def perform_tissue_classification(tiv, vem_iters, beta, scheme='mf',
noise='gauss', freeze_prop=False,
labels=('CSF','GM','WM'), mixmat=None):
"""
perform probabilistic tissue classification on TIV
"""
tiv_ = AffineImage(tiv.get_data(), tiv.get_affine(), 'scanner')
ppm_img_, _ = brain_segmentation(tiv_, beta=beta, niters=vem_iters,
labels=labels, mixmat=mixmat,
noise=noise, freeze_prop=freeze_prop,
scheme=scheme)
count_tiv = len(np.where(tiv.get_data() > 0)[0])
return Nifti1Image(ppm_img_.get_data()[..., 1], tiv.get_affine()), \
Nifti1Image(ppm_img_.get_data()[..., 0], tiv.get_affine()), \
count_tiv
# Labels
labels = []
labels += ['CSF' for k in range(k_csf)]
labels += ['GM' for k in range(k_gm)]
labels += ['WM' for k in range(k_wm)]
mixmat = []
mixmat += [[1, 0, 0] for k in range(k_csf)]
mixmat += [[0, 1, 0] for k in range(k_gm)]
mixmat += [[0, 0, 1] for k in range(k_wm)]
mixmat = np.array(mixmat)
# Perform tissue classification
ppm_img, label_img = brain_segmentation(img,
mask_img=mask_img,
beta=beta, niters=niters,
labels=labels, mixmat=mixmat,
noise=noise, freeze_prop=freeze_prop,
scheme=SCHEME)
outfile = join(mkdtemp(), 'hard_classif.nii')
save_image(label_img, outfile)
print('Label image saved in: %s' % outfile)
# Compute fuzzy Dice indices if a 3-class fuzzy model is provided
if not args.probc == None and \
not args.probg == None and \
not args.probw == None:
print('Computing Dice index')
gpm = [load_image(args.probc),
load_image(args.probg),
load_image(args.probw)]
d = fuzzy_dice(gpm, ppm_img, np.where(mask_img.get_data() > 0))