class SegmentationPluginOutside(_SegmentationPlugin):
LABEL = 'Ring around primary region'
NAME = 'outside'
COLOR = '#00FF00'
DOC = ":additional_segmentation_plugins"
REQUIRES = ['primary_segmentation']
PARAMS = [('expansion_size', IntTrait(10, 0, 4000, label='Expansion size')),
('separation_size', IntTrait(5, 0, 4000, label='Separation size')),
]
@stopwatch()
def _run(self, meta_image, container):
image = meta_image.image
if self.params['expansion_size'] > 0 and self.params['expansion_size'] > self.params['separation_size']:
nr_objects = container.img_labels.getMinmax()[1] + 1
img_labels = ccore.seeded_region_expansion(image,
container.img_labels,
ccore.SrgType.KeepContours,
nr_objects,
0,
self.params['expansion_size'],
self.params['separation_size'])
img_labels = ccore.substractImages(img_labels, container.img_labels)
return ccore.ImageMaskContainer(image, img_labels, False, True, True)
else:
raise ValueError("Parameters are not valid. Requirements: 'expansion_size' > 0 and "
"'expansion_size' > 'separation_size'")
class SegmentationPluginPropagate(_SegmentationPlugin):
LABEL = 'Propagate region from primary'
NAME = 'propagate'
COLOR = '#FFFF99'
DOC = ":additional_segmentation_plugins"
REQUIRES = ['primary_segmentation']
PARAMS = [('presegmentation_median_radius', IntTrait(1, 0, 100, label='Median radius')),
('presegmentation_alpha', FloatTrait(1.0, 0, 4000, label='Otsu factor', digits=2)),
('lambda', FloatTrait(0.05, 0, 4000, label='Lambda', digits=2)),
('delta_width', IntTrait(1, 1, 4, label='Delta width')),
]
@stopwatch()
def _run(self, meta_image, container):
image = meta_image.image
img_prefiltered = ccore.disc_median(image, self.params['presegmentation_median_radius'])
t = int(ccore.get_otsu_threshold(img_prefiltered) * self.params['presegmentation_alpha'])
img_bin = ccore.threshold_image(img_prefiltered, t)
img_labels = ccore.segmentation_propagate(img_prefiltered, img_bin,
container.img_labels,
self.params['lambda'],
self.params['delta_width'])
return ccore.ImageMaskContainer(image, img_labels, False, True, True)
class SectionTracking(SectionCore):
SECTION_NAME = SECTION_NAME_TRACKING
OPTIONS = [
('tracking', [
('region', SelectionTrait2(None, [], label='Region name')),
('tracking_maxobjectdistance',
IntTrait(0, 0, 4000, label='Max object x-y distance')),
('tracking_maxtrackinggap',
IntTrait(0, 0, 4000, label='Max time-point gap')),
('tracking_maxsplitobjects',
IntTrait(0, 0, 4000, label='Max split events')),
('tracking_exporttrackfeatures',
BooleanTrait(True, label='Export tracks')),
('tracking_compressiontrackfeatures',
SelectionTrait(COMPRESSION_FORMATS[0],
COMPRESSION_FORMATS,
label='Compression')),
]),
('visualization', [
('tracking_visualization',
BooleanTrait(False, label='Visualization')),
('tracking_visualize_track_length',
IntTrait(5, -1, 10000, label='Max. time-points')),
('tracking_centroid_radius',
IntTrait(3, -1, 50, label='Centroid radius')),
]),
]
class SectionEventSelection(SectionCore):
SECTION_NAME = SECTION_NAME_EVENT_SELECTION
OPTIONS = [
('event_selection',
[('eventchannel',
SelectionTrait2(None, [], label='Channel')),
('event_selection',
BooleanTrait(True, label='Event Selection')),
('backwardrange',
FloatTrait(0, -1, 4000, label='Duration [pre]')),
('forwardrange',
FloatTrait(0, -1, 4000, label='Duration [post]')),
('duration_unit',
SelectionTrait2(TimeConverter.FRAMES,
TimeConverter.units,
label='Duration unit')),
('backwardrange_min', BooleanTrait(False, label='Min.')),
('forwardrange_min', BooleanTrait(False, label='Min.')),
('maxindegree', IntTrait(1, 0, 4000, label='Max in-degree')),
('maxoutdegree', IntTrait(2, 0, 4000, label='Max out-degree')),
]),
('supervised_event_selection',
[('supervised_event_selection',
BooleanTrait(True, label='Supervised',
widget_info=BooleanTrait.RADIOBUTTON)),
('labeltransitions',
StringTrait('', 200, label='Class transition motif(s)')),
('backwardlabels',
StringTrait('', 200, label='Class filter [pre]',
mask='(\d+,)*\d+')),
('forwardlabels',
StringTrait('', 200, label='Class filter [post]',
mask='(\d+,)*\d+')),
('backwardcheck',
FloatTrait(2, 0, 4000, label='Filter duration [pre]')),
('forwardcheck',
FloatTrait(2, 0, 4000, label='Filter duration [post]')),
]),
('unsupervised_event_selection',
[('unsupervised_event_selection',
BooleanTrait(False, label='Unsupervised',
widget_info=BooleanTrait.RADIOBUTTON)),
('min_event_duration',
IntTrait(3, 1, 100, label='Min. event duration')),
('num_clusters',
IntTrait(6, 2, 15, label='Number of clusters',)),
('min_cluster_size',
IntTrait(2, 1, 10, label='Min. cluster size',)),
])
]
class SectionTracking(SectionCore):
SECTION_NAME = SECTION_NAME_TRACKING
OPTIONS = [
('tracking', [
('region', SelectionTrait2(None, [], label='Region name')),
('tracking_maxobjectdistance',
IntTrait(0, 0, 4000, label='Max object x-y distance')),
('tracking_maxtrackinggap',
IntTrait(0, 0, 4000, label='Max time-point gap')),
('tracking_maxsplitobjects',
IntTrait(0, 0, 4000, label='Max split events')),
]),
]
class SegmentationPluginExpanded(_SegmentationPlugin):
LABEL = 'Expanded region from primary'
NAME = 'expanded'
COLOR = '#00FFFF'
DOC = ":additional_segmentation_plugins"
REQUIRES = ['primary_segmentation']
PARAMS = [('expansion_size', IntTrait(10, 0, 4000, label='Expansion size')),
]
@stopwatch()
def _run(self, meta_image, container):
image = meta_image.image
if self.params['expansion_size'] > 0:
nr_objects = container.img_labels.getMinmax()[1] + 1
img_labels = ccore.seeded_region_expansion(image,
container.img_labels,
ccore.SrgType.KeepContours,
nr_objects,
0,
self.params['expansion_size'],
0)
else:
img_labels = container.img_labels
return ccore.ImageMaskContainer(image, img_labels, False, True, True)
class SegmentationPluginInside(_SegmentationPlugin):
LABEL = 'Shrinked region from primary'
NAME = 'inside'
COLOR = '#FFFF00'
DOC = ":additional_segmentation_plugins"
REQUIRES = ['primary_segmentation']
PARAMS = [('shrinking_size', IntTrait(5, 0, 4000, label='Shrinking size')),
]
@stopwatch()
def _run(self, meta_image, container):
image = meta_image.image
if self.params['shrinking_size'] > 0:
nr_objects = container.img_labels.getMinmax()[1] + 1
img_labels = ccore.seeded_region_shrinking(image,
container.img_labels,
nr_objects,
self.params['shrinking_size'])
else:
img_labels = container.img_labels
return ccore.ImageMaskContainer(image, img_labels, False, True, True)
class SectionCluster(SectionCore):
SECTION_NAME = SECTION_NAME_CLUSTER
OPTIONS = [('cluster', [
('position_granularity',
IntTrait(1, 1, 1000, label='Batch size (non-timelapse)')),
])]
class SegmentationPluginModification(_SegmentationPlugin):
LABEL = 'Expansion/shrinking of primary region'
NAME = 'modification'
COLOR = '#FF00FF'
DOC = ":additional_segmentation_plugins"
REQUIRES = ['primary_segmentation']
PARAMS = [('expansion_size', IntTrait(5, 0, 4000, label='Expansion size')),
('shrinking_size', IntTrait(5, 0, 4000, label='Shrinking size')),
]
@stopwatch()
def _run(self, meta_image, container):
image = meta_image.image
if self.params['expansion_size'] > 0 or self.params['shrinking_size'] > 0:
nr_objects = container.img_labels.getMinmax()[1] + 1
if self.params['shrinking_size'] > 0:
img_labelsA = ccore.seeded_region_shrinking(image,
container.img_labels,
nr_objects,
self.params['shrinking_size'])
else:
img_labelsA = container.img_labels
if self.params['expansion_size'] > 0:
img_labelsB = ccore.seeded_region_expansion(image,
container.img_labels,
ccore.SrgType.KeepContours,
nr_objects,
0,
self.params['expansion_size'],
0)
else:
img_labelsB = container.img_labels
img_labels = ccore.substractImages(img_labelsB, img_labelsA)
return ccore.ImageMaskContainer(image, img_labels, False, True, True)
else:
raise ValueError("Parameters are not valid. Requirements: 'expansion_size' > 0 and/or "
"'shrinking_size' > 0")
class SectionGeneral(SectionCore):
SECTION_NAME = SECTION_NAME_GENERAL
OPTIONS = [('general', [
('pathin',
StringTrait('',
label='Image Directory',
widget_info=StringTrait.STRING_PATH)),
('has_multiple_plates', BooleanTrait(False, label='Multiple Plates')),
('pathout',
StringTrait('',
label='Output Directory',
widget_info=StringTrait.STRING_PATH)),
('plate_layout',
StringTrait('',
label='Plate Layout',
widget_info=StringTrait.STRING_PATH)),
('namingscheme',
SelectionTrait(CecogEnvironment.naming_schema.sections()[0],
CecogEnvironment.naming_schema.sections(),
label='Image Naming Scheme')),
('framerange', BooleanTrait(False, label='Timelapse')),
('framerange_begin', IntTrait(1, 0, 10000, label='first')),
('framerange_end', IntTrait(1, 0, 1000, label='last')),
('frameincrement', IntTrait(1, 1, 100, label='increment')),
('skip_finished', BooleanTrait(False,
label='Skip finished positions')),
('constrain_positions', BooleanTrait(False, label='Positions')),
('positions', StringTrait('', 1000, label="", mask='(\w+,)*\w+')),
('process_primary', BooleanTrait(True, label='primary')),
('process_secondary', BooleanTrait(False, label='secondary')),
('process_tertiary', BooleanTrait(False, label='tertiary')),
('process_merged', BooleanTrait(False, label='merged')),
('crop_image', BooleanTrait(False, label='Image cropping')),
('crop_image_x0', IntTrait(-1, -1, 4000, label='upper left x:')),
('crop_image_y0', IntTrait(-1, -1, 4000, label='upper left y')),
('crop_image_x1', IntTrait(-1, -1, 4000, label='lower right x')),
('crop_image_y1', IntTrait(-1, -1, 4000, label='lower right y')),
('crop_image', BooleanTrait(False, label='Image cropping')),
('rendering', DictTrait({}, label='Rendering')),
('version',
StringTrait('',
6,
label='Cecog %s, file version:' % version,
widget_info=StringTrait.STRING_GRAYED)),
('rendering_class', DictTrait({}, label='Rendering class')),
('primary_featureextraction_exportfeaturenames',
ListTrait(['n2_avg', 'n2_stddev', 'roisize'],
label='Primary channel')),
('secondary_featureextraction_exportfeaturenames',
ListTrait(['n2_avg', 'n2_stddev', 'roisize'],
label='Secondary channel')),
('tertiary_featureextraction_exportfeaturenames',
ListTrait(['n2_avg', 'n2_stddev', 'roisize'],
label='Tertiary channel')),
])]
class SegmentationPluginConstrainedWatershed(_SegmentationPlugin):
LABEL = 'Constrained watershed from primary'
NAME = 'constrained_watershed'
COLOR = '#FF99FF'
REQUIRES = ['primary_segmentation']
PARAMS = [
('gauss_filter_size', IntTrait(2, 1, 4, label='Gauss filter size')),
]
@stopwatch()
def _run(self, meta_image, container):
image = meta_image.image
img_labels = self._constrained_watershed(
image,
container.img_labels,
filter_size=self.params['gauss_filter_size'])
return ccore.ImageMaskContainer(image, img_labels, False, True, True)
def _constrained_watershed(self, img_in, img_labels, filter_size=2):
maxlabel = img_labels.getMinmax()[1]
img_bin = ccore.threshold(img_labels, 1, maxlabel, 0, 255)
# internal marker
img_ero = ccore.erode(img_bin, 3, 8)
img_internal_marker = ccore.anchoredSkeleton(img_bin, img_ero)
# external marker
img_inv = ccore.linearRangeMapping(img_bin, 255, 0, 0, 255)
img_voronoi = ccore.watershed(img_inv)
img_external_marker = ccore.threshold(img_voronoi, 0, 0, 0, 255)
# full marker image
img_marker = ccore.supremum(img_internal_marker, img_external_marker)
# gradient image
img_filtered = ccore.gaussianFilter(img_in, filter_size)
img_grad = ccore.morphoGradient(img_filtered, 1, 8)
# Watershed result: 0 is WSL, 1 is Background, all other values correspond to labels.
img_grad_watershed = ccore.constrainedWatershed(img_grad, img_marker)
# we first get the regions
maxreslab = img_grad_watershed.getMinmax()[1]
img_bin2 = ccore.threshold(img_grad_watershed, 2, maxreslab, 0, 255)
img_temp = ccore.copyImageIf(img_labels, img_bin2)
img_out = ccore.relabelImage(img_bin2, img_temp)
return img_out
class SectionFeatureExtraction(SectionCore):
SECTION_NAME = SECTION_NAME_FEATURE_EXTRACTION
OPTIONS = [
('primary_features',
[('primary_featurecategory_%s' % name, BooleanTrait(True, label=desc))
for name, desc in zip(FEATURE_CATEGORIES, FEATURE_CATEGORY_DESC)] + [
('primary_dist_haralick',
StringTrait('1,2,4,8',
200,
label='Haralick: Distances for cooccurence')),
('primary_se_granugrey',
StringTrait('1,2,3,5,7',
200,
label='Granulometry Sizes (Structuring Element)')),
('primary_diameter_spotfeatures',
IntTrait(5, 1, 30, label="Diameter")),
('primary_thresh_spotfeatures',
IntTrait(8, 1, 255, label="Threshold")),
]),
('secondary_features',
[('secondary_featurecategory_%s' % name, BooleanTrait(True,
label=desc))
for name, desc in zip(FEATURE_CATEGORIES, FEATURE_CATEGORY_DESC)] + [
('secondary_dist_haralick',
StringTrait('1,2,4,8',
200,
label='Haralick: Distances for cooccurence')),
('secondary_se_granugrey',
StringTrait('1,2,3,5,7',
200,
label='Granulometry Sizes (Structuring Element)')),
('secondary_diameter_spotfeatures',
IntTrait(5, 1, 30, label="Diameter")),
('secondary_thresh_spotfeatures',
IntTrait(8, 1, 255, label="Threshold")),
]),
('tertiary_features',
[('tertiary_featurecategory_%s' % name, BooleanTrait(True,
label=desc))
for name, desc in zip(FEATURE_CATEGORIES, FEATURE_CATEGORY_DESC)] + [
('tertiary_dist_haralick',
StringTrait('1,2,4,8',
200,
label='Haralick: Distances for cooccurence')),
('tertiary_se_granugrey',
StringTrait('1,2,3,5,7',
200,
label='Granulometry Sizes (Structuring Element)')),
('tertiary_diameter_spotfeatures',
IntTrait(5, 1, 30, label="Diameter")),
('tertiary_thresh_spotfeatures',
IntTrait(8, 1, 255, label="Threshold")),
]),
]
class SectionFeatureExtraction(SectionCore):
SECTION_NAME = SECTION_NAME_FEATURE_EXTRACTION
OPTIONS = [
('primary_features', [
('primary_%s' % name, BooleanTrait(val, label=desc))
for name, (desc, val) in GUI_LABELS.iteritems()
] + [
('primary_dist_haralick',
StringTrait(
'1,2,4,8', 200, label='Haralick: Distances for cooccurence')),
('primary_se_granulometry',
StringTrait('1,2,3,5,7',
200,
label='Granulometry Sizes (Structuring Element)')),
('primary_diameter_spotfeatures',
IntTrait(5, 1, 30, label="Diameter")),
('primary_thresh_spotfeatures',
IntTrait(8, 1, 255, label="Threshold")),
]),
('secondary_features', [
('secondary_%s' % name, BooleanTrait(val, label=desc))
for name, (desc, val) in GUI_LABELS.iteritems()
] + [
('secondary_dist_haralick',
StringTrait(
'1,2,4,8', 200, label='Haralick: Distances for cooccurence')),
('secondary_se_granulometry',
StringTrait('1,2,3,5,7',
200,
label='Granulometry Sizes (Structuring Element)')),
('secondary_diameter_spotfeatures',
IntTrait(5, 1, 30, label="Diameter")),
('secondary_thresh_spotfeatures',
IntTrait(8, 1, 255, label="Threshold")),
]),
('tertiary_features', [
('tertiary_%s' % name, BooleanTrait(val, label=desc))
for name, (desc, val) in GUI_LABELS.iteritems()
] + [
('tertiary_dist_haralick',
StringTrait(
'1,2,4,8', 200, label='Haralick: Distances for cooccurence')),
('tertiary_se_granulometry',
StringTrait('1,2,3,5,7',
200,
label='Granulometry Sizes (Structuring Element)')),
('tertiary_diameter_spotfeatures',
IntTrait(5, 1, 30, label="Diameter")),
('tertiary_thresh_spotfeatures',
IntTrait(8, 1, 255, label="Threshold")),
]),
]
class SectionErrorcorrection(SectionCore):
SECTION_NAME = SECTION_NAME_ERRORCORRECTION
OPTIONS = [
('error_correction',
[('primary', BooleanTrait(True, label='primary')),
('secondary', BooleanTrait(False, label='secondary')),
('tertiary', BooleanTrait(False, label='tertiary')),
('merged', BooleanTrait(False, label='merged')),
('constrain_graph',
BooleanTrait(True, label='Constrain graph')),
('primary_graph',
StringTrait('', 1000, label='Primary file',
widget_info=StringTrait.STRING_FILE)),
('secondary_graph',
StringTrait('', 1000, label='Secondary file',
widget_info=StringTrait.STRING_FILE)),
('tertiary_graph',
StringTrait('', 1000, label='Tertiary file',
widget_info=StringTrait.STRING_FILE)),
('merged_graph',
StringTrait('', 1000, label='Merged ch. file',
widget_info=StringTrait.STRING_FILE)),
('groupby_position',
BooleanTrait(True, label='Position',
widget_info=BooleanTrait.RADIOBUTTON)),
('groupby_sirna',
BooleanTrait(False, label='siRNA',
widget_info=BooleanTrait.RADIOBUTTON)),
('hmm_smoothing',
BooleanTrait(True, label='Smoothing Model',
widget_info=BooleanTrait.RADIOBUTTON)),
('hmm_baumwelch',
BooleanTrait(False, label='Baum-Welch',
widget_info=BooleanTrait.RADIOBUTTON)),
('groupby_genesymbol',
BooleanTrait(False, label='Gene Symbol',
widget_info=BooleanTrait.RADIOBUTTON)),
('overwrite_time_lapse',
BooleanTrait(False, label='Overwrite time-lapse')),
('timelapse',
FloatTrait(1, 0, 2000, digits=2,
label='Time-lapse [min]')),
('max_time',
FloatTrait(-1, -1, 2000, digits=2,
label='Max. time in plot [min]')),
('ignore_tracking_branches',
BooleanTrait(False, label='Ignore tracking branches')),
('enable_sorting',
BooleanTrait(False, label='Sort by phase duration')),
('sorting_sequence',
StringTrait('', 1000, label='Label sequence',
mask='(\w+,)*\w+')),
('primary_sort',
StringTrait('', 100)),
('secondary_sort',
StringTrait('', 100)),
('compose_galleries',
BooleanTrait(False, label='Compose gallery images')),
('compose_galleries_sample',
IntTrait(-1, -1, 10000, label='Max. number of random samples')),
('resampling_factor',
FloatTrait(0.4, 0.01, 1.0, label="Resampling factor")),
('size_gallery_image',
IntTrait(60, 1, 1000, label='Size of gallery images (px)'))
])
]
class SectionPostProcessing(SectionCore):
SECTION_NAME = SECTION_NAME_POST_PROCESSING
OPTIONS = [('post_processing', [
('ibb_analysis', BooleanTrait(True, label='IBB analysis')),
('mappingfile_path',
StringTrait('',
1000,
label='Mapping file path',
widget_info=StringTrait.STRING_PATH)),
('single_plot', BooleanTrait(
True,
label='Export single event',
)),
('single_plot_max_plots', IntTrait(
1,
1,
2000,
label='Max. number',
)),
('ibb_ratio_signal_threshold',
FloatTrait(
1.2,
0.5,
5,
label='IBB minimum ratio signal threshold',
)),
('ibb_range_signal_threshold',
FloatTrait(
3,
0.5,
5,
label='IBB minimum range threshold',
)),
('ibb_onset_factor_threshold',
FloatTrait(
1.2,
1,
5,
label='IBB onset slope threshold',
)),
('nebd_onset_factor_threshold',
FloatTrait(
1.2,
1,
5,
label='NEBD onset slope threshold',
)),
('plot_ylim1_low', IntTrait(
0,
0,
2000,
label='Y-axis limit (low)',
)),
('plot_ylim1_high',
IntTrait(
100,
1,
4000,
label='Y-axis limit (high)',
)),
('single_plot_ylim_low',
FloatTrait(
1,
0,
10,
label='Y-axis ratio range (low)',
)),
('single_plot_ylim_high',
IntTrait(
5,
1,
30,
label='Y-axis ratio range (high)',
)),
('group_by_position',
BooleanTrait(True,
label='Position',
widget_info=BooleanTrait.RADIOBUTTON)),
('group_by_oligoid',
BooleanTrait(False,
label='Oligo ID',
widget_info=BooleanTrait.RADIOBUTTON)),
('group_by_genesymbol',
BooleanTrait(False,
label='Gene symbol',
widget_info=BooleanTrait.RADIOBUTTON)),
('group_by_group',
BooleanTrait(False,
label='Group',
widget_info=BooleanTrait.RADIOBUTTON)),
('color_sort_by_position',
BooleanTrait(False,
label='Position',
widget_info=BooleanTrait.RADIOBUTTON)),
('color_sort_by_oligoid',
BooleanTrait(True,
label='Oligo ID',
widget_info=BooleanTrait.RADIOBUTTON)),
('color_sort_by_genesymbol',
BooleanTrait(False,
label='Gene symbol',
widget_info=BooleanTrait.RADIOBUTTON)),
('color_sort_by_group',
BooleanTrait(False,
label='Group',
widget_info=BooleanTrait.RADIOBUTTON)),
('securin_analysis', BooleanTrait(True, label='Securin analysis')),
])]
class SectionTracking(SectionCore):
SECTION_NAME = SECTION_NAME_TRACKING
OPTIONS = [
('tracking', [
('tracking_regionname',
SelectionTrait2(None, [], label='Region name')),
('tracking_maxobjectdistance',
IntTrait(0, 0, 4000, label='Max object x-y distance')),
('tracking_maxtrackinggap',
IntTrait(0, 0, 4000, label='Max time-point gap')),
('tracking_maxsplitobjects',
IntTrait(0, 0, 4000, label='Max split events')),
('tracking_maxindegree', IntTrait(1,
0,
4000,
label='Max in-degree')),
('tracking_maxoutdegree',
IntTrait(2, 0, 4000, label='Max out-degree')),
('tracking_exporttrackfeatures',
BooleanTrait(True, label='Export tracks')),
('tracking_compressiontrackfeatures',
SelectionTrait(COMPRESSION_FORMATS[0],
COMPRESSION_FORMATS,
label='Compression')),
]),
('event_selection', [
('tracking_labeltransitions',
StringTrait('',
200,
label='Class transition motif(s)',
mask='(\(\d+,\d+\),)*\(\d+,\d+\)')),
('tracking_backwardrange',
FloatTrait(0, -1, 4000, label='Duration [pre]')),
('tracking_forwardrange',
FloatTrait(0, -1, 4000, label='Duration [post]')),
('tracking_backwardlabels',
StringTrait('',
200,
label='Class filter [pre]',
mask='(\d+,)*\d+')),
('tracking_forwardlabels',
StringTrait('',
200,
label='Class filter [post]',
mask='(\d+,)*\d+')),
('tracking_backwardcheck',
FloatTrait(2, 0, 4000, label='Filter duration [pre]')),
('tracking_forwardcheck',
FloatTrait(2, 0, 4000, label='Filter duration [post]')),
('tracking_duration_unit',
SelectionTrait2(TRACKING_DURATION_UNITS_DEFAULT[0],
TRACKING_DURATION_UNITS_DEFAULT,
label='Duration unit')),
('tracking_backwardrange_min', BooleanTrait(False, label='Min.')),
('tracking_forwardrange_min', BooleanTrait(False, label='Min.')),
]),
('visualization', [
('tracking_visualization',
BooleanTrait(False, label='Visualization')),
('tracking_visualize_track_length',
IntTrait(5, -1, 10000, label='Max. time-points')),
('tracking_centroid_radius',
IntTrait(3, -1, 50, label='Centroid radius')),
]),
]
class SectionOutput(SectionCore):
SECTION_NAME = SECTION_NAME_OUTPUT
OPTIONS = [
('output', [
('text_output',
BooleanTrait(False,
label='Text Output',
widget_info=BooleanTrait.RADIOBUTTON)),
('rendering_labels_discwrite',
BooleanTrait(False, label='Label images')),
('rendering_contours_discwrite',
BooleanTrait(False, label='Contour images')),
('rendering_contours_showids',
BooleanTrait(False, label='Show object IDs')),
('rendering_class_discwrite',
BooleanTrait(False, label='Classification images')),
('rendering_class_showids',
BooleanTrait(False, label='Show object IDs')),
('rendering_channel_gallery',
BooleanTrait(False, label='Merged channel gallery')),
('export_object_counts',
BooleanTrait(False, label='Export object counts')),
('export_object_counts_ylim_max',
IntTrait(-1, -1, 1000, label='Max. count in plot')),
('export_object_details',
BooleanTrait(False, label='Export detailed object data')),
('export_file_names',
BooleanTrait(False, label='Export raw image file names')),
('export_tracking_as_dot',
BooleanTrait(False, label='Export tracking as GraphViz .dot')),
('export_track_data',
BooleanTrait(False, label='Export tracks for head nodes')),
('export_events', BooleanTrait(False, label='Export event data')),
('events_export_all_features',
BooleanTrait(False, label='Export all features per event')),
('events_export_gallery_images',
BooleanTrait(False, label='Export gallery images')),
('events_gallery_image_size',
IntTrait(50, 1, 1000, label='Gallery image size (pixel)')),
]),
('hdf5', [
('hdf5_create_file',
BooleanTrait(True,
label='Create CellH5',
widget_info=BooleanTrait.RADIOBUTTON)),
('hdf5_reuse', BooleanTrait(False, label='Reuse CellH5')),
('minimal_effort', BooleanTrait(False,
label='Only necessary steps')),
('hdf5_include_raw_images',
BooleanTrait(False, label='Include 8-bit image data')),
('hdf5_include_label_images',
BooleanTrait(False, label='Include segmentation images')),
('hdf5_include_crack',
BooleanTrait(False, label='Include crack contours')),
('hdf5_include_features',
BooleanTrait(False, label='Include features')),
('hdf5_include_classification',
BooleanTrait(False, label='Include classification')),
('hdf5_include_tracking',
BooleanTrait(False, label='Include tracking')),
('hdf5_include_events', BooleanTrait(False,
label='Include events')),
('hdf5_compression',
BooleanTrait(True,
label='Enable gzip compression (recommended!)')),
('hdf5_merge_positions',
BooleanTrait(True, label='Merge positions into one file')),
]),
]
class SectionGeneral(SectionCore):
SECTION_NAME = SECTION_NAME_GENERAL
OPTIONS = [('general', [
('pathin',
StringTrait('',
1000,
label='Image folder',
widget_info=StringTrait.STRING_PATH)),
('has_multiple_plates', BooleanTrait(False, label='Multiple plates')),
('pathout',
StringTrait('',
1000,
label='Analysis folder',
widget_info=StringTrait.STRING_PATH)),
('image_import_namingschema',
BooleanTrait(True,
label='Import via naming schema',
widget_info=BooleanTrait.RADIOBUTTON)),
('image_import_structurefile',
BooleanTrait(False,
label='Import via coordinate file',
widget_info=BooleanTrait.RADIOBUTTON)),
('namingscheme',
SelectionTrait(CecogEnvironment.naming_schema.sections()[0],
CecogEnvironment.naming_schema.sections(),
label='Naming scheme')),
('structure_filename',
StringTrait('',
1000,
label='Coordinate filename',
widget_info=StringTrait.STRING_FILE)),
('structure_file_pathin',
BooleanTrait(True,
label='Image folder',
widget_info=BooleanTrait.RADIOBUTTON)),
('structure_file_pathout',
BooleanTrait(False,
label='Analysis folder',
widget_info=BooleanTrait.RADIOBUTTON)),
('structure_file_extra_path',
BooleanTrait(False,
label='Different location',
widget_info=BooleanTrait.RADIOBUTTON)),
('structure_file_extra_path_name',
StringTrait('',
1000,
label='Path',
widget_info=StringTrait.STRING_PATH)),
('constrain_positions', BooleanTrait(False,
label='Constrain positions')),
('positions',
StringTrait('', 1000, label='Positions', mask='(\w+,)*\w+')),
('crop_image', BooleanTrait(False, label='Crop image')),
('crop_image_x0', IntTrait(-1, -1, 4000, label='Upper left X')),
('crop_image_y0', IntTrait(-1, -1, 4000, label='Upper left Y')),
('crop_image_x1', IntTrait(-1, -1, 4000, label='Lower right X')),
('crop_image_y1', IntTrait(-1, -1, 4000, label='Lower right Y')),
('crop_image', BooleanTrait(False, label='Crop image')),
('redofailedonly', BooleanTrait(True,
label='Skip processed positions')),
('framerange', BooleanTrait(False, label='Constrain timepoints')),
('framerange_begin', IntTrait(1, 0, 10000, label='Begin')),
('framerange_end', IntTrait(1, 0, 1000, label='End')),
('frameincrement', IntTrait(1, 1, 100, label='Timepoint increment')),
('rendering', DictTrait({}, label='Rendering')),
('version',
StringTrait('',
6,
label='Cecog %s, file version:' % VERSION,
widget_info=StringTrait.STRING_GRAYED)),
('rendering_class', DictTrait({}, label='Rendering class')),
('primary_featureextraction_exportfeaturenames',
ListTrait(['n2_avg', 'n2_stddev', 'roisize'],
label='Primary channel')),
('secondary_featureextraction_exportfeaturenames',
ListTrait(['n2_avg', 'n2_stddev', 'roisize'],
label='Secondary channel')),
('tertiary_featureextraction_exportfeaturenames',
ListTrait(['n2_avg', 'n2_stddev', 'roisize'],
label='Tertiary channel')),
])]
class SectionObjectdetection(SectionCore):
SECTION_NAME = SECTION_NAME_OBJECTDETECTION
OPTIONS = [
('primary_image',
[('primary_channelid',
SelectionTrait2(None, [], label='Primary channel ID')),
('primary_normalizemin',
IntTrait(0, -2**16, 2**16, label='Min.')),
('primary_normalizemax',
IntTrait(255, -2**16, 2**16, label='Max.')),
('primary_zslice_selection',
BooleanTrait(True, label='Z-slice selection',
widget_info=BooleanTrait.RADIOBUTTON)),
('primary_flat_field_correction',
BooleanTrait(False, label='Z-slice flat field correction',
widget_info=BooleanTrait.CHECKBOX)),
('primary_flat_field_correction_image_dir',
StringTrait('', 1000, label='Correction image directory',
widget_info=StringTrait.STRING_PATH)),
('primary_zslice_selection_slice',
IntTrait(1, 0, 1000, label='Slice')),
('primary_zslice_projection',
BooleanTrait(False, label='Z-slice projection',
widget_info=BooleanTrait.RADIOBUTTON)),
('primary_zslice_projection_method',
SelectionTrait(ZSLICE_PROJECTION_METHODS[0],
ZSLICE_PROJECTION_METHODS, label='Method')),
('primary_zslice_projection_begin',
IntTrait(1, 0, 1000, label='Begin')),
('primary_zslice_projection_end',
IntTrait(1, 0, 1000, label='End')),
('primary_zslice_projection_step',
IntTrait(1, 1, 1000, label='Step')),
# these two options are nopes, just to have no
# special casing in channel classes
('primary_channelregistration_x',
IntTrait(0, -99999, 99999, label='Shift X')),
('primary_channelregistration_y',
IntTrait(0, -99999, 99999, label='Shift Y')),
]),
] + \
unlist(
[[('%s_image' % prefix,
[('%s_channelid' % prefix,
SelectionTrait2(None, [], label='%s channel ID' % name)),
('%s_normalizemin' % prefix,
IntTrait(0, -2**16, 2**16, label='Min.')),
('%s_normalizemax' % prefix,
IntTrait(255, -2**16, 2**16, label='Max.')),
('%s_zslice_selection' % prefix,
BooleanTrait(True, label='Z-slice selection',
widget_info=BooleanTrait.RADIOBUTTON)),
('%s_flat_field_correction' % prefix,
BooleanTrait(False, label='Z-slice flat field correction',
widget_info=BooleanTrait.CHECKBOX)),
('%s_flat_field_correction_image_dir' % prefix,
StringTrait('', 1000, label='Correction image directory',
widget_info=StringTrait.STRING_PATH)),
('%s_zslice_selection_slice' % prefix,
IntTrait(1, 0, 1000, label='Slice')),
('%s_zslice_projection' % prefix,
BooleanTrait(False, label='Z-slice projection',
widget_info=BooleanTrait.RADIOBUTTON)),
('%s_zslice_projection_method' % prefix,
SelectionTrait(ZSLICE_PROJECTION_METHODS[0],
ZSLICE_PROJECTION_METHODS, label='Method')),
('%s_zslice_projection_begin' % prefix,
IntTrait(1, 0, 1000, label='Begin')),
('%s_zslice_projection_end' % prefix,
IntTrait(1, 0, 1000, label='End')),
('%s_zslice_projection_step' % prefix,
IntTrait(1, 1, 1000, label='Step')),
]),
('%s_registration' % prefix,
[('%s_channelregistration_x' % prefix,
IntTrait(0, -99999, 99999, label='Shift X')),
('%s_channelregistration_y' % prefix,
IntTrait(0, -99999, 99999, label='Shift Y')),
]),
]
for name, prefix in [('Secondary', 'secondary'),
('Tertiary', 'tertiary'),
# moste merged channel options are nopes
# to avoid special casing
('Merged', 'merged')
]]
)
class SectionErrorcorrection(SectionCore):
SECTION_NAME = SECTION_NAME_ERRORCORRECTION
OPTIONS = [('error_correction', [
('filename_to_r',
StringTrait('',
1000,
label='R-project executable',
widget_info=StringTrait.STRING_FILE)),
('constrain_graph', BooleanTrait(True, label='Constrain graph')),
('primary_graph',
StringTrait('',
1000,
label='Primary file',
widget_info=StringTrait.STRING_FILE)),
('secondary_graph',
StringTrait('',
1000,
label='Secondary file',
widget_info=StringTrait.STRING_FILE)),
('skip_processed_plates',
BooleanTrait(False, label='Skip processed plates')),
('position_labels', BooleanTrait(False, label='Position labels')),
('mappingfile_path',
StringTrait('',
1000,
label='Path',
widget_info=StringTrait.STRING_PATH)),
('groupby_position',
BooleanTrait(True,
label='Position',
widget_info=BooleanTrait.RADIOBUTTON)),
('groupby_oligoid',
BooleanTrait(False,
label='Oligo ID',
widget_info=BooleanTrait.RADIOBUTTON)),
('groupby_genesymbol',
BooleanTrait(False,
label='Gene symbol',
widget_info=BooleanTrait.RADIOBUTTON)),
('overwrite_time_lapse',
BooleanTrait(False, label='Overwrite time-lapse')),
('timelapse', FloatTrait(1,
0,
2000,
digits=2,
label='Time-lapse [min]')),
('max_time',
FloatTrait(100, 1, 2000, digits=2, label='Max. time in plot [min]')),
('ignore_tracking_branches',
BooleanTrait(False, label='Ignore tracking branches')),
('show_html', BooleanTrait(True, label='Open in browser')),
('enable_sorting', BooleanTrait(False,
label='Sort by phase duration')),
('sorting_sequence',
StringTrait('', 1000, label='Label sequence', mask='(\w+,)*\w+')),
('primary_sort', StringTrait('', 100)),
('secondary_sort', StringTrait('', 100)),
('compose_galleries',
BooleanTrait(False, label='Compose gallery images')),
('compose_galleries_sample',
IntTrait(-1, -1, 10000, label='Max. number of random samples')),
])]
class SegmentationPluginPrimary(_SegmentationPlugin):
LABEL = 'Local adaptive threshold w/ split&merge'
NAME = 'primary'
COLOR = '#FF0000'
REQUIRES = None
PARAMS = [('medianradius', IntTrait(2, 0, 1000, label='Median radius')),
('latwindowsize', IntTrait(20, 1, 1000, label='Window size')),
('latlimit', IntTrait(1, 0, 255, label='Min. contrast')),
('lat2', BooleanTrait(False, label='Local adaptive threshold 2')),
('latwindowsize2', IntTrait(20, 1, 1000, label='Window size')),
('latlimit2', IntTrait(1, 0, 255, label='Min. contrast')),
('shapewatershed', BooleanTrait(False, label='Split & merge by shape')),
('shapewatershed_gausssize', IntTrait(1, 0, 1000000, label='Gauss radius')),
('shapewatershed_maximasize', IntTrait(1, 0, 1000000, label='Min. seed distance')),
('shapewatershed_minmergesize', IntTrait(1, 0, 1000000, label='Object size threshold')),
('intensitywatershed', BooleanTrait(False, label='Split & merge by intensity')),
('intensitywatershed_gausssize', IntTrait(1, 0, 1000000, label='Gauss radius')),
('intensitywatershed_maximasize', IntTrait(1, 0, 1000000, label='Min. seed distance')),
('intensitywatershed_minmergesize', IntTrait(1, 0, 1000000, label='Object size threshold')),
('postprocessing', BooleanTrait(False, label='Object filter')),
('postprocessing_roisize_min', IntTrait(-1, -1, 1000000, label='Min. object size')),
('postprocessing_roisize_max', IntTrait(-1, -1, 1000000, label='Max. object size')),
('postprocessing_intensity_min', IntTrait(-1, -1, 1000000, label='Min. average intensity')),
('postprocessing_intensity_max', IntTrait(-1, -1, 1000000, label='Max. average intensity')),
('removeborderobjects', BooleanTrait(True, label='Remove border objects')),
('holefilling', BooleanTrait(True, label='Fill holes')),
]
# the : at the beginning indicates a QRC link with alias 'plugins/segmentation/local_adaptive_threshold'
DOC = ':local_adaptive_threshold'
def render_to_gui(self, panel):
panel.add_group(None,
[('medianradius', (0, 0, 1, 1)),
('latwindowsize', (0, 1, 1, 1)),
('latlimit', (0, 2, 1, 1)),
], link='lat', label='Local adaptive threshold')
panel.add_group('lat2',
[('latwindowsize2', (0, 0, 1, 1)),
('latlimit2', (0, 1, 1, 1)),
])
panel.add_input('holefilling')
panel.add_input('removeborderobjects')
panel.add_group('shapewatershed',
[('shapewatershed_gausssize', (0, 0, 1, 1)),
('shapewatershed_maximasize', (0, 1, 1, 1)),
('shapewatershed_minmergesize', (1, 0, 1, 1)),
])
panel.add_group('postprocessing',
[('postprocessing_roisize_min', (0, 0, 1, 1)),
('postprocessing_roisize_max', (0, 1, 1, 1)),
('postprocessing_intensity_min', (1, 0, 1, 1)),
('postprocessing_intensity_max', (1, 1, 1, 1)),
])
@stopwatch()
def prefilter(self, img_in, radius=None):
if radius is None:
radius = self.params['medianradius']
img_out = ccore.disc_median(img_in, radius)
return img_out
@stopwatch()
def threshold(self, img_in, size, limit):
img_out = ccore.window_average_threshold(img_in, size, limit)
return img_out
@stopwatch()
def correct_segmetation(self, img_in, img_bin, border, gauss_size,
max_dist, min_merge_size, kind='shape'):
if kind == 'shape':
f = ccore.segmentation_correction_shape
else:
f = ccore.segmentation_correction_intensity
return f(img_in, img_bin, border, gauss_size, max_dist, min_merge_size)
@stopwatch()
def postprocessing(self, container, is_active, roisize_minmax,
intensity_minmax, delete_objects=True):
valid_ids = container.getObjects().keys()
rejected_ids = []
if is_active:
feature_categories = set()
conditions = []
for idx, (roisize, intensity) in enumerate( \
zip(roisize_minmax, intensity_minmax)):
cmprt = '>=' if idx == 0 else '<='
if roisize > -1:
feature_categories.add('roisize')
conditions.append('roisize %s %d' % (cmprt, roisize))
if intensity > -1:
feature_categories.add('normbase2')
conditions.append('n2_avg %s %d' % (cmprt, intensity))
if len(conditions) > 0:
conditions_str = ' and '.join(conditions)
# extract features needed for the filter
# FIXME: features are currently kept in the ObjectContainer and used for classification automatically
for feature in feature_categories:
container.applyFeature(feature)
valid_ids = []
rejected_ids = []
# get a dict copy, because we delete elements from the dict
objects = container.getObjects()
for obj_id, obj in objects.iteritems():
# eval condition string based on the feature dict (provides values for the features above)
if not eval(conditions_str, obj.getFeatures()):
if delete_objects:
container.delObject(obj_id)
rejected_ids.append(obj_id)
else:
valid_ids.append(obj_id)
# store valid and rejected object IDs to the container
container.valid_ids = valid_ids
container.rejected_ids = rejected_ids
@stopwatch()
def _run(self, meta_image):
image = meta_image.image
img_prefiltered = self.prefilter(image)
img_bin1 = self.threshold(img_prefiltered, self.params['latwindowsize'], self.params['latlimit'])
if self.params['holefilling']:
ccore.fill_holes(img_bin1, False)
if self.params['lat2']:
img_bin2 = self.threshold(img_prefiltered, self.params['latwindowsize2'],
self.params['latlimit2'])
# replacement for not working ccore.projectImage
img_bin = numpy.zeros((img_bin2.height, img_bin2.width),
dtype=meta_image.format)
img_bin = ccore.numpy_to_image(img_bin, copy=True)
ccore.zproject(img_bin, [img_bin1, img_bin2], ccore.ProjectionType.MaxProjection)
else:
img_bin = img_bin1
if self.params['shapewatershed']:
img_bin = self.correct_segmetation(img_prefiltered, img_bin,
self.params['latwindowsize'],
self.params['shapewatershed_gausssize'],
self.params['shapewatershed_maximasize'],
self.params['shapewatershed_minmergesize'],
kind='shape')
if self.params['intensitywatershed']:
img_bin = self.correct_segmetation(img_prefiltered, img_bin,
self.params['latwindowsize'],
self.params['intensitywatershed_gausssize'],
self.params['intensitywatershed_maximasize'],
self.params['intensitywatershed_minmergesize'],
kind='intensity')
container = ccore.ImageMaskContainer(image, img_bin, self.params['removeborderobjects'])
self.postprocessing(container, self.params['postprocessing'],
(self.params['postprocessing_roisize_min'], self.params['postprocessing_roisize_max']),
(self.params['postprocessing_intensity_min'], self.params['postprocessing_intensity_max']))
return container
class SegmentationPluginPrimary3(_SegmentationPlugin):
LABEL = 'Local adaptive threshold, togglemappings, split by minima depth, background corrected object filter'
NAME = 'primary3'
COLOR = '#FF0000'
REQUIRES = None
PARAMS = [('medianradius', IntTrait(2, 0, 1000, label='Median radius')),
('togglemappings', BooleanTrait(False, label='Toggle Mappings')),
('tm_size', IntTrait(1, 0, 20, label='Toggle Mappings Size')),
('latwindowsize', IntTrait(20, 1, 1000, label='Window size')),
('latlimit', IntTrait(1, 0, 255, label='Min. contrast')),
('lat2', BooleanTrait(False, label='Local adaptive threshold 2')),
('latwindowsize2', IntTrait(20, 1, 1000, label='Window size')),
('latlimit2', IntTrait(1, 0, 255, label='Min. contrast')),
('watershed_distance', BooleanTrait(False, label='Watershed (Distance)')),
('watershed_dynamic', IntTrait(2, 0, 1000, label='Minimal depth')),
('watershed_used_distance', IntTrait(2, 0, 2, label='Distance Metric (0: chessboard, 1: L1, 2: L2)')),
('postprocessing', BooleanTrait(False, label='Object filter')),
('postprocessing_roisize_min', IntTrait(-1, -1, 1000000, label='Min. object size')),
('postprocessing_roisize_max', IntTrait(-1, -1, 1000000, label='Max. object size')),
('postprocessing_intensity_min', IntTrait(-1, -1, 1000000, label='Min. average intensity above background')),
('postprocessing_intensity_max', IntTrait(-1, -1, 1000000, label='Max. average intensity above background')),
('removeborderobjects', BooleanTrait(True, label='Remove border objects')),
('holefilling', BooleanTrait(True, label='Fill holes')),
]
# the : at the beginning indicates a QRC link with alias 'plugins/segmentation/local_adaptive_threshold'
DOC = ':local_adaptive_threshold'
def render_to_gui(self, panel):
panel.add_group('togglemappings',
[('tm_size', (0, 0, 1, 1)),
])
panel.add_group(None,
[('medianradius', (0, 0, 1, 1)),
('latwindowsize', (0, 1, 1, 1)),
('latlimit', (0, 2, 1, 1)),
], link='lat', label='Local adaptive threshold')
panel.add_group('lat2',
[('latwindowsize2', (0, 0, 1, 1)),
('latlimit2', (0, 1, 1, 1)),
])
panel.add_input('holefilling')
panel.add_input('removeborderobjects')
panel.add_group('watershed_distance',
[('watershed_dynamic', (0, 0, 1, 1)),
('watershed_used_distance', (0, 1, 1, 1)),
])
# panel.add_group('shapewatershed',
# [('shapewatershed_gausssize', (0, 0, 1, 1)),
# ('shapewatershed_maximasize', (0, 1, 1, 1)),
# ('shapewatershed_minmergesize', (1, 0, 1, 1)),
# ])
panel.add_group('postprocessing',
[('postprocessing_roisize_min', (0, 0, 1, 1)),
('postprocessing_roisize_max', (0, 1, 1, 1)),
('postprocessing_intensity_min', (1, 0, 1, 1)),
('postprocessing_intensity_max', (1, 1, 1, 1)),
])
@stopwatch()
def prefilter(self, img_in, radius=None):
img_temp = img_in
if self.params['togglemappings']:
img_temp = ccore.toggle_mapping(img_in, self.params['tm_size'])
if radius is None:
radius = self.params['medianradius']
img_out = ccore.disc_median(img_temp, radius)
return img_out
@stopwatch()
def threshold(self, img_in, size, limit):
img_out = ccore.window_average_threshold(img_in, size, limit)
return img_out
@stopwatch()
def OLD_correct_segmetation(self, img_in, img_bin, border, gauss_size,
max_dist, min_merge_size, kind='shape'):
if kind == 'shape':
f = ccore.segmentation_correction_shape
else:
f = ccore.segmentation_correction_intensity
return f(img_in, img_bin, border, gauss_size, max_dist, min_merge_size)
@stopwatch()
def correct_segmetation(self, img_in, img_bin, dyn, distance=2):
if distance==2:
# Euclidean distance
res = ccore.watershed_dynamic_split(img_bin, dyn, 4, 2)
elif distance==1:
# we use connectivity 4 (for the watershed) and distance mode 1 (which
# corresponds to the L1 norm which corresponds to the graph distance
# of a 4-neighborhood graph
res = ccore.watershed_dynamic_split(img_bin, dyn, 4, 1)
elif distance==0:
# the chessboard distance and 8 connectivity for the watershed algorithm.
# However, the distances are "deeper" for 4-connectivity.
res = ccore.watershed_dynamic_split(img_bin, dyn, 8, 0)
else:
print 'not implemented'
res = img_bin
return res
@stopwatch()
def postprocessing(self, container, is_active, roisize_minmax,
intensity_minmax, delete_objects=True,
offset=0):
valid_ids = container.getObjects().keys()
rejected_ids = []
if is_active:
feature_categories = set()
conditions = []
for idx, (roisize, intensity) in enumerate( \
zip(roisize_minmax, intensity_minmax)):
cmprt = '>=' if idx == 0 else '<='
if roisize > -1:
feature_categories.add('roisize')
conditions.append('roisize %s %d' % (cmprt, roisize))
if intensity > -1:
feature_categories.add('normbase2')
conditions.append('n2_avg %s %d' % (cmprt, intensity+offset))
if len(conditions) > 0:
conditions_str = ' and '.join(conditions)
# extract features needed for the filter
# FIXME: features are currently kept in the ObjectContainer and used for classification automatically
# Features can be removed from the container, but it remains much better a choice
# to restrict the feature sets used for classification.
for feature in feature_categories:
container.applyFeature(feature)
valid_ids = []
rejected_ids = []
# get a dict copy, because we delete elements from the dict
objects = container.getObjects()
for obj_id, obj in objects.iteritems():
# eval condition string based on the feature dict (provides values for the features above)
if not eval(conditions_str, obj.getFeatures()):
if delete_objects:
container.delObject(obj_id)
rejected_ids.append(obj_id)
else:
valid_ids.append(obj_id)
#pdb.set_trace()
#img_v = container.img.
# delete features that were added by the object filter
for feature in ['roisize', 'normbase2']:
container.deleteFeatureCategory(feature)
# store valid and rejected object IDs to the container
container.valid_ids = valid_ids
container.rejected_ids = rejected_ids
@stopwatch()
def _run(self, meta_image):
image = meta_image.image
img_prefiltered = self.prefilter(image)
img_bin1 = self.threshold(img_prefiltered, self.params['latwindowsize'], self.params['latlimit'])
if self.params['holefilling']:
ccore.fill_holes(img_bin1, False)
if self.params['lat2']:
img_bin2 = self.threshold(img_prefiltered, self.params['latwindowsize2'],
self.params['latlimit2'])
# replacement for not working ccore.projectImage
img_bin = numpy.zeros((img_bin2.height, img_bin2.width),
dtype=meta_image.format)
img_bin = ccore.numpy_to_image(img_bin, copy=True)
ccore.zproject(img_bin, [img_bin1, img_bin2], ccore.ProjectionType.MaxProjection)
else:
img_bin = img_bin1
if self.params['watershed_distance']:
img_bin = self.correct_segmetation(img_prefiltered, img_bin,
self.params['watershed_dynamic'],
self.params['watershed_used_distance'])
# if self.params['shapewatershed']:
# img_bin = self.correct_segmetation(img_prefiltered, img_bin,
# self.params['latwindowsize'],
# self.params['shapewatershed_gausssize'],
# self.params['shapewatershed_maximasize'],
# self.params['shapewatershed_minmergesize'],
# kind='shape')
# if self.params['intensitywatershed']:
# img_bin = self.correct_segmetation(img_prefiltered, img_bin,
# self.params['latwindowsize'],
# self.params['intensitywatershed_gausssize'],
# self.params['intensitywatershed_maximasize'],
# self.params['intensitywatershed_minmergesize'],
# kind='intensity')
container = ccore.ImageMaskContainer(image, img_bin, self.params['removeborderobjects'])
# calculate offset: mean on the background region, as given by the segmentation result
# no locality: simply a global mean on the image.
np_image = image.toArray(True)
np_img_bin = img_bin.toArray(True)
offset = np_image[np_img_bin==0].mean()
self.postprocessing(container, self.params['postprocessing'],
(self.params['postprocessing_roisize_min'], self.params['postprocessing_roisize_max']),
(self.params['postprocessing_intensity_min'], self.params['postprocessing_intensity_max']),
offset=offset)
return container
class SegmentationPluginIlastik(SegmentationPluginPrimary):
LABEL = 'Local adaptive threshold w/ split&merge using trained ilastik classifier'
NAME = 'primary_ilastik'
COLOR = '#FF0000'
REQUIRES = None
PARAMS = [('ilastik_classifier', StringTrait('', 1000, label='ilastik Classifier File',
widget_info=StringTrait.STRING_FILE)),
('ilastik_class_selector', IntTrait(1, 0, 1000, label='Output class')),
('medianradius', IntTrait(2, 0, 1000, label='Median radius')),
('latwindowsize', IntTrait(20, 1, 1000, label='Window size')),
('latlimit', IntTrait(1, 0, 255, label='Min. contrast')),
('lat2', BooleanTrait(False, label='Local adaptive threshold 2')),
('latwindowsize2', IntTrait(20, 1, 1000, label='Window size')),
('latlimit2', IntTrait(1, 0, 255, label='Min. contrast')),
('shapewatershed', BooleanTrait(False, label='Split & merge by shape')),
('shapewatershed_gausssize', IntTrait(1, 0, 10000, label='Gauss radius')),
('shapewatershed_maximasize', IntTrait(1, 0, 10000, label='Min. seed distance')),
('shapewatershed_minmergesize', IntTrait(1, 0, 10000, label='Object size threshold')),
('intensitywatershed', BooleanTrait(False, label='Split & merge by intensity')),
('intensitywatershed_gausssize', IntTrait(1, 0, 10000, label='Gauss radius')),
('intensitywatershed_maximasize', IntTrait(1, 0, 10000, label='Min. seed distance')),
('intensitywatershed_minmergesize', IntTrait(1, 0, 10000, label='Object size threshold')),
('postprocessing', BooleanTrait(False, label='Object filter')),
('postprocessing_roisize_min', IntTrait(-1, -1, 10000, label='Min. object size')),
('postprocessing_roisize_max', IntTrait(-1, -1, 10000, label='Max. object size')),
('postprocessing_intensity_min', IntTrait(-1, -1, 10000, label='Min. average intensity')),
('postprocessing_intensity_max', IntTrait(-1, -1, 10000, label='Max. average intensity')),
('removeborderobjects', BooleanTrait(True, label='Remove border objects')),
('holefilling', BooleanTrait(True, label='Fill holes')),
]
# the : at the beginning indicates a QRC link with alias 'plugins/segmentation/local_adaptive_threshold'
DOC = ':local_adaptive_threshold'
@stopwatch()
def prefilter(self, img_in):
img = SegmentationPluginPrimary.prefilter(self, img_in)
np_img = img.toArray(True)
return self._predict_image_with_ilastik(np_img)
def threshold(self, img_in, *args):
np_img = img_in.toArray(True)
return ccore.numpy_to_image((np_img > 128).astype(numpy.uint8), True)
def render_to_gui(self, panel):
panel.add_group(None, [('ilastik_classifier', (0, 0, 1, 1)),
('ilastik_class_selector', (1, 0, 1, 1)),
], label='ilastik')
SegmentationPluginPrimary.render_to_gui(self, panel)
def _predict_image_with_ilastik(self, image_):
import ilastik
from ilastik.core.dataMgr import DataMgr, DataItemImage
from ilastik.modules.classification.core.featureMgr import FeatureMgr
from ilastik.modules.classification.core.classificationMgr import ClassificationMgr
from ilastik.modules.classification.core.features.featureBase import FeatureBase
from ilastik.modules.classification.core.classifiers.classifierRandomForest import ClassifierRandomForest
from ilastik.modules.classification.core.classificationMgr import ClassifierPredictThread
from ilastik.core.volume import DataAccessor
import numpy, h5py
dataMgr = DataMgr()
# Transform input image to ilastik convention s
# 3D = (time,x,y,z,channel)
# 2D = (time,1,x,y,channel)
# Note, this work for 2D images right now. Is there a need for 3D
image_.shape = (1,1) + image_.shape
# Check if image_ has channels, if not add singelton dimension
if len(image_.shape) == 4:
image_.shape = image_.shape + (1,)
# Add data item di to dataMgr
di = DataItemImage('')
di.setDataVol(DataAccessor(image_))
dataMgr.append(di, alreadyLoaded=True)
fileName = self.params["ilastik_classifier"]
ilastik_class = self.params["ilastik_class_selector"]
hf = h5py.File(fileName,'r')
temp = hf['classifiers'].keys()
# If hf is not closed this leads to an error in win64 and mac os x
hf.close()
del hf
classifiers = []
for cid in temp:
cidpath = 'classifiers/' + cid
classifiers.append(ClassifierRandomForest.loadRFfromFile(fileName, str(cidpath)))
dataMgr.module["Classification"]["classificationMgr"].classifiers = classifiers
# Restore user selection of feature items from hdf5
featureItems = []
f = h5py.File(fileName,'r')
for fgrp in f['features'].values():
featureItems.append(FeatureBase.deserialize(fgrp))
f.close()
del f
fm = FeatureMgr(dataMgr, featureItems)
# Create FeatureMgr
# Compute features
fm.prepareCompute(dataMgr)
fm.triggerCompute()
fm.joinCompute(dataMgr)
# Predict with loaded classifier
classificationPredict = ClassifierPredictThread(dataMgr)
classificationPredict.start()
classificationPredict.wait()
if ilastik_class >= classificationPredict._prediction[0].shape[-1]:
raise RuntimeError('ilastik output class not valid...')
# Produce output image and select the probability map
probMap = (classificationPredict._prediction[0][0,0,:,:, ilastik_class] * 255).astype(numpy.uint8)
img_out = ccore.numpy_to_image(probMap, True)
return img_out