def test_compute_fsd_features(self):
Fs = 6
expected_feature_list = ['Shape.FSD' + str(i + 1) for i in range(Fs)]
fdata = htk_features.compute_fsd_features(cfg.im_nuclei_seg_mask,
Fs=Fs)
check_fdata_sanity(fdata, expected_feature_list)
check_fdata_sanity(cfg.fdata_nuclei,
expected_feature_list,
match_feature_count=False)
if GENERATE_GROUNDTRUTH:
fdata.to_csv(os.path.join(tempfile.gettempdir(),
'Easy1_nuclei_fsd_features.csv'),
index=False)
fdata_gtruth = pd.read_csv(
utilities.getTestFilePath('Easy1_nuclei_fsd_features.csv'))
pd.testing.assert_frame_equal(fdata,
fdata_gtruth,
check_less_precise=2)
def test_compute_morphometry_features(self):
expected_feature_list = [
'Size.Area',
'Size.MajorAxisLength',
'Size.MinorAxisLength',
'Size.Perimeter',
'Shape.Circularity',
'Shape.Eccentricity',
'Shape.EquivalentDiameter',
'Shape.Extent',
'Shape.MinorMajorAxisRatio',
'Shape.Solidity',
]
fdata = htk_features.compute_morphometry_features(
cfg.im_nuclei_seg_mask)
check_fdata_sanity(fdata, expected_feature_list)
check_fdata_sanity(cfg.fdata_nuclei,
expected_feature_list,
match_feature_count=False)
if GENERATE_GROUNDTRUTH:
fdata.to_csv(os.path.join(tempfile.gettempdir(),
'Easy1_nuclei_morphometry_features.csv'),
index=False)
fdata_gtruth = pd.read_csv(
utilities.getTestFilePath('Easy1_nuclei_morphometry_features.csv'))
pd.testing.assert_frame_equal(fdata,
fdata_gtruth,
check_less_precise=2)
def test_prep(girderClient): # noqa
cfg.gc = girderClient
cfg.iteminfo = cfg.gc.get('/item', parameters={
'text': "TCGA-A2-A0YE-01Z-00-DX1"})[0]
# read GTCodes dataframe
gtcodePath = getTestFilePath('sample_GTcodes.csv')
cfg.GTcodes = read_csv(gtcodePath)
cfg.GTcodes.index = cfg.GTcodes.loc[:, 'group']
# other params
cfg.get_roi_mask_kwargs = {
'iou_thresh': 0.0,
'crop_to_roi': True,
'use_shapely': True,
'verbose': False
}
cfg.get_contours_kwargs = {
'groups_to_get': None,
'roi_group': 'roi',
'get_roi_contour': True,
'discard_nonenclosed_background': True,
'background_group': 'mostly_stroma',
'MIN_SIZE': 10, 'MAX_SIZE': None,
'verbose': False, 'monitorPrefix': ""
}
# Microns-per-pixel / Magnification (either or)
cfg.MPP = 5.0
cfg.MAG = None
# get annotations for slide
cfg.slide_annotations = cfg.gc.get('/annotation/item/' + cfg.iteminfo['_id'])
# scale up/down annotations by a factor
sf, _ = get_scale_factor_and_appendStr(
gc=cfg.gc, slide_id=cfg.iteminfo['_id'], MPP=cfg.MPP, MAG=cfg.MAG)
cfg.slide_annotations = scale_slide_annotations(cfg.slide_annotations, sf=sf)
# get bounding box information for all annotations
cfg.element_infos = get_bboxes_from_slide_annotations(cfg.slide_annotations)
# params for get_image_and_mask_from_slide()
cfg.get_kwargs = {
'gc': cfg.gc, 'slide_id': cfg.iteminfo['_id'],
'GTCodes_dict': cfg.GTcodes.T.to_dict(),
'bounds': {
'XMIN': 58000, 'XMAX': 63000,
'YMIN': 35000, 'YMAX': 39000},
'MPP': cfg.MPP,
'MAG': cfg.MAG,
'get_roi_mask_kwargs': cfg.get_roi_mask_kwargs,
'get_contours_kwargs': cfg.get_contours_kwargs,
'get_rgb': True,
'get_contours': True,
'get_visualization': True,
}
def test_compute_haralick_features(self):
f = [
'Haralick.ASM',
'Haralick.Contrast',
'Haralick.Correlation',
'Haralick.SumOfSquares',
'Haralick.IDM',
'Haralick.SumAverage',
'Haralick.SumVariance',
'Haralick.SumEntropy',
'Haralick.Entropy',
'Haralick.DifferenceVariance',
'Haralick.DifferenceEntropy',
'Haralick.IMC1',
'Haralick.IMC2',
]
expected_feature_list = []
for col in f:
expected_feature_list.append(col + '.Mean')
expected_feature_list.append(col + '.Range')
fdata = htk_features.compute_haralick_features(cfg.im_nuclei_seg_mask,
cfg.im_nuclei_stain)
check_fdata_sanity(fdata, expected_feature_list)
check_fdata_sanity(cfg.fdata_nuclei,
expected_feature_list,
prefix='Nucleus.',
match_feature_count=False)
check_fdata_sanity(cfg.fdata_nuclei,
expected_feature_list,
prefix='Cytoplasm.',
match_feature_count=False)
if GENERATE_GROUNDTRUTH:
fdata.to_csv(os.path.join(tempfile.gettempdir(),
'Easy1_nuclei_haralick_features.csv'),
index=False)
fdata_gtruth = pd.read_csv(
utilities.getTestFilePath('Easy1_nuclei_haralick_features.csv'))
pd.testing.assert_frame_equal(fdata,
fdata_gtruth,
check_less_precise=2)
def test_compute_intensity_features(self):
expected_feature_list = [
'Intensity.Min',
'Intensity.Max',
'Intensity.Mean',
'Intensity.Median',
'Intensity.MeanMedianDiff',
'Intensity.Std',
'Intensity.IQR',
'Intensity.MAD',
'Intensity.Skewness',
'Intensity.Kurtosis',
'Intensity.HistEnergy',
'Intensity.HistEntropy',
]
fdata = htk_features.compute_intensity_features(
cfg.im_nuclei_seg_mask, cfg.im_nuclei_stain)
check_fdata_sanity(fdata, expected_feature_list)
check_fdata_sanity(cfg.fdata_nuclei,
expected_feature_list,
prefix='Nucleus.',
match_feature_count=False)
check_fdata_sanity(cfg.fdata_nuclei,
expected_feature_list,
prefix='Cytoplasm.',
match_feature_count=False)
if GENERATE_GROUNDTRUTH:
fdata.to_csv(os.path.join(tempfile.gettempdir(),
'Easy1_nuclei_intensity_features.csv'),
index=False)
fdata_gtruth = pd.read_csv(
utilities.getTestFilePath('Easy1_nuclei_intensity_features.csv'),
index_col=None)
pd.testing.assert_frame_equal(fdata,
fdata_gtruth,
check_less_precise=2)
def test_prep(girderClient): # noqa
cfg.gc = girderClient
cfg.iteminfo = cfg.gc.get('/item', parameters={
'text': "TCGA-A2-A0YE-01Z-00-DX1"})[0]
cfg.GTcodes = read_csv(getTestFilePath('saliency_GTcodes.csv'))
def test_prep(girderClient): # noqa
cfg.gc = girderClient
iteminfo = cfg.gc.get('/item', parameters={
'text': "TCGA-A2-A0YE-01Z-00-DX1"})[0]
# read GTCodes dataframe
gtcodePath = getTestFilePath('sample_GTcodes.csv')
GTCodes_dict = read_csv(gtcodePath)
GTCodes_dict.index = GTCodes_dict.loc[:, 'group']
GTCodes_dict = GTCodes_dict.to_dict(orient='index')
# just a temp directory to save masks for now
cfg.BASE_SAVEPATH = tempfile.mkdtemp()
cfg.SAVEPATHS = {
'contours': os.path.join(cfg.BASE_SAVEPATH, 'contours'),
'rgb': os.path.join(cfg.BASE_SAVEPATH, 'rgbs'),
'visualization': os.path.join(cfg.BASE_SAVEPATH, 'vis'),
'mask': os.path.join(cfg.BASE_SAVEPATH, 'masks'),
}
for _, savepath in cfg.SAVEPATHS.items():
if not os.path.exists(savepath):
os.mkdir(savepath)
# Microns-per-pixel / Magnification (either or)
cfg.MPP = 5.0
cfg.MAG = None
# get annotations for slide
cfg.slide_annotations = cfg.gc.get('/annotation/item/' + iteminfo['_id'])
# scale up/down annotations by a factor
sf, _ = get_scale_factor_and_appendStr(
gc=cfg.gc, slide_id=iteminfo['_id'], MPP=cfg.MPP, MAG=cfg.MAG)
cfg.slide_annotations = scale_slide_annotations(cfg.slide_annotations, sf=sf)
# get bounding box information for all annotations
cfg.element_infos = get_bboxes_from_slide_annotations(cfg.slide_annotations)
# common params for annotations_to_contours_no_mask()
annotations_to_contours_kwargs = {
'MPP': cfg.MPP, 'MAG': cfg.MAG,
'linewidth': 0.2,
'get_rgb': True, 'get_visualization': True,
}
# params for TESTING annotations_to_contours_no_mask()
cfg.test_annots_to_contours_kwargs = copy.deepcopy(
annotations_to_contours_kwargs)
cfg.test_annots_to_contours_kwargs.update({
'gc': cfg.gc,
'slide_id': iteminfo['_id'],
'bounds': {
'XMIN': 58000, 'XMAX': 63000,
'YMIN': 35000, 'YMAX': 39000},
})
# params for getting all rois for slide
cfg.get_all_rois_kwargs = {
'gc': cfg.gc,
'slide_id': iteminfo['_id'],
'GTCodes_dict': GTCodes_dict,
'save_directories': cfg.SAVEPATHS,
'annotations_to_contours_kwargs': annotations_to_contours_kwargs,
'slide_name': 'TCGA-A2-A0YE',
'verbose': False,
'monitorprefix': 'test',
}
