision_number == 1:
#
# Added hole size questions
#
setting_values = setting_values + ["Yes", "10"]
variable_revision_number = 2
if variable_revision_number == 2:
#
# Added graph stuff
#
setting_values = setting_values + [
"No",
"None",
Directory.static_join_string(DEFAULT_OUTPUT_FOLDER_NAME, "None"),
"None",
"None",
]
variable_revision_number = 3
return setting_values, variable_revision_number
def make_objskeleton_graph(
self, skeleton, skeleton_labels, trunks, branchpoints, endpoints, image
):
"""Make a table that captures the graph relationship of the skeleton
skeleton - binary skeleton image + outline of seed objects
skeleton_labels - labels matrix of skeleton
trunks - binary image with trunk points as 1
branchpoints - binary image with branchpoints as 1
endpoints - binary image with endpoints as 1
image - image for intensity measurement
returns two tables.
Table 1: edge table
The edge table is a numpy record array with the following named
columns in the following order:
v1: index into vertex table of first vertex of edge
v2: index into vertex table of second vertex of edge
length: # of intermediate pixels + 2 (for two vertices)
total_intensity: sum of intensities along the edge
Table 2: vertex table
The vertex table is a numpy record array:
i: I coordinate of the vertex
j: J coordinate of the vertex
label: the vertex's label
kind: kind of vertex = "T" for trunk, "B" for branchpoint or "E" for endpoint.
"""
i, j = numpy.mgrid[0 : skeleton.shape[0], 0 : skeleton.shape[1]]
#
# Give each point of interest a unique number
#
points_of_interest = trunks | branchpoints | endpoints
number_of_points = numpy.sum(points_of_interest)
#
# Make up the vertex table
#
tbe = numpy.zeros(points_of_interest.shape, "|S1")
tbe[trunks] = "T"
tbe[branchpoints] = "B"
tbe[endpoints] = "E"
i_idx = i[points_of_interest]
j_idx = j[points_of_interest]
poe_labels = skeleton_labels[points_of_interest]
tbe = tbe[points_of_interest]
vertex_table = {
self.VF_I: i_idx,
self.VF_J: j_idx,
self.VF_LABELS: poe_labels,
self.VF_KIND: tbe,
}
#
# First, break the skeleton by removing the branchpoints, endpoints
# and trunks
#
broken_skeleton = skeleton & (~points_of_interest)
#
# Label the broken skeleton: this labels each edge differently
#
edge_labels, nlabels = centrosome.cpmorphology.label_skeleton(skeleton)
#
# Reindex after removing the points of interest
#
edge_labels[points_of_interest] = 0
if nlabels > 0:
indexer = numpy.arange(nlabels + 1)
unique_labels = numpy.sort(numpy.unique(edge_labels))
nlabels = len(unique_labels) - 1
indexer[unique_labels] = numpy.arange(len(unique_labels))
edge_labels = indexer[edge_labels]
#
# find magnitudes and lengths for all edges
#
magnitudes = fix(
scipy.ndimage.sum(
image, edge_labels, numpy.arange(1, nlabels + 1, dtype=numpy.int32)
)
)
lengths = fix(
scipy.ndimage.sum(
numpy.ones(edge_labels.shape),
edge_labels,
numpy.arange(1, nlabels + 1, dtype=numpy.int32),
)
).astype(int)
else:
magnitudes = numpy.zeros(0)
lengths = numpy.zeros(0, int)
#
# combine the edge labels and indexes of points of interest with padding
#
edge_mask = edge_labels != 0
all_labels = numpy.zeros(numpy.array(edge_labels.shape) + 2, int)
all_labels[1:-1, 1:-1][edge_mask] = edge_labels[edge_mask] + number_of_points
all_labels[i_idx + 1, j_idx + 1] = numpy.arange(1, number_of_points + 1)
#
# Collect all 8 neighbors for each point of interest
#
p1 = numpy.zeros(0, int)
p2 = numpy.zeros(0, int)
for i_off, j_off in (
(0, 0),
(0, 1),
(0, 2),
(1, 0),
(1, 2),
(2, 0),
(2, 1),
(2, 2),
):
p1 = numpy.hstack((p1, numpy.arange(1, number_of_points + 1)))
p2 = numpy.hstack((p2, all_labels[i_idx + i_off, j_idx + j_off]))
#
# Get rid of zeros which are background
#
p1 = p1[p2 != 0]
p2 = p2[p2 != 0]
#
# Find point_of_interest -> point_of_interest connections.
#
p1_poi = p1[(p2 <= number_of_points) & (p1 < p2)]
p2_poi = p2[(p2 <= number_of_points) & (p1 < p2)]
#
# Make sure matches are labeled the same
#
same_labels = (
skeleton_labels[i_idx[p1_poi - 1], j_idx[p1_poi - 1]]
== skeleton_labels[i_idx[p2_poi - 1], j_idx[p2_poi - 1]]
)
p1_poi = p1_poi[same_labels]
p2_poi = p2_poi[same_labels]
#
# Find point_of_interest -> edge
#
p1_edge = p1[p2 > number_of_points]
edge = p2[p2 > number_of_points]
#
# Now, each value that p2_edge takes forms a group and all
# p1_edge whose p2_edge are connected together by the edge.
# Possibly they touch each other without the edge, but we will
# take the minimum distance connecting each pair to throw out
# the edge.
#
edge, p1_edge, p2_edge = centrosome.cpmorphology.pairwise_permutations(
edge, p1_edge
)
indexer = edge - number_of_points - 1
lengths = lengths[indexer]
magnitudes = magnitudes[indexer]
#
# OK, now we make the edge table. First poi<->poi. Length = 2,
# magnitude = magnitude at each point
#
poi_length = numpy.ones(len(p1_poi)) * 2
poi_magnitude = (
image[i_idx[p1_poi - 1], j_idx[p1_poi - 1]]
+ image[i_idx[p2_poi - 1], j_idx[p2_poi - 1]]
)
#
# Now the edges...
#
poi_edge_length = lengths + 2
poi_edge_magnitude = (
image[i_idx[p1_edge - 1], j_idx[p1_edge - 1]]
+ image[i_idx[p2_edge - 1], j_idx[p2_edge - 1]]
+ magnitudes
)
#
# Put together the columns
#
v1 = numpy.hstack((p1_poi, p1_edge))
v2 = numpy.hstack((p2_poi, p2_edge))
lengths = numpy.hstack((poi_length, poi_edge_length))
magnitudes = numpy.hstack((poi_magnitude, poi_edge_magnitude))
#
# Sort by p1, p2 and length in order to pick the shortest length
#
indexer = numpy.lexsort((lengths, v1, v2))
v1 = v1[indexer]
v2 = v2[indexer]
lengths = lengths[indexer]
magnitudes = magnitudes[indexer]
if len(v1) > 0:
to_keep = numpy.hstack(([True], (v1[1:] != v1[:-1]) | (v2[1:] != v2[:-1])))
v1 = v1[to_keep]
v2 = v2[to_keep]
lengths = lengths[to_keep]
magnitudes = magnitudes[to_keep]
#
# Put it all together into a table
#
edge_table = {
self.EF_V1: v1,
self.EF_V2: v2,
self.EF_LENGTH: lengths,
self.EF_TOTAL_INTENSITY: magnitudes,
}
return edge_table, vertex_table
