def buffer_image(self, image_path, Xi, Yi, Xt, Yt, color=False):
"""buffer_image Load and image and store in buffer
This method loads an image and stores it in the appropriate
position based on the stitching vector coordinates within
a large tile of the output image. It is intended to be
used as a thread to increase the reading component to
assembling the image.
Args:
image_path ([str]): Path to image to load
Xi ([list]): Xmin and Xmax of pixels to load from the image
Yi ([list]): Ymin and Ymax of pixels to load from the image
Xt ([list]): X position within the buffer to store the image
Yt ([list]): Y position within the buffer to store the image
"""
# Load the image
br = BioReader(image_path, max_workers=2)
image = br.read_image(X=Xi, Y=Yi) # only get the first z,c,t layer
# Put the image in the buffer
if color != None:
image_temp = (
255 * (image[..., 0, 0].astype(np.float32) - self.bounds[0]) /
(self.bounds[1] - self.bounds[0]))
image_temp[image_temp > 255] = 255
image_temp[image_temp < 0] = 0
image_temp = image_temp.astype(np.uint8)
self._image[Yt[0]:Yt[1], Xt[0]:Xt[1], ...] = 0
self._image[Yt[0]:Yt[1], Xt[0]:Xt[1], self.color] = image_temp
else:
self._image[Yt[0]:Yt[1], Xt[0]:Xt[1], ...] = image[:, :, :, 0, 0]
def process_image(input_img_path, output_img_path, projection, method):
# Grab a free process
with ProcessManager.process():
# initalize biowriter and bioreader
with BioReader(input_img_path, max_workers=ProcessManager._active_threads) as br, \
BioWriter(output_img_path, metadata=br.metadata, max_workers=ProcessManager._active_threads) as bw:
# output image is 2d
bw.Z = 1
# iterate along the x,y direction
for x in range(0, br.X, tile_size):
x_max = min([br.X, x + tile_size])
for y in range(0, br.Y, tile_size):
y_max = min([br.Y, y + tile_size])
ProcessManager.submit_thread(projection,
br,
bw, (x, x_max), (y, y_max),
method=method)
ProcessManager.join_threads()
def main(
input_dir: Path,
ball_radius: int,
light_background: bool,
output_dir: Path,
) -> None:
""" Main execution function
Args:
input_dir: path to directory containing the input images.
ball_radius: radius of ball to use for the rolling-ball algorithm.
light_background: whether the image has a light or dark background.
output_dir: path to directory where to store the output images.
"""
for in_path in input_dir.iterdir():
in_path = Path(in_path)
out_path = Path(output_dir).joinpath(in_path.name)
# Load the input image
with BioReader(in_path) as reader:
logger.info(f'Working on {in_path.name} with shape {reader.shape}')
# Initialize the output image
with BioWriter(out_path,
metadata=reader.metadata,
max_workers=cpu_count()) as writer:
rolling_ball(
reader=reader,
writer=writer,
ball_radius=ball_radius,
light_background=light_background,
)
return
def _get_resized_image_stack(flist):
""" Load all images in a list and resize to OPTIONS['size']
When files are parsed, the variables are used in an index to provide
a method to reference a specific file name by its dimensions. This
function returns the variable index based on the input filename pattern.
Inputs:ed th
flist - Paths of list of images to load and resize
Outputs:
img_stack - A 3D stack of 2D images
X - width of image
Y - height of image
"""
#Initialize the output
br = BioReader(str(flist[0]))
X = br.num_x()
Y = br.num_y()
C = len(flist)
img_stack = np.zeros((OPTIONS['size'], OPTIONS['size'], C),
dtype=np.float32)
# Load every image as a z-slice
for ind, fname in zip(range(len(flist)), flist):
br = BioReader(str(fname))
I = np.squeeze(br.read_image())
img_stack[:, :, ind] = cv2.resize(
I, (OPTIONS['size'], OPTIONS['size']),
interpolation=cv2.INTER_LINEAR).astype(np.float32)
return img_stack, X, Y
def validate_and_copy(
file: Path,
outDir: Path,
) -> None:
# Enter context manager to verify the file is a tiled tiff
with BioReader(file['file'], backend='python') as br:
shutil.copy2(file['file'], outDir.joinpath(file['file'].name))
def image_to_zarr(inp_image: Path, out_dir: Path) -> None:
with ProcessManager.process():
with BioReader(inp_image) as br:
# Loop through timepoints
for t in range(br.T):
# Loop through channels
for c in range(br.C):
extension = "".join([
suffix for suffix in inp_image.suffixes[-2:]
if len(suffix) < 5
])
out_path = out_dir.joinpath(
inp_image.name.replace(extension, FILE_EXT))
if br.C > 1:
out_path = out_dir.joinpath(
out_path.name.replace(FILE_EXT,
f"_c{c}" + FILE_EXT))
if br.T > 1:
out_path = out_dir.joinpath(
out_path.name.replace(FILE_EXT,
f"_t{t}" + FILE_EXT))
with BioWriter(
out_path,
max_workers=ProcessManager._active_threads,
metadata=br.metadata,
) as bw:
bw.C = 1
bw.T = 1
bw.channel_names = [br.channel_names[c]]
# Loop through z-slices
for z in range(br.Z):
# Loop across the length of the image
for y in range(0, br.Y, TILE_SIZE):
y_max = min([br.Y, y + TILE_SIZE])
bw.max_workers = ProcessManager._active_threads
br.max_workers = ProcessManager._active_threads
# Loop across the depth of the image
for x in range(0, br.X, TILE_SIZE):
x_max = min([br.X, x + TILE_SIZE])
bw[y:y_max, x:x_max, z:z + 1, 0,
0] = br[y:y_max, x:x_max, z:z + 1, c, t]
def main(
input_dir: Path,
file_pattern: str,
output_dir: Path,
):
fp = filepattern.FilePattern(input_dir, file_pattern)
files = [Path(file[0]['file']).resolve() for file in fp]
files = list(filter(
lambda file_path: file_path.name.endswith('.ome.tif') or file_path.name.endswith('.ome.zarr'),
files
))
executor = (ThreadPoolExecutor if utils.USE_GPU else ProcessPoolExecutor)(utils.NUM_THREADS)
processes: List[Future[bool]] = list()
for in_file in files:
with BioReader(in_file) as reader:
x_shape, y_shape, z_shape = reader.X, reader.Y, reader.Z
metadata = reader.metadata
ndims = 2 if z_shape == 1 else 3
out_file = output_dir.joinpath(utils.replace_extension(in_file, extension='_flow.ome.zarr'))
init_zarr_file(out_file, ndims, metadata)
tile_count = 0
for z in range(0, z_shape, utils.TILE_SIZE):
z = None if ndims == 2 else z
for y in range(0, y_shape, utils.TILE_SIZE):
for x in range(0, x_shape, utils.TILE_SIZE):
coordinates = x, y, z
device = (tile_count % utils.NUM_THREADS) if utils.USE_GPU else None
tile_count += 1
# flow_thread(in_file, out_file, coordinates, device)
processes.append(executor.submit(
flow_thread,
in_file,
out_file,
coordinates,
device,
))
done, not_done = wait(processes, 0)
while len(not_done) > 0:
logger.info(f'Percent complete: {100 * len(done) / len(processes):6.3f}%')
for r in done:
r.result()
done, not_done = wait(processes, 5)
executor.shutdown()
return
def _merge_layers(input_dir, input_files, output_dir, output_file):
zs = [z for z in input_files.keys()] # sorted list of filenames by z-value
zs.sort()
# Initialize the output file
br = BioReader(
str(Path(input_dir).joinpath(input_files[zs[0]][0]).absolute()))
bw = BioWriter(str(Path(output_dir).joinpath(output_file).absolute()),
metadata=br.read_metadata())
bw.num_z(Z=len(zs))
del br
# Load each image and save to the volume file
for z, i in zip(zs, range(len(zs))):
br = BioReader(
str(Path(input_dir).joinpath(input_files[z][0]).absolute()))
bw.write_image(br.read_image(), Z=[i, i + 1])
del br
# Close the output image and delete
bw.close_image()
del bw
# Set up the number of threads for each task
read_workers = max([cpu_count()//3,1])
write_workers = max([cpu_count()-1,2])
loop_workers = max([3*cpu_count()//4,2])
# extract filenames from registration_string and similar_transformation_string
registration_set=registration_string.split()
similar_transformation_set=similar_transformation_string.split()
filename_len=len(template)
# seperate the filename of the moving image from the complete path
moving_image_name=registration_set[1][-1*filename_len:]
# read and downscale reference image
br_ref = BioReader(registration_set[0],max_workers=write_workers)
scale_factor=get_scale_factor(br_ref.num_y(),br_ref.num_x())
logger.info('Scale factor: {}'.format(scale_factor))
# intialize the scale factor and scale matrix(to be used to upscale the transformation matrices)
if method == 'Projective':
scale_matrix = np.array([[1,1,scale_factor],[1,1,scale_factor],[1/scale_factor,1/scale_factor,1]])
else:
scale_matrix = np.array([[1/scale_factor,1/scale_factor,1],[1/scale_factor,1/scale_factor,1]])
logger.info('Reading and downscaling reference image: {}'.format(Path(registration_set[0]).name))
reference_image_downscaled,max_val,min_val = get_scaled_down_images(br_ref,scale_factor,get_max=True)
br_ref.max_workers = read_workers
# read moving image
logger.info('Reading and downscaling moving image: {}'.format(Path(registration_set[1]).name))
def main(
_opName: str,
_in1: Path,
_sigma: str,
_calibration: str,
_out: Path,
) -> None:
"""Initialize ImageJ"""
# Bioformats throws a debug message, disable the loci debugger to mute it
def disable_loci_logs():
DebugTools = scyjava.jimport("loci.common.DebugTools")
DebugTools.setRootLevel("WARN")
scyjava.when_jvm_starts(disable_loci_logs)
# This is the version of ImageJ pre-downloaded into the docker container
logger.info("Starting ImageJ...")
ij = imagej.init("sc.fiji:fiji:2.1.1+net.imagej:imagej-legacy:0.37.4",
headless=True)
# ij_converter.ij = ij
logger.info("Loaded ImageJ version: {}".format(ij.getVersion()))
""" Validate and organize the inputs """
args = []
argument_types = []
arg_len = 0
# Validate opName
opName_values = [
"DefaultTubeness",
]
assert _opName in opName_values, "opName must be one of {}".format(
opName_values)
# Validate in1
in1_types = {
"DefaultTubeness": "RandomAccessibleInterval",
}
# Check that all inputs are specified
if _in1 is None and _opName in list(in1_types.keys()):
raise ValueError("{} must be defined to run {}.".format(
"in1", _opName))
elif _in1 != None:
in1_type = in1_types[_opName]
# switch to images folder if present
if _in1.joinpath("images").is_dir():
_in1 = _in1.joinpath("images").absolute()
args.append([f for f in _in1.iterdir() if f.is_file()])
arg_len = len(args[-1])
else:
argument_types.append(None)
args.append([None])
# Validate sigma
sigma_types = {
"DefaultTubeness": "double",
}
# Check that all inputs are specified
if _sigma is None and _opName in list(sigma_types.keys()):
raise ValueError("{} must be defined to run {}.".format(
"sigma", _opName))
else:
sigma = None
# Validate calibration
calibration_types = {
"DefaultTubeness": "double[]",
}
# Check that all inputs are specified
if _calibration is None and _opName in list(calibration_types.keys()):
raise ValueError("{} must be defined to run {}.".format(
"calibration", _opName))
else:
calibration = None
for i in range(len(args)):
if len(args[i]) == 1:
args[i] = args[i] * arg_len
""" Set up the output """
out_types = {
"DefaultTubeness": "IterableInterval",
}
""" Run the plugin """
try:
for ind, (in1_path, ) in enumerate(zip(*args)):
if in1_path != None:
# Load the first plane of image in in1 collection
logger.info("Processing image: {}".format(in1_path))
in1_br = BioReader(in1_path)
# Convert to appropriate numpy array
in1 = ij_converter.to_java(ij,
np.squeeze(in1_br[:, :, 0:1, 0, 0]),
in1_type)
metadata = in1_br.metadata
fname = in1_path.name
dtype = ij.py.dtype(in1)
if _sigma is not None:
sigma = ij_converter.to_java(ij, _sigma, sigma_types[_opName],
dtype)
if _calibration is not None:
calibration = ij_converter.to_java(ij, _calibration,
calibration_types[_opName],
dtype)
logger.info("Running op...")
if _opName == "DefaultTubeness":
out = ij.op().filter().tubeness(in1, sigma, calibration)
logger.info("Completed op!")
if in1_path != None:
in1_br.close()
# Saving output file to out
logger.info("Saving...")
out_array = ij_converter.from_java(ij, out, out_types[_opName])
bw = BioWriter(_out.joinpath(fname), metadata=metadata)
bw.Z = 1
bw.dtype = out_array.dtype
bw[:] = out_array.astype(bw.dtype)
bw.close()
except:
logger.error("There was an error, shutting down jvm before raising...")
raise
finally:
# Exit the program
logger.info("Shutting down jvm...")
del ij
jpype.shutdownJVM()
logger.info("Complete!")
def _parse_stitch(self, stitchPath, imagePath):
""" Load and parse image stitching vectors
This function creates a list of file dictionaries that include the filename and
pixel position and dimensions within a stitched image. It also determines the
size of the final stitched image and the suggested name of the output image based
on differences in file names in the stitching vector.
This method originally appeared in the image assembler plugin:
https://github.com/Nicholas-Schaub/polus-plugins/blob/imageassembler/polus-image-assembler-plugin/src/main.py
Inputs:
stitchPath - A path to stitching vectors
imagePath - A path to tiled tiff images
timepointName - Use the vector timeslice as the image name
Outputs:
out_dict - Dictionary with keys (width, height, name, filePos)
"""
# Initialize the output
out_dict = {'width': int(0), 'height': int(0), 'filePos': []}
# Set the regular expression used to parse each line of the stitching vector
line_regex = r"file: (.*); corr: (.*); position: \((.*), (.*)\); grid: \((.*), (.*)\);"
# Get a list of all images in imagePath
images = [p.name for p in Path(imagePath).iterdir()]
# Open each stitching vector
fpath = str(Path(stitchPath).absolute())
name_pos = {}
with open(fpath, 'r') as fr:
# Read the first line to get the filename for comparison to all other filenames
line = fr.readline()
stitch_groups = re.match(line_regex, line)
stitch_groups = {
key: val
for key, val in zip(STITCH_VARS, stitch_groups.groups())
}
name = stitch_groups['file']
name_ind = [i for i in range(len(name))]
fr.seek(0) # reset to the first line
# Read each line in the stitching vector
for line in fr:
# Read and parse values from the current line
stitch_groups = re.match(line_regex, line)
stitch_groups = {
key: get_number(val)
for key, val in zip(STITCH_VARS, stitch_groups.groups())
}
# If an image in the vector doesn't match an image in the collection, then skip it
if stitch_groups['file'] not in images:
continue
# Get the image size
stitch_groups['width'], stitch_groups[
'height'] = BioReader.image_size(
str(
Path(imagePath).joinpath(
stitch_groups['file']).absolute()))
if out_dict['width'] < stitch_groups['width'] + stitch_groups[
'posX']:
out_dict['width'] = stitch_groups['width'] + stitch_groups[
'posX']
if out_dict['height'] < stitch_groups[
'height'] + stitch_groups['posY']:
out_dict['height'] = stitch_groups[
'height'] + stitch_groups['posY']
# Set the stitching vector values in the file dictionary
out_dict['filePos'].append(stitch_groups)
return out_dict
files = list({% for inp,val in cookiecutter._inputs.items() if val.type=='collection' -%}
{% if loop.first %}{{ inp }}{% endif %}
{%- endfor -%}.iterdir())
for file in files:
{%- endif %}
{#- Use bfio if requested #}
{%- if cookiecutter.use_bfio == "True" %}
{%- filter indent(level2,True) %}
logger.info(f'Processing image: {file["file"]}')
# Load the input image
logger.debug(f'Initializing BioReader for {file["file"]}')
with BioReader(file['file']) as br:
input_extension = ''.join([s for s in file['file'].suffixes[-2:] if len(s) < 5])
out_name = file['file'].name.replace(input_extension,POLUS_EXT)
out_path = {{ cookiecutter._outputs.keys()|first }}.joinpath(out_name)
# Initialize the output image
logger.debug(f'Initializing BioReader for {out_path}')
with BioWriter(out_path,metadata=br.metadata) as bw:
# This is where the magic happens, replace this part with your method
bw[:] = awesome_function(br[:])
{%- endfilter %}
{%- endif %}
if __name__=="__main__":
datefmt='%d-%b-%y %H:%M:%S')
logger = logging.getLogger("do_flat")
logger.setLevel(logging.INFO)
# Set up the FilePattern object
images = FilePattern(inpDir, filepattern)
''' Start the javabridge '''
logger.info("Starting the javabridge...")
log_config = Path(__file__).parent.joinpath("log4j.properties")
jutil.start_vm(args=[
"-Dlog4j.configuration=file:{}".format(str(log_config.absolute()))
],
class_path=bioformats.JARS)
''' Load the flatfielding data '''
logger.info("Loading the flatfield data...")
flat_br = BioReader(brightfield)
flat_image = np.squeeze(flat_br.read_image())
del flat_br
# Normalize the brightfield image if it isn't done already
flat_image = flat_image.astype(np.float32)
flat_image = np.divide(flat_image, np.mean(flat_image))
# Load the darkfield and photobleach offsets if they are specified
if darkfield != None:
dark_br = BioReader(darkfield)
dark_image = np.squeeze(dark_br.read_image())
del dark_br
else:
dark_image = np.zeros(flat_image.shape, dtype=np.float32)
if photobleach != None:
def flow_thread(
file_name: Path,
zarr_path: Path,
coordinates: Tuple[int, int, Optional[int]],
device: Optional[int],
) -> bool:
x, y, z = coordinates
ndims = 2 if z is None else 3
z = 0 if z is None else z
# Load the data
with BioReader(file_name) as reader:
x_shape, y_shape, z_shape = reader.X, reader.Y, reader.Z
x_min = max(0, x - utils.TILE_OVERLAP)
x_max = min(x_shape, x + utils.TILE_SIZE + utils.TILE_OVERLAP)
y_min = max(0, y - utils.TILE_OVERLAP)
y_max = min(y_shape, y + utils.TILE_SIZE + utils.TILE_OVERLAP)
z_min = max(0, z - utils.TILE_OVERLAP)
z_max = min(z_shape, z + utils.TILE_SIZE + utils.TILE_OVERLAP)
masks = numpy.squeeze(reader[y_min:y_max, x_min:x_max, z_min:z_max, 0, 0])
masks = masks if ndims == 2 else numpy.transpose(masks, (2, 0, 1))
masks_shape = masks.shape
# Calculate index and offsets
x_overlap = x - x_min
x_min, x_max = x, min(x_shape, x + utils.TILE_SIZE)
cx_min, cx_max = x_overlap, x_max - x_min + x_overlap
y_overlap = y - y_min
y_min, y_max = y, min(y_shape, y + utils.TILE_SIZE)
cy_min, cy_max = y_overlap, y_max - y_min + y_overlap
z_overlap = z - z_min
z_min, z_max = z, min(z_shape, z + utils.TILE_SIZE)
cz_min, cz_max = z_overlap, z_max - z_min + z_overlap
# Save the mask before transforming
if ndims == 2:
masks_original = masks[numpy.newaxis, numpy.newaxis, numpy.newaxis, :, :]
else:
masks_original = masks[numpy.newaxis, numpy.newaxis, :, :]
masks_original = masks_original[:, :, cz_min:cz_max, cy_min:cy_max, cx_min:cx_max]
# noinspection PyTypeChecker
zarr_root = zarr.open(str(zarr_path))[0]
zarr_root[0:1, 0:1, z_min:z_max, y_min:y_max, x_min:x_max] = numpy.asarray(masks_original != 0, dtype=numpy.float32)
zarr_root[0:1, ndims + 1:ndims + 2, z_min:z_max, y_min:y_max, x_min:x_max] = numpy.asarray(masks_original, dtype=numpy.float32)
if not numpy.any(masks):
logger.debug(f'Tile (x, y, z) = {x, y, z} in file {file_name.name} has no objects. Setting flows to zero...')
flows = numpy.zeros((ndims, *masks.shape), dtype=numpy.float32)
else:
# Normalize
labels, masks = numpy.unique(masks, return_inverse=True)
if len(labels) == 1:
logger.debug(f'Tile (x, y, z) = {x, y, z} in file {file_name.name} has only one object.')
masks += 1
masks = numpy.reshape(masks, newshape=masks_shape)
flows = dynamics.masks_to_flows(masks, device=device)
logger.debug(f'Computed flows on tile (x, y, z) = {x, y, z} in file {file_name.name}')
# Zarr axes ordering should be (t, c, z, y, x). Add missing t, c, and z axes
if ndims == 2:
flows = flows[numpy.newaxis, :, numpy.newaxis, :, :]
else:
flows = flows[numpy.newaxis, :, :, :]
flows = flows[:, :, cz_min:cz_max, cy_min:cy_max, cx_min:cx_max]
zarr_root[0:1, 1:ndims + 1, z_min:z_max, y_min:y_max, x_min:x_max] = flows
return True
def extract_fovs(file_path: Path,
out_path: Path):
""" Extract individual FOVs from a czi file
When CZI files are loaded by BioFormats, it will generally try to mosaic
images together by stage position if the image was captured with the
intention of mosaicing images together. At the time this function was
written, there was no clear way of extracting individual FOVs so this
algorithm was created.
Every field of view in each z-slice, channel, and timepoint contained in a
CZI file is saved as an individual image.
Args:
file_path (Path): Path to CZI file
out_path (Path): Path to output directory
"""
with ProcessManager.process(file_path.name):
logger.info('Starting extraction from ' + str(file_path) + '...')
if Path(file_path).suffix != '.czi':
TypeError("Path must be to a czi file.")
base_name = Path(file_path.name).stem
# Load files without mosaicing
czi = czifile.CziFile(file_path,detectmosaic=False)
subblocks = [s for s in czi.filtered_subblock_directory if s.mosaic_index is not None]
ind = {'X': [],
'Y': [],
'Z': [],
'C': [],
'T': [],
'Row': [],
'Col': []}
# Get the indices of each FOV
for s in subblocks:
scene = [dim.start for dim in s.dimension_entries if dim.dimension=='S']
if scene is not None and scene[0] != 0:
continue
for dim in s.dimension_entries:
if dim.dimension=='X':
ind['X'].append(dim.start)
elif dim.dimension=='Y':
ind['Y'].append(dim.start)
elif dim.dimension=='Z':
ind['Z'].append(dim.start)
elif dim.dimension=='C':
ind['C'].append(dim.start)
elif dim.dimension=='T':
ind['T'].append(dim.start)
row_conv = {y:row for (y,row) in zip(np.unique(np.sort(ind['Y'])),range(0,len(np.unique(ind['Y']))))}
col_conv = {x:col for (x,col) in zip(np.unique(np.sort(ind['X'])),range(0,len(np.unique(ind['X']))))}
ind['Row'] = [row_conv[y] for y in ind['Y']]
ind['Col'] = [col_conv[x] for x in ind['X']]
with BioReader(file_path) as br:
metadata = br.metadata
chan_names = br.cnames
for s,i in zip(subblocks,range(0,len(subblocks))):
Z = None if len(ind['Z'])==0 else ind['Z'][i]
C = None if len(ind['C'])==0 else ind['C'][i]
T = None if len(ind['T'])==0 else ind['T'][i]
out_file_path = out_path.joinpath(_get_image_name(base_name,
row=ind['Row'][i],
col=ind['Col'][i],
Z=Z,
C=C,
T=T))
dims = [_get_image_dim(s,'Y'),
_get_image_dim(s,'X'),
_get_image_dim(s,'Z'),
_get_image_dim(s,'C'),
_get_image_dim(s,'T')]
data = s.data_segment().data().reshape(dims)
write_thread(out_file_path,
data,
metadata,
chan_names[C])
"-Dlog4j.configuration=file:{}".format(str(log_config.absolute()))
],
class_path=bioformats.JARS)
# Make the output directory
image = Path(input_dir).joinpath(image)
if pyramid_type == "Neuroglancer":
out_dir = Path(output_dir).joinpath(image.name)
elif pyramid_type == "DeepZoom":
out_dir = Path(output_dir).joinpath('{}_files'.format(image_num))
out_dir.mkdir()
out_dir = str(out_dir.absolute())
# Create the BioReader object
logger.info('Getting the BioReader...')
bf = BioReader(str(image.absolute()))
# Create the output path and info file
if pyramid_type == "Neuroglancer":
file_info = utils.neuroglancer_info_file(bf, out_dir)
elif pyramid_type == "DeepZoom":
file_info = utils.dzi_file(bf, out_dir, image_num)
else:
ValueError("pyramid_type must be Neuroglancer or DeepZoom")
logger.info("data_type: {}".format(file_info['data_type']))
logger.info("num_channels: {}".format(file_info['num_channels']))
logger.info("number of scales: {}".format(len(file_info['scales'])))
logger.info("type: {}".format(file_info['type']))
# Create the classes needed to generate a precomputed slice
logger.info("Creating encoder and file writer...")
for files in fp.iterate(group_by='c'):
# Get the filenames in the channel order
paths = []
for c in channelOrder:
for file in files:
if file['c'] == c:
paths.append(file)
break
# make sure that files were found in the current loop
if len(paths) == 0:
continue
# Initialize the output file
br = BioReader(paths[0]['file'])
file_name = filepattern.output_name(
filePattern, paths,
{c: paths[0][c]
for c in fp.variables if c != 'c'})
logger.info('Writing: {}'.format(file_name))
bw = BioWriter(str(Path(outDir).joinpath(file_name)),
metadata=br.read_metadata())
del br
# Modify the metadata to make sure channels are written correctly
bw.num_c(len(paths))
bw._metadata.image().Pixels.channel_count = bw.num_c()
# Process the data in tiles
threads = []
def flow_thread(input_path: Path, zfile: Path, use_gpu: bool,
dev: torch.device, x: int, y: int, z: int) -> bool:
""" Converts labels to flows
This function converts labels in each tile to vector field.
Args:
input_path(path): Path of input image collection
zfile(path): Path where output zarr file will be saved
x(int): Start index of the tile in x dimension of image
y(int): Start index of the tile in y dimension of image
z(int): Z slice of the image
"""
logging.basicConfig(
format='%(asctime)s - %(name)-8s - %(levelname)-8s - %(message)s',
datefmt='%d-%b-%y %H:%M:%S')
logger = logging.getLogger("flow")
logger.setLevel(logging.INFO)
root = zarr.open(str(zfile))[0]
with BioReader(input_path) as br:
x_min = max([0, x - TILE_OVERLAP])
x_max = min([br.X, x + TILE_SIZE + TILE_OVERLAP])
y_min = max([0, y - TILE_OVERLAP])
y_max = min([br.Y, y + TILE_SIZE + TILE_OVERLAP])
# Normalize
I = br[y_min:y_max, x_min:x_max, z:z + 1, 0, 0].squeeze()
_, image = np.unique(I, return_inverse=True)
image = image.reshape(y_max - y_min, x_max - x_min)
flow = dynamics.masks_to_flows(image, use_gpu, dev)[0]
logger.debug('Computed flows on slice %d tile(y,x) %d:%d %d:%d ', z, y,
y_max, x, x_max)
flow_final = flow[:, :, :, np.newaxis,
np.newaxis].transpose(1, 2, 3, 0, 4)
x_overlap = x - x_min
x_min = x
x_max = min([br.X, x + TILE_SIZE])
y_overlap = y - y_min
y_min = y
y_max = min([br.Y, y + TILE_SIZE])
root[0:1, 0:1, z:z + 1, y_min:y_max, x_min:x_max, ] = (
I[y_overlap:y_max - y_min + y_overlap, x_overlap:x_max - x_min +
x_overlap, np.newaxis, np.newaxis, np.newaxis] > 0).transpose(
4, 3, 2, 0, 1)
root[0:1, 1:3, z:z + 1, y_min:y_max,
x_min:x_max] = flow_final[y_overlap:y_max - y_min + y_overlap,
x_overlap:x_max - x_min + x_overlap,
...].transpose(4, 3, 2, 0, 1)
root[0:1, 3:4, z:z + 1, y_min:y_max,
x_min:x_max, ] = I[y_overlap:y_max - y_min + y_overlap,
x_overlap:x_max - x_min + x_overlap,
np.newaxis, np.newaxis, np.newaxis].astype(
np.float32).transpose(4, 3, 2, 0, 1)
return True
def main(inpDir: Path, outDir: Path, filePattern: str = None) -> None:
""" Turn labels into flow fields.
Args:
inpDir: Path to the input directory
outDir: Path to the output directory
"""
# Use a gpu if it's available
use_gpu = torch.cuda.is_available()
if use_gpu:
dev = torch.device("cuda")
else:
dev = torch.device("cpu")
logger.info(f'Running on: {dev}')
# Determine the number of threads to run on
num_threads = max([cpu_count() // 2, 1])
logger.info(f'Number of threads: {num_threads}')
# Get all file names in inpDir image collection based on input pattern
if filePattern:
fp = filepattern.FilePattern(inpDir, filePattern)
inpDir_files = [file[0]['file'].name for file in fp()]
logger.info('Processing %d labels based on filepattern ' %
(len(inpDir_files)))
else:
inpDir_files = [f.name for f in Path(inpDir).iterdir() if f.is_file()]
# Loop through files in inpDir image collection and process
processes = []
if use_gpu:
executor = ThreadPoolExecutor(num_threads)
else:
executor = ProcessPoolExecutor(num_threads)
for f in inpDir_files:
br = BioReader(Path(inpDir).joinpath(f).absolute())
out_file = Path(outDir).joinpath(
f.replace('.ome', '_flow.ome').replace('.tif',
'.zarr')).absolute()
bw = BioWriter(out_file, metadata=br.metadata)
bw.C = 4
bw.dtype = np.float32
bw.channel_names = ['cell_probability', 'x', 'y', 'labels']
bw._backend._init_writer()
for z in range(br.Z):
for x in range(0, br.X, TILE_SIZE):
for y in range(0, br.Y, TILE_SIZE):
processes.append(
executor.submit(flow_thread,
Path(inpDir).joinpath(f).absolute(),
out_file, use_gpu, dev, x, y, z))
bw.close()
br.close()
done, not_done = wait(processes, 0)
logger.info(f'Percent complete: {100 * len(done) / len(processes):6.3f}%')
while len(not_done) > 0:
for r in done:
r.result()
done, not_done = wait(processes, 5)
logger.info(
f'Percent complete: {100 * len(done) / len(processes):6.3f}%')
executor.shutdown()
log_config = Path(__file__).parent.joinpath("log4j.properties")
jutil.start_vm(args=["-Dlog4j.configuration=file:{}".format(str(log_config.absolute()))],class_path=bioformats.JARS)
{% endif -%}
{% for inp,val in cookiecutter._inputs|dictsort -%}
{% if val.type=="collection" -%}
# Get all file names in {{ inp }} image collection
{{ inp }}_files = [f.name for f in Path({{ inp }}).iterdir() if f.is_file() and "".join(f.suffixes)=='.ome.tif']
{% endif %}
{% endfor -%}
{% for inp,val in cookiecutter._inputs|dictsort -%}
{% for out,n in cookiecutter._outputs|dictsort -%}
{% if val.type=="collection" and cookiecutter.use_bfio -%}
# Loop through files in {{ inp }} image collection and process
for i,f in enumerate({{ inp }}_files):
# Load an image
br = BioReader(Path({{ inp }}).joinpath(f))
image = np.squeeze(br.read_image())
# initialize the output
out_image = np.zeros(image.shape,dtype=br._pix['type'])
""" Do some math and science - you should replace this """
logger.info('Processing image ({}/{}): {}'.format(i,len({{ inp }}_files),f))
out_image = awesome_math_and_science_function(image)
# Write the output
bw = BioWriter(Path({{ out }}).joinpath(f),metadata=br.read_metadata())
bw.write_image(np.reshape(out_image,(br.num_y(),br.num_x(),br.num_z(),1,1)))
{%- endif %}{% endfor %}{% endfor %}
finally:
def make_tile(self, x_min, x_max, y_min, y_max, color=None):
"""make_tile Create a supertile
This method identifies images that have stitching vector positions
within the bounds of the supertile defined by the x and y input
arguments. It then spawns threads to load images and store in the
supertile buffer. Finally it returns the assembled supertile to
allow the main thread to generate the write thread.
Args:
x_min ([int]): Minimum x bound of the tile
x_max ([int]): Maximum x bound of the tile
y_min ([int]): Minimum y bound of the tile
y_max ([int]): Maximum y bound of the tile
stitchPath ([str]): Path to the stitching vector
Returns:
[type]: [description]
"""
self._X_offset = x_min
self._Y_offset = y_min
# Get the data type
br = BioReader(
str(
Path(self._file_path).joinpath(
self._file_dict['filePos'][0]['file'])))
dtype = br._pix['type']
# initialize the image
if color != None:
self._image = np.full((y_max - y_min, x_max - x_min, 4),
color,
dtype=dtype)
else:
self._image = np.zeros((y_max - y_min, x_max - x_min, 1),
dtype=dtype)
# get images in bounds of current super tile
with ThreadPoolExecutor(max([self._max_workers, 2])) as executor:
for f in self._file_dict['filePos']:
if (f['posX'] >= x_min and f['posX'] <= x_max) or (
f['posX'] + f['width'] >= x_min
and f['posX'] + f['width'] <= x_max):
if (f['posY'] >= y_min and f['posY'] <= y_max) or (
f['posY'] + f['height'] >= y_min
and f['posY'] + f['height'] <= y_max):
# get bounds of image within the tile
Xt = [max(0, f['posX'] - x_min)]
Xt.append(
min(x_max - x_min, f['posX'] + f['width'] - x_min))
Yt = [max(0, f['posY'] - y_min)]
Yt.append(
min(y_max - y_min,
f['posY'] + f['height'] - y_min))
# get bounds of image within the image
Xi = [max(0, x_min - f['posX'])]
Xi.append(min(f['width'], x_max - f['posX']))
Yi = [max(0, y_min - f['posY'])]
Yi.append(min(f['height'], y_max - f['posY']))
# self.buffer_image(str(Path(self._file_path).joinpath(f['file'])),Xi,Yi,Xt,Yt,color)
executor.submit(
self.buffer_image,
str(Path(self._file_path).joinpath(f['file'])), Xi,
Yi, Xt, Yt, color)
