def region_overlap_report(regionName, atlasBase, atlasTarget):
"""
This function is used to generate a region overlap report. It is used to find which regions a given
area in an atlas (base) correspond to in another atlas (target).
The returned list is sorted according to overlap ratio. Each element is a tuple with the region name, counts and ratio
"""
regionData = atlasBase.regions[atlasBase.ticks.index(regionName)]
atlasBaseImg = loadsave.load_nii(atlasBase.get_volume('1mm')['niifile'])
atlasTargetImg = loadsave.load_nii(
atlasTarget.get_volume('1mm')['niifile'])
atlasBaseData = atlasBaseImg.get_data()
atlasTargetData = atlasTargetImg.get_data()
maskMatrix = atlasTargetData.copy()
maskMatrix[atlasBaseData == regionData] = 1
maskMatrix[atlasBaseData != regionData] = 0
maskedTargetMatrix = np.multiply(atlasTargetData, maskMatrix)
unique, counts = np.unique(maskedTargetMatrix, return_counts=True)
resultList = []
for areaData, count in zip(unique, counts):
if areaData == 0:
continue
descriptionTuple = (
atlasTarget.ticks[atlasTarget.regions.index(areaData)],
'%d/%s' % (count, atlasTarget.get_volume('1mm')['regioncounts'][
atlasTarget.regions.index(areaData)]), float(count) / int(
atlasTarget.get_volume('1mm')['regioncounts'][
atlasTarget.regions.index(areaData)]))
resultList.append(descriptionTuple)
resultList = sorted(resultList,
key=lambda x: int(x[1].split('/')[0]),
reverse=True)
return resultList
def calc_attr(theattr):
img = load_nii('../iso2.0_dtifitresult_{}.nii.gz'.format(theattr))
atlasobj = atlas.getbywd()
atlasname = atlasobj.name
atlasimg = load_nii(os.path.join('nativespace', 'wtemplate_2.nii.gz'))
res = niicalc.calc_region_mean(img, atlasobj, atlasimg)
save_csvmat(os.path.join('nativespace', 'mean{}.csv'.format(theattr)), res)
def __init__(self, atlasobj, volumename, img, outfolder, windowLength = 100, stepSize = 1): self.img = img self.atlasobj = atlasobj self.atlasimg = load_nii(atlasobj.get_volume(volumename)['niifile']) self.outfolder = outfolder self.windowLength = windowLength self.stepSize = stepSize
def build_cmd(self, attr = None, nethold = None, out = None): #attr[1,2,3,4,5,6]
cmd = [self.exeName,"-i",self.corrPath]
if out:
cmd.extend(["-o",out+"\\"])
else:
cmd.extend(["-o",self.outfolder+"\\"])
if nethold:
cmd.extend(["-n",str(nethold)])
else:
cmd.extend(["-n",str(self.argsDict['netthreshold'])])
cmd.extend(["-w",str(self.argsDict['windowLength'])])
cmd.extend(["-t",str(self.argsDict['stepSize'])])
img = load_nii(self.niipath)
cmd.extend(["-l",str(img.header.get_data_shape()[3])])
if attr:
for i in attr:
cmd.append("-c"+str(i))
else:
for k,v in self.argsDict['calc_attr'].items():
if v:
cmd.append("-c"+str(INTER_ATTR_MAPPING[k]))
return cmd
def func(args): subject = args[0] atlasname = args[1] volumename = '3mm' atlasobj = atlas.get(atlasname) outfolder = os.path.join(mmdps_locale.MSA_ProcessedPath, subject, atlasobj.name, 'bold_net', 'dynamic 1 100') img = load_nii(os.path.join(mmdps_locale.MSA_ProcessedPath, subject, 'pBOLD.nii')) os.makedirs(outfolder, exist_ok = True) c = CalcDynamic(atlasobj, volumename, img, outfolder, 100, 1) c.run()
def color_atlas_region(atlasobj,
regions,
colors,
outfilepath,
resolution='1mm'):
"""
This function is used to color one or several regions in an atlas.
An nii file is saved where the specified region is set to 1 and other areas to 0
Input
- regions: a string or a list of strings
- colors: int or a list of int
Note: if both regions and colors are lists, these two must have the same length
"""
if type(regions) is list and type(colors) is list and len(regions) != len(
colors):
print(
'You must specify colors for each regions to label, or specify a single color'
)
raise Exception('Number of colors and regions not match')
atlasImg = loadsave.load_nii(atlasobj.get_volume(resolution)['niifile'])
atlasData = atlasImg.get_data()
newAtlasData = atlasData.copy()
if type(regions) is list and type(colors) is list:
# each color for each region
newAtlasData = np.zeros(newAtlasData.shape)
for region, color in zip(regions, colors):
regionMask = atlasData.copy()
regionMask[regionMask != atlasobj.regions[atlasobj.ticks.index(
region)]] = 0
regionMask[regionMask == atlasobj.regions[atlasobj.ticks.index(
region)]] = color
newAtlasData += regionMask
elif type(regions) is list and type(colors) is int:
# one color for each region
newAtlasData = np.zeros(newAtlasData.shape)
for region in regions:
regionMask = atlasData.copy()
regionMask[regionMask != atlasobj.regions[atlasobj.ticks.index(
region)]] = 0
regionMask[regionMask == atlasobj.regions[atlasobj.ticks.index(
region)]] = colors
newAtlasData += regionMask
elif type(regions) is str and type(colors) is int:
# one color for one region
newAtlasData[newAtlasData != atlasobj.regions[atlasobj.ticks.index(
regions)]] = 0
newAtlasData[newAtlasData == atlasobj.regions[atlasobj.ticks.index(
regions)]] = colors
else:
raise Exception(
'Unsupported combinations. regions as %s, colors as %s.' %
(type(regions), type(colors)))
newAtlasImg = nib.Nifti1Image(newAtlasData, atlasImg.affine,
atlasImg.header)
nib.save(newAtlasImg, outfilepath)
def __init__(self, atlasobj, volumename, img, outfolder, windowLength = 100, stepsize = 3): """ volumename = '3mm' is the name of the atlas volume img is the nii file loaded using load_nii function """ self.img = img self.atlasobj = atlasobj self.atlasimg = load_nii(atlasobj.get_volume(volumename)['niifile']) self.outfolder = outfolder self.stepsize = stepsize self.windowLength = windowLength
def func(args):
subject = args[0]
atlasname = args[1]
windowLength = args[2]
stepsize = args[3]
volumename = '3mm'
atlasobj = atlas.get(atlasname)
work_path = 'D:/Research/xuquan_FMRI/Dynamic_tfMRI_work/'
outfolder = os.path.join(work_path, subject, atlasname, 'bold_net',
'dynamic_%d_%d' % (stepsize, windowLength))
img = load_nii(os.path.join(work_path, subject, 'pBOLD.nii'))
os.makedirs(outfolder, exist_ok=True)
c = CalcDynamic(atlasobj, volumename, img, outfolder, windowLength,
stepsize)
c.run()
def __init__(self, atlasobj, volumename, img, outfolder):
self.img = img
self.atlasobj = atlasobj
self.atlasimg = load_nii(atlasobj.get_volume(volumename)['niifile'])
self.outfolder = outfolder
def gen_timeseries(self):
data = self.img.get_data()
atdata = self.atlasimg.get_data()
timepoints = data.shape[3]
timeseries = np.empty((self.atlasobj.count, timepoints))
for i, region in enumerate(self.atlasobj.regions):
regiondots = data[atdata == region, :]
regionts = np.mean(regiondots, axis=0)
timeseries[i, :] = regionts
return timeseries
def gen_net(self):
ts = self.gen_timeseries()
save_csvmat(self.outpath('timeseries.csv'), ts)
tscorr = np.corrcoef(ts)
save_csvmat(self.outpath('corrcoef.csv'), tscorr)
def run(self):
self.gen_net()
if __name__ == '__main__':
atlasobj = path.curatlas()
volumename = '3mm'
img = load_nii(os.path.join(path.curparent(), 'pBOLD.nii'))
outfolder = 'bold_net'
c = Calc(atlasobj, volumename, img, outfolder)
c.run()
timepoints = data.shape[3]
timeseries = np.empty((atlasobj.count, timepoints))
for i, region in enumerate(atlasobj.regions):
regiondots = data[atdata == region, :]
regionts = np.mean(regiondots, axis=0)
timeseries[i, :] = regionts
return timeseries
def gen_net(self):
ts = self.gen_timeseries()
save_csvmat(self.outpath('timeseries.csv'), ts)
tscorr = np.corrcoef(ts)
save_csvmat(self.outpath('corrcoef.csv'), tscorr)
def run(self):
self.gen_net()
if __name__ == '__main__':
atlasobj = atlas.get('bnatlas')
volumename = '3mm'
for subject in sorted(os.listdir(mmdps_locale.ChanggungAllFullPath)):
print('building subject %s' % subject)
outfolder = os.path.join(mmdps_locale.ChanggungAllFullPath, subject,
'bold_net', 'bnatlas_3')
os.makedirs(outfolder, exist_ok=True)
img = load_nii(
os.path.join(mmdps_locale.ChanggungAllFullPath, subject,
'Filtered_4DVolume.nii'))
c = Calc(atlasobj, volumename, img, outfolder)
c.run()
import os
import t1_niicalc as niicalc
from mmdps.proc import atlas
from mmdps.util.loadsave import load_nii, save_csvmat
if __name__ == '__main__':
img = load_nii('../grey.hdr')
curatlas = atlas.getbywd()
atlasimg = load_nii(curatlas.get_volume('1mm')['niifile'])
res = niicalc.calc_region_mean(img, curatlas, atlasimg)
save_csvmat(os.path.join('t1mean', 'greyvolume.csv'), res)
def get_data(self, resolution=1):
return loadsave.load_nii(
os.path.join(rootconfig.path.atlas, self.name,
'%s_%d.nii' % (self.name, resolution))).get_data()
for i, region in enumerate(atlasobj.regions):
regiondots = data[atdata==region, :]
regionts = np.mean(regiondots, axis=0)
timeseries[i, :] = regionts
return timeseries
def gen_net(self):
ts = self.gen_timeseries()
save_csvmat(self.outpath('timeseries.csv'), ts)
timepoints = ts.shape[1] # number of total timepoints
start = 0
while start + self.windowLength < timepoints:
tscorr = np.corrcoef(ts[:, start:start + self.windowLength])
save_csvmat(self.outpath('corrcoef-%d.%d.csv' % (start, start + self.windowLength)), tscorr)
start += self.stepsize
def run(self):
self.gen_net()
if __name__ == '__main__':
atlasList = ['brodmann_lr', 'brodmann_lrce', 'aicha', 'bnatlas']
volumename = '3mm'
for subject in sorted(os.listdir(mmdps_locale.ChanggungAllFullPath)):
print('building subject %s' % subject)
img = load_nii(os.path.join(mmdps_locale.ChanggungAllFullPath, subject, 'pBOLD.nii'))
for atlasname in atlasList:
atlasobj = atlas.get(atlasname)
outfolder = os.path.join(mmdps_locale.ChanggungAllFullPath, subject, atlasobj.name, 'bold_net')
os.makedirs(outfolder, exist_ok = True)
c = CalcDynamic(atlasobj, volumename, img, outfolder)
c.run()
