Skip to content

Commit

Permalink
working code - same nuclei and features
Browse files Browse the repository at this point in the history
  • Loading branch information
@subinkitware
subinkitware committed Aug 31, 2023
1 parent aaf02c0 commit 6c5c59d
Showing 1 changed file with 99 additions and 112 deletions.
211 changes: 99 additions & 112 deletions histomicstk/segmentation/label/compute_nuclei_features_and_annotations.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -46,8 +46,6 @@ def compute_nuclei_features_and_annotations(im_label,tile_info, im_nuclei=None,

#start the data counter
i = 1
X = []
Y = []
x_start = -1
y_start = -1
max_length = float('inf')
Expand Down Expand Up @@ -88,153 +86,142 @@ def compute_nuclei_features_and_annotations(im_label,tile_info, im_nuclei=None,
bx, by = _remove_thin_colinear_spurs(bx, by,0.01)

if len(bx) > 0:
X.append(bx)
Y.append(by)

#extract features
idata.at[i, 'Label'] = rprops.label
idata.at[i, 'Identifier.Xmin'] = rprops.bbox[1]
idata.at[i, 'Identifier.Ymin'] = rprops.bbox[0]
idata.at[i, 'Identifier.Xmax'] = rprops.bbox[3]
idata.at[i, 'Identifier.Ymax'] = rprops.bbox[2]
idata.at[i, 'Identifier.CentroidX'] = rprops.centroid[1]
idata.at[i, 'Identifier.CentroidY'] = rprops.centroid[0]
if im_nuclei is not None:
# intensity-weighted centroid
wcy, wcx = rprops.weighted_centroid
idata.at[i, 'Identifier.WeightedCentroidX'] = wcx
idata.at[i, 'Identifier.WeightedCentroidY'] = wcy
feature_list.append(idata)

# compute cytoplasm mask
if im_cytoplasm is not None:

cyto_mask = htk_label.dilate_xor(im_label, neigh_width=cyto_width)

cyto_props = regionprops(cyto_mask, intensity_image=im_cytoplasm)
# get boundary points and convert to base pixel space
num_points = len(bx)

# ensure that cytoplasm props order corresponds to the nuclei
lablocs = {v['label']: i for i, v in enumerate(cyto_props)}
cyto_props = [cyto_props[lablocs[v['label']]] if v['label'] in lablocs else None
for v in nuclei_props]
if num_points < 3:
continue

# compute morphometry features
if morphometry_features_flag:
cur_points = np.zeros((num_points, 3))
cur_points[:, 0] = np.round(gx + bx * wfrac, 2)
cur_points[:, 1] = np.round(gy + by * hfrac, 2)
cur_points = cur_points.tolist()

fmorph = compute_morphometry_features(im_label, rprops=nuclei_props)
# create annotation json
cur_annot = {
'type': 'polyline',
'points': cur_points,
'closed': True,
'fillColor': 'rgba(0,0,0,0)',
'lineColor': 'rgb(0,255,0)'
}

feature_list.append(fmorph)
nuclei_annot_list.append(cur_annot)

# compute FSD features
if fsd_features_flag:
#extract features
idata.at[i, 'Label'] = rprops.label
idata.at[i, 'Identifier.Xmin'] = rprops.bbox[1]
idata.at[i, 'Identifier.Ymin'] = rprops.bbox[0]
idata.at[i, 'Identifier.Xmax'] = rprops.bbox[3]
idata.at[i, 'Identifier.Ymax'] = rprops.bbox[2]
idata.at[i, 'Identifier.CentroidX'] = rprops.centroid[1]
idata.at[i, 'Identifier.CentroidY'] = rprops.centroid[0]
if im_nuclei is not None:
# intensity-weighted centroid
wcy, wcx = rprops.weighted_centroid
idata.at[i, 'Identifier.WeightedCentroidX'] = wcx
idata.at[i, 'Identifier.WeightedCentroidY'] = wcy
feature_list.append(idata)

ffsd = compute_fsd_features(im_label, fsd_bnd_pts, fsd_freq_bins,
cyto_width, rprops=nuclei_props)
# # compute cytoplasm mask
# if im_cytoplasm is not None:

feature_list.append(ffsd)
# cyto_mask = htk_label.dilate_xor(im_label, neigh_width=cyto_width)

# compute nuclei intensity features
if intensity_features_flag:
# cyto_props = regionprops(cyto_mask, intensity_image=im_cytoplasm)

fint_nuclei = compute_intensity_features(im_label, im_nuclei,
rprops=nuclei_props)
fint_nuclei.columns = ['Nucleus.' + col
for col in fint_nuclei.columns]
# # ensure that cytoplasm props order corresponds to the nuclei
# lablocs = {v['label']: i for i, v in enumerate(cyto_props)}
# cyto_props = [cyto_props[lablocs[v['label']]] if v['label'] in lablocs else None
# for v in nuclei_props]

feature_list.append(fint_nuclei)
# # compute morphometry features
if morphometry_features_flag:

# compute cytoplasm intensity features
if intensity_features_flag and im_cytoplasm is not None:
fmorph = compute_morphometry_features(im_label, rprops=nuclei_props)

fint_cytoplasm = compute_intensity_features(cyto_mask, im_cytoplasm,
rprops=cyto_props)
fint_cytoplasm.columns = ['Cytoplasm.' + col
for col in fint_cytoplasm.columns]
feature_list.append(fmorph)

feature_list.append(fint_cytoplasm)
# # compute FSD features
# if fsd_features_flag:

# compute nuclei gradient features
if gradient_features_flag:
# ffsd = compute_fsd_features(im_label, fsd_bnd_pts, fsd_freq_bins,
# cyto_width, rprops=nuclei_props)

fgrad_nuclei = compute_gradient_features(im_label, im_nuclei,
rprops=nuclei_props)
fgrad_nuclei.columns = ['Nucleus.' + col
for col in fgrad_nuclei.columns]
# feature_list.append(ffsd)

feature_list.append(fgrad_nuclei)
# # compute nuclei intensity features
# if intensity_features_flag:

# compute cytoplasm gradient features
if gradient_features_flag and im_cytoplasm is not None:
# fint_nuclei = compute_intensity_features(im_label, im_nuclei,
# rprops=nuclei_props)
# fint_nuclei.columns = ['Nucleus.' + col
# for col in fint_nuclei.columns]

fgrad_cytoplasm = compute_gradient_features(cyto_mask, im_cytoplasm,
rprops=cyto_props)
fgrad_cytoplasm.columns = ['Cytoplasm.' + col
for col in fgrad_cytoplasm.columns]
# feature_list.append(fint_nuclei)

feature_list.append(fgrad_cytoplasm)
# # compute cytoplasm intensity features
# if intensity_features_flag and im_cytoplasm is not None:

# compute nuclei haralick features
if haralick_features_flag:
# fint_cytoplasm = compute_intensity_features(cyto_mask, im_cytoplasm,
# rprops=cyto_props)
# fint_cytoplasm.columns = ['Cytoplasm.' + col
# for col in fint_cytoplasm.columns]

fharalick_nuclei = compute_haralick_features(
im_label, im_nuclei,
num_levels=num_glcm_levels,
rprops=nuclei_props
)
# feature_list.append(fint_cytoplasm)

fharalick_nuclei.columns = ['Nucleus.' + col
for col in fharalick_nuclei.columns]
# # compute nuclei gradient features
# if gradient_features_flag:

feature_list.append(fharalick_nuclei)
# fgrad_nuclei = compute_gradient_features(im_label, im_nuclei,
# rprops=nuclei_props)
# fgrad_nuclei.columns = ['Nucleus.' + col
# for col in fgrad_nuclei.columns]

# compute cytoplasm haralick features
if haralick_features_flag and im_cytoplasm is not None:
# feature_list.append(fgrad_nuclei)

fharalick_cytoplasm = compute_haralick_features(
cyto_mask, im_cytoplasm,
num_levels=num_glcm_levels,
rprops=cyto_props
)
# # compute cytoplasm gradient features
# if gradient_features_flag and im_cytoplasm is not None:

fharalick_cytoplasm.columns = ['Cytoplasm.' + col
for col in fharalick_cytoplasm.columns]
# fgrad_cytoplasm = compute_gradient_features(cyto_mask, im_cytoplasm,
# rprops=cyto_props)
# fgrad_cytoplasm.columns = ['Cytoplasm.' + col
# for col in fgrad_cytoplasm.columns]

feature_list.append(fharalick_cytoplasm)
# feature_list.append(fgrad_cytoplasm)

print('>>concat all the feature data')
# # compute nuclei haralick features
# if haralick_features_flag:

# Merge all features
fdata = pd.concat(feature_list, axis=1)
# fharalick_nuclei = compute_haralick_features(
# im_label, im_nuclei,
# num_levels=num_glcm_levels,
# rprops=nuclei_props
# )

print('>>concat all the nuclei annotations')
# fharalick_nuclei.columns = ['Nucleus.' + col
# for col in fharalick_nuclei.columns]

#merge all nuclei segs
for j in range(len(X)):
bx, by = X[j],Y[j]
# feature_list.append(fharalick_nuclei)

for i in range(len(bx)):
# get boundary points and convert to base pixel space
# # compute cytoplasm haralick features
# if haralick_features_flag and im_cytoplasm is not None:

num_points = len(bx)
# fharalick_cytoplasm = compute_haralick_features(
# cyto_mask, im_cytoplasm,
# num_levels=num_glcm_levels,
# rprops=cyto_props
# )

if num_points < 3:
continue
# fharalick_cytoplasm.columns = ['Cytoplasm.' + col
# for col in fharalick_cytoplasm.columns]

cur_points = np.zeros((num_points, 3))
cur_points[:, 0] = np.round(gx + bx[i] * wfrac, 2)
cur_points[:, 1] = np.round(gy + by[i] * hfrac, 2)
cur_points = cur_points.tolist()
# feature_list.append(fharalick_cytoplasm)

# create annotation json
cur_annot = {
'type': 'polyline',
'points': cur_points,
'closed': True,
'fillColor': 'rgba(0,0,0,0)',
'lineColor': 'rgb(0,255,0)'
}
print('>>concat all the feature data')

nuclei_annot_list.append(cur_annot)
# Merge all features
fdata = pd.concat(feature_list, axis=1)

return fdata, nuclei_annot_list

Expand Down

0 comments on commit 6c5c59d

Please sign in to comment.