def normalize(C):
# briefly transform to msi to be able to apply standard normalizer
msi = Msi()
msi.set_image(C.copy())
standard_normalizer.normalize(msi)
# back to array format used here:
return msi.get_image()
def run(self):
tiff_ring_reader = TiffRingReader()
nr_filters = len(sc.other["RECORDED_WAVELENGTHS"])
# analyze all the first image files
image_files = get_image_files_from_folder(self.flatfield_folder)
image_files = filter(lambda image_name: "F0" in image_name,
image_files)
# helper function to take maximum of two images
def maximum_of_two_images(image_1, image_name_2):
image_2 = tiff_ring_reader.read(
os.path.join(self.flatfield_folder, image_name_2),
nr_filters)[0].get_image()
return np.maximum(image_1, image_2)
# now reduce to maximum of all the single images
flat_maximum = reduce(lambda x, y: maximum_of_two_images(x, y),
image_files, 0)
msi = Msi(image=flat_maximum)
msi.set_wavelengths(sc.other["RECORDED_WAVELENGTHS"])
# write flatfield as nrrd
writer = NrrdWriter(msi)
writer.write(self.output().path)
def test_add_dummy_wavelengths_automatically(self):
msi_no_wavelengths_set = Msi()
msi_no_wavelengths_set.set_image(self.msi.get_image())
nr_wavelengths = msi_no_wavelengths_set.get_image().shape[-1]
np.testing.assert_equal(msi_no_wavelengths_set.get_wavelengths(),
np.arange(nr_wavelengths),
"correct dummy wavelength values set")
def getFakeMsi():
# build a fake multispectral image with 5 dimensions.
image = np.concatenate((np.ones((5, 5, 1)), np.ones(
(5, 5, 1)) * 2, np.ones((5, 5, 1)) * 3, np.ones(
(5, 5, 1)) * 4, np.ones((5, 5, 1)) * 5),
axis=-1)
msi = Msi(image)
msi.set_wavelengths(np.array([5, 4, 3, 2, 1]))
return msi
def getFakeMsi():
# build a fake multispectral image with 5 dimensions.
image = np.concatenate((np.ones((5, 5, 1)),
np.ones((5, 5, 1)) * 2,
np.ones((5, 5, 1)) * 3,
np.ones((5, 5, 1)) * 4,
np.ones((5, 5, 1)) * 5),
axis=-1)
msi = Msi(image)
msi.set_wavelengths(np.array([5, 4, 3, 2, 1]))
return msi
def read(self, file_to_read):
# our spectrometer like to follow german standards in files, we need
# to switch to english ones
transformed=""
replacements = {',': '.', '\r\n': ''}
with open(file_to_read) as infile:
for line in infile:
for src, target in replacements.iteritems():
line = line.replace(src, target)
transformed = "\n".join([transformed, line])
for num, line in enumerate(transformed.splitlines(), 1):
if ">>>>>Begin" in line:
break
for num_end, line in enumerate(transformed.splitlines(), 1):
if ">>>>>End" in line:
num_end -= 1
break
string_only_spectrum = "\n".join(transformed.splitlines()[num:num_end])
data_vector = np.fromstring(string_only_spectrum,
sep="\t").reshape(-1, 2)
msi = Msi(data_vector[:, 1],
{'wavelengths': data_vector[:, 0] * 10 ** -9})
return msi
def read(self,
file_to_read,
resize_factor=None,
sorting_function=sort_by_filter):
""" read the msi from tiffs.
The fileToRead is a string prefix, all files starting with this
prefix will be summarized to one msi. they will be sorted as specified
in the sorting_function
Args:
sorting_function: the function which defines the sorting of the
strings that match the prefix. Pass none if normal
lexicographical sorting is wished
file_to_read: the prefix of the tiff file which shall be read
"""
if resize_factor is None:
resize_factor = TiffReader.RESIZE_FACTOR
path, file_prefix = os.path.split(file_to_read)
files = os.listdir(path)
files_to_read = [
os.path.join(path, f) for f in files if file_prefix[2:] in f
]
files_to_read.sort(cmp=sorting_function)
image_array = [
self.to_image(f, resize_factor=resize_factor)
for f in files_to_read
]
image = reduce(lambda x, y: np.dstack((x, y)), image_array)
msi = Msi(image)
return msi
def read(self, fileToRead):
""" read the nrrd image.
TODO: properties are not correctly extracted from nrrd."""
image = None
try:
reader = sitk.ImageFileReader()
reader.SetFileName(fileToRead)
image = reader.Execute()
image = sitk.GetArrayFromImage(image)
except RuntimeError as re:
# image could not be read
logging.warning("image " + fileToRead +
" could not be loaded: " +
str(re))
# rethrow exception after logging
raise
# if image is too low dimensional singleton dimensions
# are added when saving. Done because sitk can only handle dimensions
# 2,3,4. This removes these singleton dimensions again.
squeezed_image = np.squeeze(image)
msi = Msi(squeezed_image)
return msi
def read(self, fileToRead, n, resize_factor=None, segmentation=None):
""" read the msi from tiffs.
The fileToRead is the first file to read,
then n files will be read to one msi from a
sorted file list
segmentation: tiff filename of the segmentation. If none, it will be
tried to get a segmentation from npy files with filenames like the
tiff files + _seg.tiff. If this fails, no segmentation will be
assumed
"""
if resize_factor is None:
resize_factor = TiffRingReader.RESIZE_FACTOR
path, file_name = os.path.split(fileToRead)
files = os.listdir(path)
files_in_folder = [
os.path.join(path, f) for f in files
if os.path.isfile(os.path.join(path, f)) and f.endswith('.tiff')
]
files_in_folder.sort()
position = files_in_folder.index(fileToRead)
# take n images from found position
image_array = [
self.to_image(f, resize_factor)
for f in files_in_folder[position:position + n]
]
image = reduce(lambda x, y: np.dstack((x, y)), image_array)
# in case of 1 dimensional image: add a fake last dimension, since
# we always assume the last dimension to be the wavelength domain.
# TODO SW: Test this and implement for other readers
if n is 1:
image = np.expand_dims(image, -1)
msi = Msi(image)
# we pass an explicic image as segmentation
if segmentation is not None:
segmentation = self.to_image(segmentation, resize_factor)
else: # otherwise: search for numpy segmentations
try:
segmentation_array = [
to_segmentation(f)
for f in files_in_folder[position:position + n]
]
if do_resize(resize_factor):
segmentation = reduce(lambda x, y: x & y,
segmentation_array)
segmentation = scipy.misc.imresize(segmentation,
resize_factor,
interp="bilinear")
except:
logging.info("didn't find segmentation for all images")
return msi, segmentation
def test_add_dummy_wavelengths_automatically(self):
msi_no_wavelengths_set = Msi()
msi_no_wavelengths_set.set_image(self.msi.get_image())
nr_wavelengths = msi_no_wavelengths_set.get_image().shape[-1]
np.testing.assert_equal(msi_no_wavelengths_set.get_wavelengths(),
np.arange(nr_wavelengths),
"correct dummy wavelength values set")
def run(self):
tiff_ring_reader = TiffRingReader()
nr_filters = len(sc.other["RECORDED_WAVELENGTHS"])
# analyze all the first image files
image_files = get_image_files_from_folder(self.dark_folder,
suffix="F0.tiff")
# returns the mean dark image vector of all inputted dark image
# overly complicated TODO SW: make this simple code readable.
dark_means = map(
lambda image_name: msimani.calculate_mean_spectrum(
tiff_ring_reader.read(
os.path.join(self.dark_folder, image_name), nr_filters)[0]
), image_files)
dark_means_sum = reduce(lambda x, y: x + y.get_image(), dark_means, 0)
final_dark_mean = dark_means_sum / len(dark_means)
msi = Msi(image=final_dark_mean)
msi.set_wavelengths(sc.other["RECORDED_WAVELENGTHS"])
# write flatfield as nrrd
writer = NrrdWriter(msi)
writer.write(self.output().path)
def run(self):
tiff_ring_reader = TiffRingReader()
nr_filters = len(sc.other["RECORDED_WAVELENGTHS"])
# analyze all the first image files
image_files = get_image_files_from_folder(self.dark_folder,
suffix="F0.tiff")
# returns the mean dark image vector of all inputted dark image
# overly complicated TODO SW: make this simple code readable.
dark_means = map(lambda image_name:
msimani.calculate_mean_spectrum(
tiff_ring_reader.read(os.path.join(self.dark_folder, image_name),
nr_filters)[0]),
image_files)
dark_means_sum = reduce(lambda x, y: x+y.get_image(), dark_means, 0)
final_dark_mean = dark_means_sum / len(dark_means)
msi = Msi(image=final_dark_mean)
msi.set_wavelengths(sc.other["RECORDED_WAVELENGTHS"])
# write flatfield as nrrd
writer = NrrdWriter(msi)
writer.write(self.output().path)
def run(self):
tiff_ring_reader = TiffRingReader()
nr_filters = len(sc.other["RECORDED_WAVELENGTHS"])
# analyze all the first image files
image_files = get_image_files_from_folder(self.flatfield_folder)
image_files = filter(lambda image_name: "F0" in image_name, image_files)
# helper function to take maximum of two images
def maximum_of_two_images(image_1, image_name_2):
image_2 = tiff_ring_reader.read(os.path.join(self.flatfield_folder,
image_name_2),
nr_filters)[0].get_image()
return np.maximum(image_1, image_2)
# now reduce to maximum of all the single images
flat_maximum = reduce(lambda x, y: maximum_of_two_images(x, y),
image_files, 0)
msi = Msi(image=flat_maximum)
msi.set_wavelengths(sc.other["RECORDED_WAVELENGTHS"])
# write flatfield as nrrd
writer = NrrdWriter(msi)
writer.write(self.output().path)
def read(self, fileToRead):
""" read the msi from pngs.
The fileToRead is a string prefix, all files starting with this
prefix will be summarized to one msi"""
path, file_prefix = os.path.split(fileToRead)
files = os.listdir(path)
files_to_read = [
os.path.join(path, f) for f in files if f.startswith(file_prefix)
]
files_to_read.sort()
image_array = [toImage(f) for f in files_to_read]
image = reduce(lambda x, y: np.dstack((x, y)), image_array)
msi = Msi(image)
return msi
def test_properties_not_shared(self):
msi1 = Msi()
msi2 = Msi()
msi1.add_property({"integration time": np.array([1, 2, 3])})
self.assertTrue('integration time' not in msi2.get_properties())
def test_properties_not_shared(self):
msi1 = Msi()
msi2 = Msi()
msi1.add_property({"integration time": np.array([1, 2, 3])})
self.assertTrue('integration time' not in msi2.get_properties())
