def __init__(self,
subj_fol,
atlas_name='yeo7_1000',
weight_by='3_2_AxPasi7',
tractography_type='wholebrain_5d_labmask_msmt',
atlas_main_folder=r'C:\Users\Admin\my_scripts\aal\yeo'):
self.folder_name = subj_fol #full path of subj folder
self.tractography_type = tractography_type + '.trk'
self.nii_file_name = load_dwi_files(self.folder_name)[5]
self.streamline_name = self._streamline_name()
self.weight_name = weight_by
self.streamlines = self._load_streamlines()
self.atlas_type = str.lower(atlas_name)
self.atlas_main_folder = atlas_main_folder
self.mask = self._load_gm_mask()
if 'yeo' in self.atlas_type:
self.atlas_file_name = os.path.join(
self.folder_name, 'r' + atlas_name + '_atlas.nii')
self.idx_to_txt = f'{self.atlas_main_folder}{os.sep}{self.atlas_type}{os.sep}index2label.txt'
self.mni_atlas_label = f'{self.atlas_main_folder}{os.sep}{self.atlas_type}{os.sep}{self.atlas_type}_atlas.nii'
elif 'bna' in self.atlas_type:
self.atlas_type = atlas_name
self.atlas_file_name = os.path.join(
self.folder_name, 'rBN_Atlas_274_combined_1mm.nii')
self.idx_to_txt = f'{self.atlas_main_folder}{os.sep}BNA_with_cerebellum.csv'
self.mni_atlas_label = f'{self.atlas_main_folder}{os.sep}BN_Atlas_274_combined_1mm.nii'
def extract_one_bundle(moved, target, full_bund_name, bundle_num, subj,
folder_name, rt, mct):
dipy_home = find_home()
#moved = set_number_of_points(moved, 20)
recognized_bundle, bundle_labels, model = find_bundle(
dipy_home, moved, bundle_num, rt, mct)
nii_file = load_dwi_files(folder_name)[5]
bundle = target[bundle_labels]
if len(bundle) < 20:
model = []
recognized_bundle = []
bundle_labels = []
print(f"Couldn't find {full_bund_name} for {subj}")
else:
keep_s, keep_i = remove_cci_outliers(bundle)
new_s = []
new_s += [bundle[s1] for s1 in keep_i]
save_ft(folder_name,
'',
new_s,
nii_file,
file_name=rf'\{full_bund_name}.trk')
return model, recognized_bundle, bundle_labels
def _load_gm_mask(self):
'''
Load tissue scan (from Fast in preprocessing) to choose GM mask
'''
labels = load_dwi_files(self.folder_name)[3]
mask = (labels == 2)
return mask
def mean_bundle_weight(folder_name, bundle_file_name):
if not os.path.exists(bundle_file_name):
mean_bundle = np.nan
else:
bundle = load_trk(bundle_file_name, "same", bbox_valid_check=False)
bundle = bundle.streamlines
bvec_file = load_dwi_files(folder_name)[6]
mean_vols = weighting_streamlines(folder_name,
bundle,
bvec_file,
weight_by='AxPasi7')
mean_bundle = np.nanmean(mean_vols)
return mean_bundle
out=np.zeros_like(vols_sum),
where=counts != 0)
return vox_vol
if __name__ == '__main__':
main_feo = r'F:\Hila\balance\eo'
main_fec = r'F:\Hila\balance\ec'
subj_folder = glob(os.path.join(main_feo, '*/')) + glob(
os.path.join(main_fec, '*/'))
all_subj_folders = []
for sfol in subj_folder[::]:
for folder_name in glob(os.path.join(sfol, '*/'))[::]:
dir_name = folder_name + 'streamlines'
gtab, data, affine, labels, white_matter, nii_file, bvec_file = load_dwi_files(
folder_name, small_delta=15)
#tract_path = f'{dir_name}{n}_wholebrain_5d_labmask_msmt.trk'
for tfiles in glob(os.path.join(dir_name, '*')):
if 'wholebrain_4d_labmask.trk' in tfiles:
tract_path = os.path.join(dir_name, tfiles)
continue
streamlines = load_ft(tract_path, nii_file)
mean_vol_per_tract = weighting_streamlines(folder_name,
streamlines,
bvec_file,
weight_by='3_2_AxPasi7')
vox_vol = vol_map(streamlines, affine, mean_vol_per_tract,
white_matter.shape)
empty_header = nib.Nifti1Header()
vec_mean = []
vec_median = []
for subj in os.listdir(group):
folder_name = os.path.join(group, subj)
full_bund_name = f'{subj}_{file_bundle_name}'
bundle_file_name = rf'{folder_name}\streamlines\{full_bund_name}.trk'
if not os.path.exists(bundle_file_name):
print('Moving on!')
vec_mean.append(np.nan)
vec_median.append(np.nan)
continue
bundle = load_trk(bundle_file_name, "same", bbox_valid_check=False)
bundle = bundle.streamlines
bvec_file = load_dwi_files(folder_name)[6]
mean_vols = weighting_streamlines(
folder_name,
bundle,
bvec_file,
weight_by='ADD_along_streamlines_WMmasked')
vec_mean.append(np.nanmean(mean_vols))
vec_median.append(np.nanmedian(mean_vols))
val_dict[bundle_name + '_mean_' +
str.split(group, os.sep)[-2]] = vec_mean
val_dict[bundle_name + '_median_' +
str.split(group, os.sep)[-2]] = vec_median
table_vals = pd.DataFrame(val_dict, index=os.listdir(group))
def clean_non_cc(grouping, idx, clean_grouping, clean_streamlines):
for pair, tracts in grouping.items():
if pair[0] == 0 or pair[1] == 0:
continue
else:
if pair[0] in list(np.asarray(idx[0:int(len(idx)/2)])+1) and pair[1] in list(np.asarray(idx[int(len(idx)/2):len(idx)])+1):
clean_streamlines+=tracts
if pair[1] in list(np.asarray(idx[0:int(len(idx)/2)])+1) and pair[0] in list(np.asarray(idx[int(len(idx)/2):len(idx)])+1):
clean_streamlines+=tracts
clean_streamlines, keep_streamlines_idx = remove_cci_outliers(clean_streamlines)
return clean_streamlines
if __name__ == '__main__':
subj = all_subj_folders
names = all_subj_names
fig_type = 'cc_10d_labmask'
weight_by='2_2_AxPasi7'
for s, n in zip(subj[26::], names[26::]):
main_folder = subj_folder + s
gtab, data, affine, labels, white_matter, nii_file, bvec_file = load_dwi_files(main_folder)
labels_headers, idx, m, grouping, h = load_fibers(main_folder, n, index_to_text_file,fig_type,nii_file)
clean_grouping, m_clean, clean_streamlines, m_weighted = create_empty_vars()
weight_by_data, affine = extract_weighted_data(bvec_file,weight_by)
clean_streamlines = clean_non_cc(grouping, idx, clean_grouping, clean_streamlines)
save_ft(main_folder, n, clean_streamlines, nii_file, file_name='_'+ fig_type+'_cleaned.trk')