def save_roisubset(trkfile, roislist, roisexcel, labelmask):
#loads trk file, list of rois, the full correspondance of structure => label and the label mask, and saves the
# tracts traversing each region
trkdata = load_trk(trkfile, 'same')
trkdata.to_vox()
if hasattr(trkdata, 'space_attribute'):
header = trkdata.space_attribute
elif hasattr(trkdata, 'space_attributes'):
header = trkdata.space_attributes
trkstreamlines = trkdata.streamlines
import pandas as pd
df = pd.read_excel(roisexcel, sheet_name='Sheet1')
df['Structure'] = df['Structure'].str.lower()
for rois in roislist:
labelslist = []
for roi in rois:
rslt_df = df.loc[df['Structure'] == roi.lower()]
if rois[0].lower() == "wholebrain" or rois[0].lower() == "brain":
labelslist = None
else:
labelslist = np.concatenate((labelslist, np.array(rslt_df.index2)))
print(labelslist)
if isempty(labelslist) and roi.lower() != "wholebrain" and roi.lower() != "brain":
txt = "Warning: Unrecognized roi, will take whole brain as ROI. The roi specified was: " + roi
print(txt)
if isempty(labelslist):
if labelmask is None:
roimask = (fdwi_data[:, :, :, 0] > 0)
else:
roimask = np.where(labelmask == 0, False, True)
else:
if labelmask is None:
raise ("File not found error: labels requested but labels file could not be found at "+dwipath+ " for subject " + subject)
roimask = np.zeros(np.shape(labelmask),dtype=int)
for label in labelslist:
roimask = roimask + (labelmask == label)
trkroipath = trkfile.replace(".trk", "_" + rois + ".trk")
if not os.path.exists(trkroipath):
affinetemp = np.eye(4)
trkroistreamlines = target(trkstreamlines, affinetemp, roimask, include=True, strict="longstring")
trkroistreamlines = Streamlines(trkroistreamlines)
myheader = create_tractogram_header(trkroipath, *header)
roi_sl = lambda: (s for s in trkroistreamlines)
tract_save.save_trk_heavy_duty(trkroipath, streamlines=roi_sl,
affine=header[0], header=myheader)
roistring = roistring #+ "_"
#str_identifier = '_stepsize_' + str(stepsize) + saved_streamlines+ roistring
str_identifier = '_stepsize_' + str(stepsize) + classifiertype + roistring + saved_streamlines
labelslist=[]
if targetrois and (targetrois[0]!="wholebrain" or len(targetrois) > 1):
df = pd.read_excel(atlas_legends, sheet_name='Sheet1')
df['Structure'] = df['Structure'].str.lower()
for roi in targetrois:
rslt_df = df.loc[df['Structure'] == roi.lower()]
if roi.lower() == "wholebrain" or roi.lower() == "brain":
labelslist=None
else:
labelslist=np.concatenate((labelslist, np.array(rslt_df.index)))
print(labelslist)
if isempty(labelslist) and roi.lower() != "wholebrain" and roi.lower() != "brain":
txt = "Warning: Unrecognized roi, will take whole brain as ROI. The roi specified was: " + roi
print(txt)
bvec_orient=[1,2,-3]
tall = time()
tract_results = []
if verbose:
txt=("Process running with % d max processes available on % d subjects with % d subjects in parallel each using % d processes"
% (mp.cpu_count(), np.size(l), subject_processes, function_processes))
print(txt)
send_mail(txt,subject="Main process start msg ")
def save_roisubset(streamlines, roislist, roisexcel, labelmask, stringstep, ratios, trkpath, subject, affine, header):
#atlas_legends = BIGGUS_DISKUS + "/atlases/CHASSSYMM3AtlasLegends.xlsx"
df = pd.read_excel(roisexcel, sheet_name='Sheet1')
df['Structure'] = df['Structure'].str.lower()
for rois in roislist:
if len(rois)==1:
roiname = "_" + rois[0] + "_"
elif len(rois)>1:
roiname="_"
for roi in rois:
roiname = roiname + roi[0:4]
roiname = roiname + "_"
labelslist=[]#fimbria
for roi in rois:
rslt_df = df.loc[df['Structure'] == roi.lower()]
if roi.lower() == "wholebrain" or roi.lower() == "brain":
labelslist=None
else:
labelslist=np.concatenate((labelslist,np.array(rslt_df.index2)))
if isempty(labelslist) and roi.lower() != "wholebrain" and roi.lower() != "brain":
txt = "Warning: Unrecognized roi, will take whole brain as ROI. The roi specified was: " + roi
print(txt)
#bvec_orient=[-2,1,3]
if isempty(labelslist):
roimask = np.where(labelmask == 0, False, True)
else:
if labelmask is None:
raise ("Bad label data, could not define ROI for streams")
roimask = np.zeros(np.shape(labelmask),dtype=int)
for label in labelslist:
roimask = roimask + (labelmask == label)
if not isempty(labelslist):
trkroipath = trkpath + '/' + subject + roiname + "_stepsize_" + stringstep + '.trk'
if not os.path.exists(trkroipath):
affinetemp=np.eye(4)
trkstreamlines = target(streamlines, affinetemp, roimask, include=True, strict="longstring")
trkstreamlines = Streamlines(trkstreamlines)
myheader = create_tractogram_header(trkroipath, *header)
roi_sl = lambda: (s for s in trkstreamlines)
save_trk_heavy_duty(trkroipath, streamlines=roi_sl,
affine=affine, header=myheader)
else:
trkdata = load_trk(trkroipath, 'same')
trkdata.to_vox()
if hasattr(trkdata, 'space_attribute'):
header = trkdata.space_attribute
elif hasattr(trkdata, 'space_attributes'):
header = trkdata.space_attributes
trkstreamlines = trkdata.streamlines
for ratio in ratios:
if ratio != 1:
trkroiminipath = trkpath + '/' + subject + '_ratio_' + ratios + roiname + "_stepsize_" + stringstep + '.trk'
if not os.path.exists(trkroiminipath):
ministream = []
for idx, stream in enumerate(trkstreamlines):
if (idx % ratio) == 0:
ministream.append(stream)
trkstreamlines = ministream
myheader = create_tractogram_header(trkminipath, *header)
ratioed_roi_sl_gen = lambda: (s for s in trkstreamlines)
if allsave:
save_trk_heavy_duty(trkroiminipath, streamlines=ratioed_roi_sl_gen,
affine=affine, header=myheader)
else:
trkdata = load_trk(trkminipath, 'same')
trkdata.to_vox()
if hasattr(trkdata, 'space_attribute'):
header = trkdata.space_attribute
elif hasattr(trkdata, 'space_attributes'):
header = trkdata.space_attributes
trkstreamlines = trkdata.streamlines