test_images = list_folders(path_test_cs) # all test cases
cnt=0
path_test = path_test_cs
for case in test_images:
path_timepoints = jp(path_test, case)
list_images = get_dictionary_with_paths_cs([case], path_test, eval(parameters_dict['input_data']))
list_rois = get_dictionary_with_paths_cs([case], path_test, [parameters_dict['brain_mask']])
patch_half = tuple([idx // 2 for idx in eval(parameters_dict['patch_size'])])
timepoints = list_folders(path_timepoints)
# get candidate voxels
all_infer_patches, all_coordenates = get_inference_patches( path_test=path_test,
case = case,
input_data=eval(parameters_dict['input_data']),
roi=parameters_dict['brain_mask'],
patch_shape=eval(parameters_dict['patch_size']),
step=eval(parameters_dict['sampling_step']),
normalize=eval(parameters_dict['normalize']),
mode = "cs"
)
for tp in range(len(timepoints)):
cls()
print("Fold: ", f)
print("Patient", cnt+1, "/", len(test_images))
print("Timepoint ", tp+1)
infer_patches = all_infer_patches[tp]
coordenates = all_coordenates[tp]
#Evaluate all test images
columns = ['Case', 'DSC', 'HD']
df = pd.DataFrame(columns=columns)
test_images = list_folders(path_test)
i_row = 0
cnt = 0
for case in test_images:
print(cnt + 1, "/", len(test_images))
scan_path = jp(path_test, case)
infer_patches, coordenates = get_inference_patches(
scan_path=scan_path,
input_data=options['input_data'],
roi="T1_bet_mask.nii.gz",
patch_shape=options['patch_size'],
step=options['sampling_step'],
normalize=options['normalize'])
batch_size = options['batch_size']
lesion_out = build_image(infer_patches, lesion_model, device, 2, options)
scan_numpy = nib.load(jp(scan_path, "FLAIR_masked.nii.gz")).get_fdata()
all_probs = np.zeros(
(scan_numpy.shape[0], scan_numpy.shape[1], scan_numpy.shape[2], 2))
for i in range(options['num_classes']):
all_probs[:, :, :, i] = reconstruct_image(lesion_out[:,
i], coordenates,