def _preprocess_segment(fname_t2, fname_t2_seg, contrast_test, dim_3=False): tmp_folder = sct.TempFolder() tmp_folder_path = tmp_folder.get_path() tmp_folder.chdir() img = Image(fname_t2) gt = Image(fname_t2_seg) fname_t2_RPI, fname_t2_seg_RPI = 'img_RPI.nii.gz', 'seg_RPI.nii.gz' img.change_orientation('RPI').save(fname_t2_RPI) gt.change_orientation('RPI').save(fname_t2_seg_RPI) input_resolution = gt.dim[4:7] del img, gt fname_res, fname_ctr, _ = deepseg_sc.find_centerline(algo='svm', image_fname=fname_t2_RPI, contrast_type=contrast_test, brain_bool=False, folder_output=tmp_folder_path, remove_temp_files=1, centerline_fname=None) fname_t2_seg_RPI_res = 'seg_RPI_res.nii.gz' new_resolution = 'x'.join(['0.5', '0.5', str(input_resolution[2])]) resample_file(fname_t2_seg_RPI, fname_t2_seg_RPI_res, new_resolution, 'mm', 'linear', verbose=0) img, ctr, gt = Image(fname_res), Image(fname_ctr), Image(fname_t2_seg_RPI_res) _, _, _, img = deepseg_sc.crop_image_around_centerline(im_in=img, ctr_in=ctr, crop_size=64) _, _, _, gt = deepseg_sc.crop_image_around_centerline(im_in=gt, ctr_in=ctr, crop_size=64) del ctr img = deepseg_sc.apply_intensity_normalization(im_in=img) if dim_3: # If 3D kernels fname_t2_RPI_res_crop, fname_t2_seg_RPI_res_crop = 'img_RPI_res_crop.nii.gz', 'seg_RPI_res_crop.nii.gz' img.save(fname_t2_RPI_res_crop) gt.save(fname_t2_seg_RPI_res_crop) del img, gt fname_t2_RPI_res_crop_res = 'img_RPI_res_crop_res.nii.gz' fname_t2_seg_RPI_res_crop_res = 'seg_RPI_res_crop_res.nii.gz' resample_file(fname_t2_RPI_res_crop, fname_t2_RPI_res_crop_res, new_resolution, 'mm', 'linear', verbose=0) resample_file(fname_t2_seg_RPI_res_crop, fname_t2_seg_RPI_res_crop_res, new_resolution, 'mm', 'linear', verbose=0) img, gt = Image(fname_t2_RPI_res_crop_res), Image(fname_t2_seg_RPI_res_crop_res) tmp_folder.chdir_undo() tmp_folder.cleanup() return img, gt
def _preprocess_segment(fname_t2, fname_t2_seg, contrast_test, dim_3=False): tmp_folder = sct.TempFolder() tmp_folder_path = tmp_folder.get_path() tmp_folder.chdir() img = Image(fname_t2) gt = Image(fname_t2_seg) fname_t2_RPI, fname_t2_seg_RPI = 'img_RPI.nii.gz', 'seg_RPI.nii.gz' img.change_orientation('RPI').save(fname_t2_RPI) gt.change_orientation('RPI').save(fname_t2_seg_RPI) input_resolution = gt.dim[4:7] del img, gt fname_res, fname_ctr = deepseg_sc.find_centerline(algo='svm', image_fname=fname_t2_RPI, contrast_type=contrast_test, brain_bool=False, folder_output=tmp_folder_path, remove_temp_files=1, centerline_fname=None) fname_t2_seg_RPI_res = 'seg_RPI_res.nii.gz' new_resolution = 'x'.join(['0.5', '0.5', str(input_resolution[2])]) resample_file(fname_t2_seg_RPI, fname_t2_seg_RPI_res, new_resolution, 'mm', 'linear', verbose=0) img, ctr, gt = Image(fname_res), Image(fname_ctr), Image(fname_t2_seg_RPI_res) _, _, _, img = deepseg_sc.crop_image_around_centerline(im_in=img, ctr_in=ctr, crop_size=64) _, _, _, gt = deepseg_sc.crop_image_around_centerline(im_in=gt, ctr_in=ctr, crop_size=64) del ctr img = deepseg_sc.apply_intensity_normalization(im_in=img) if dim_3: # If 3D kernels fname_t2_RPI_res_crop, fname_t2_seg_RPI_res_crop = 'img_RPI_res_crop.nii.gz', 'seg_RPI_res_crop.nii.gz' img.save(fname_t2_RPI_res_crop) gt.save(fname_t2_seg_RPI_res_crop) del img, gt fname_t2_RPI_res_crop_res = 'img_RPI_res_crop_res.nii.gz' fname_t2_seg_RPI_res_crop_res = 'seg_RPI_res_crop_res.nii.gz' resample_file(fname_t2_RPI_res_crop, fname_t2_RPI_res_crop_res, new_resolution, 'mm', 'linear', verbose=0) resample_file(fname_t2_seg_RPI_res_crop, fname_t2_seg_RPI_res_crop_res, new_resolution, 'mm', 'linear', verbose=0) img, gt = Image(fname_t2_RPI_res_crop_res), Image(fname_t2_seg_RPI_res_crop_res) tmp_folder.chdir_undo() tmp_folder.cleanup() return img, gt
def deep_segmentation_spinalcord(im_image, contrast_type, ctr_algo='cnn', ctr_file=None, brain_bool=True, kernel_size='2d', remove_temp_files=1, verbose=1): """Pipeline""" # create temporary folder with intermediate results tmp_folder = sct.TempFolder(verbose=verbose) tmp_folder_path = tmp_folder.get_path() if ctr_algo == 'file': # if the ctr_file is provided tmp_folder.copy_from(ctr_file) file_ctr = os.path.basename(ctr_file) else: file_ctr = None tmp_folder.chdir() # re-orient image to RPI logger.info("Reorient the image to RPI, if necessary...") original_orientation = im_image.orientation fname_orient = 'image_in_RPI.nii' im_image.change_orientation('RPI').save(fname_orient) input_resolution = im_image.dim[4:7] # find the spinal cord centerline - execute OptiC binary logger.info("Finding the spinal cord centerline...") fname_res, centerline_filename = find_centerline( algo=ctr_algo, image_fname=fname_orient, contrast_type=contrast_type, brain_bool=brain_bool, folder_output=tmp_folder_path, remove_temp_files=remove_temp_files, centerline_fname=file_ctr) im_nii, ctr_nii = Image(fname_res), Image(centerline_filename) # crop image around the spinal cord centerline logger.info("Cropping the image around the spinal cord...") crop_size = 96 if (kernel_size == '3d' and contrast_type == 't2s') else 64 X_CROP_LST, Y_CROP_LST, Z_CROP_LST, im_crop_nii = crop_image_around_centerline( im_in=im_nii, ctr_in=ctr_nii, crop_size=crop_size) del ctr_nii # normalize the intensity of the images logger.info("Normalizing the intensity...") im_norm_in = apply_intensity_normalization(im_in=im_crop_nii) del im_crop_nii if kernel_size == '2d': # segment data using 2D convolutions logger.info( "Segmenting the spinal cord using deep learning on 2D patches...") segmentation_model_fname = \ os.path.join(sct.__sct_dir__, 'data', 'deepseg_sc_models', '{}_sc.h5'.format(contrast_type)) seg_crop_data = segment_2d(model_fname=segmentation_model_fname, contrast_type=contrast_type, input_size=(crop_size, crop_size), im_in=im_norm_in) elif kernel_size == '3d': # segment data using 3D convolutions logger.info( "Segmenting the spinal cord using deep learning on 3D patches...") segmentation_model_fname = \ os.path.join(sct.__sct_dir__, 'data', 'deepseg_sc_models', '{}_sc_3D.h5'.format(contrast_type)) seg_crop_data = segment_3d(model_fname=segmentation_model_fname, contrast_type=contrast_type, im_in=im_norm_in) del im_norm_in # reconstruct the segmentation from the crop data logger.info("Reassembling the image...") seg_uncrop_nii = uncrop_image(ref_in=im_nii, data_crop=seg_crop_data, x_crop_lst=X_CROP_LST, y_crop_lst=Y_CROP_LST, z_crop_lst=Z_CROP_LST) fname_res_seg = sct.add_suffix(fname_res, '_seg') seg_uncrop_nii.save(fname_res_seg) del seg_crop_data # resample to initial resolution logger.info( "Resampling the segmentation to the original image resolution...") initial_resolution = 'x'.join([ str(input_resolution[0]), str(input_resolution[1]), str(input_resolution[2]) ]) fname_res_seg_downsamp = sct.add_suffix(fname_res_seg, '_downsamp') resampling.resample_file(fname_res_seg, fname_res_seg_downsamp, initial_resolution, 'mm', 'linear', verbose=0) im_image_res_seg_downsamp = Image(fname_res_seg_downsamp) if ctr_algo == 'viewer': # resample and reorient the viewer labels fname_res_labels = sct.add_suffix(fname_orient, '_labels-centerline') resampling.resample_file(fname_res_labels, fname_res_labels, initial_resolution, 'mm', 'linear', verbose=0) im_image_res_labels_downsamp = Image( fname_res_labels).change_orientation(original_orientation) else: im_image_res_labels_downsamp = None if verbose == 2: fname_res_ctr = sct.add_suffix(fname_orient, '_ctr') resampling.resample_file(fname_res_ctr, fname_res_ctr, initial_resolution, 'mm', 'linear', verbose=0) im_image_res_ctr_downsamp = Image(fname_res_ctr).change_orientation( original_orientation) else: im_image_res_ctr_downsamp = None # binarize the resampled image to remove interpolation effects logger.info( "Binarizing the segmentation to avoid interpolation effects...") thr = 0.0001 if contrast_type in ['t1', 'dwi'] else 0.5 # TODO: optimize speed --> np.where is slow im_image_res_seg_downsamp.data[np.where( im_image_res_seg_downsamp.data >= thr)] = 1 im_image_res_seg_downsamp.data[np.where( im_image_res_seg_downsamp.data < thr)] = 0 # post processing step to z_regularized im_image_res_seg_downsamp_postproc = post_processing_volume_wise( im_in=im_image_res_seg_downsamp) tmp_folder.chdir_undo() # remove temporary files if remove_temp_files: logger.info("Remove temporary files...") tmp_folder.cleanup() # reorient to initial orientation return im_image_res_seg_downsamp_postproc.change_orientation(original_orientation), \ im_nii, \ seg_uncrop_nii.change_orientation('RPI'), \ im_image_res_labels_downsamp, \ im_image_res_ctr_downsamp
def find_centerline(algo, image_fname, contrast_type, brain_bool, folder_output, remove_temp_files, centerline_fname): """ Assumes RPI orientation :param algo: :param image_fname: :param contrast_type: :param brain_bool: :param folder_output: :param remove_temp_files: :param centerline_fname: :return: """ # TODO: remove unnecessary i/o if Image(image_fname).dim[2] == 1: # isct_spine_detect requires nz > 1 im_concat = concat_data([image_fname, image_fname], dim=2) im_concat.save(sct.add_suffix(image_fname, '_concat')) image_fname = sct.add_suffix(image_fname, '_concat') bool_2d = True else: bool_2d = False # TODO: maybe change 'svm' for 'optic', because this is how we call it in sct_get_centerline if algo == 'svm': # run optic on a heatmap computed by a trained SVM+HoG algorithm # optic_models_fname = os.path.join(path_sct, 'data', 'optic_models', '{}_model'.format(contrast_type)) # # TODO: replace with get_centerline(method=optic) img_ctl, arr_ctl, _ = get_centerline(Image(image_fname), algo_fitting='optic', contrast=contrast_type) centerline_filename = sct.add_suffix(image_fname, "_ctr") img_ctl.save(centerline_filename) elif algo == 'cnn': # CNN parameters dct_patch_ctr = { 't2': { 'size': (80, 80), 'mean': 51.1417, 'std': 57.4408 }, 't2s': { 'size': (80, 80), 'mean': 68.8591, 'std': 71.4659 }, 't1': { 'size': (80, 80), 'mean': 55.7359, 'std': 64.3149 }, 'dwi': { 'size': (80, 80), 'mean': 55.744, 'std': 45.003 } } dct_params_ctr = { 't2': { 'features': 16, 'dilation_layers': 2 }, 't2s': { 'features': 8, 'dilation_layers': 3 }, 't1': { 'features': 24, 'dilation_layers': 3 }, 'dwi': { 'features': 8, 'dilation_layers': 2 } } # load model ctr_model_fname = os.path.join(sct.__sct_dir__, 'data', 'deepseg_sc_models', '{}_ctr.h5'.format(contrast_type)) ctr_model = nn_architecture_ctr( height=dct_patch_ctr[contrast_type]['size'][0], width=dct_patch_ctr[contrast_type]['size'][1], channels=1, classes=1, features=dct_params_ctr[contrast_type]['features'], depth=2, temperature=1.0, padding='same', batchnorm=True, dropout=0.0, dilation_layers=dct_params_ctr[contrast_type]['dilation_layers']) ctr_model.load_weights(ctr_model_fname) logger.info("Resample the image to 0.5x0.5 mm in-plane resolution...") fname_res = sct.add_suffix(image_fname, '_resampled') input_resolution = Image(image_fname).dim[4:7] new_resolution = 'x'.join(['0.5', '0.5', str(input_resolution[2])]) resampling.resample_file(image_fname, fname_res, new_resolution, 'mm', 'linear', verbose=0) # compute the heatmap fname_heatmap = sct.add_suffix(image_fname, "_heatmap") img_filename = ''.join(sct.extract_fname(fname_heatmap)[:2]) fname_heatmap_nii = img_filename + '.nii' z_max = heatmap(filename_in=fname_res, filename_out=fname_heatmap_nii, model=ctr_model, patch_shape=dct_patch_ctr[contrast_type]['size'], mean_train=dct_patch_ctr[contrast_type]['mean'], std_train=dct_patch_ctr[contrast_type]['std'], brain_bool=brain_bool) # run optic on the heatmap centerline_filename = sct.add_suffix(fname_heatmap, "_ctr") heatmap2optic(fname_heatmap=fname_heatmap_nii, lambda_value=7 if contrast_type == 't2s' else 1, fname_out=centerline_filename, z_max=z_max if brain_bool else None) elif algo == 'viewer': im_labels = _call_viewer_centerline(Image(image_fname)) im_centerline, arr_centerline, _ = get_centerline(im_labels) centerline_filename = sct.add_suffix(image_fname, "_ctr") labels_filename = sct.add_suffix(image_fname, "_labels-centerline") im_centerline.save(centerline_filename) im_labels.save(labels_filename) elif algo == 'file': centerline_filename = sct.add_suffix(image_fname, "_ctr") # Re-orient the manual centerline Image(centerline_fname).change_orientation('RPI').save( centerline_filename) else: logger.error( 'The parameter "-centerline" is incorrect. Please try again.') sys.exit(1) if algo != 'cnn': logger.info("Resample the image to 0.5x0.5 mm in-plane resolution...") fname_res = sct.add_suffix(image_fname, '_resampled') input_resolution = Image(image_fname).dim[4:7] new_resolution = 'x'.join(['0.5', '0.5', str(input_resolution[2])]) resampling.resample_file(image_fname, fname_res, new_resolution, 'mm', 'linear', verbose=0) resampling.resample_file(centerline_filename, centerline_filename, new_resolution, 'mm', 'linear', verbose=0) if bool_2d: im_split_lst = split_data(Image(centerline_filename), dim=2) im_split_lst[0].save(centerline_filename) return fname_res, centerline_filename
def main(args=None): # initializations param = Param() # check user arguments if not args: args = sys.argv[1:] # Get parser info parser = get_parser() arguments = parser.parse(args) fname_data = arguments['-i'] fname_seg = arguments['-s'] if '-l' in arguments: fname_landmarks = arguments['-l'] label_type = 'body' elif '-ldisc' in arguments: fname_landmarks = arguments['-ldisc'] label_type = 'disc' else: sct.printv('ERROR: Labels should be provided.', 1, 'error') if '-ofolder' in arguments: path_output = arguments['-ofolder'] else: path_output = '' param.path_qc = arguments.get("-qc", None) path_template = arguments['-t'] contrast_template = arguments['-c'] ref = arguments['-ref'] param.remove_temp_files = int(arguments.get('-r')) verbose = int(arguments.get('-v')) sct.init_sct(log_level=verbose, update=True) # Update log level param.verbose = verbose # TODO: not clean, unify verbose or param.verbose in code, but not both param_centerline = ParamCenterline( algo_fitting=arguments['-centerline-algo'], smooth=arguments['-centerline-smooth']) # registration parameters if '-param' in arguments: # reset parameters but keep step=0 (might be overwritten if user specified step=0) paramreg = ParamregMultiStep([step0]) if ref == 'subject': paramreg.steps['0'].dof = 'Tx_Ty_Tz_Rx_Ry_Rz_Sz' # add user parameters for paramStep in arguments['-param']: paramreg.addStep(paramStep) else: paramreg = ParamregMultiStep([step0, step1, step2]) # if ref=subject, initialize registration using different affine parameters if ref == 'subject': paramreg.steps['0'].dof = 'Tx_Ty_Tz_Rx_Ry_Rz_Sz' # initialize other parameters zsubsample = param.zsubsample # retrieve template file names file_template_vertebral_labeling = get_file_label(os.path.join(path_template, 'template'), 'vertebral labeling') file_template = get_file_label(os.path.join(path_template, 'template'), contrast_template.upper() + '-weighted template') file_template_seg = get_file_label(os.path.join(path_template, 'template'), 'spinal cord') # start timer start_time = time.time() # get fname of the template + template objects fname_template = os.path.join(path_template, 'template', file_template) fname_template_vertebral_labeling = os.path.join(path_template, 'template', file_template_vertebral_labeling) fname_template_seg = os.path.join(path_template, 'template', file_template_seg) fname_template_disc_labeling = os.path.join(path_template, 'template', 'PAM50_label_disc.nii.gz') # check file existence # TODO: no need to do that! sct.printv('\nCheck template files...') sct.check_file_exist(fname_template, verbose) sct.check_file_exist(fname_template_vertebral_labeling, verbose) sct.check_file_exist(fname_template_seg, verbose) path_data, file_data, ext_data = sct.extract_fname(fname_data) # sct.printv(arguments) sct.printv('\nCheck parameters:', verbose) sct.printv(' Data: ' + fname_data, verbose) sct.printv(' Landmarks: ' + fname_landmarks, verbose) sct.printv(' Segmentation: ' + fname_seg, verbose) sct.printv(' Path template: ' + path_template, verbose) sct.printv(' Remove temp files: ' + str(param.remove_temp_files), verbose) # check input labels labels = check_labels(fname_landmarks, label_type=label_type) vertebral_alignment = False if len(labels) > 2 and label_type == 'disc': vertebral_alignment = True path_tmp = sct.tmp_create(basename="register_to_template", verbose=verbose) # set temporary file names ftmp_data = 'data.nii' ftmp_seg = 'seg.nii.gz' ftmp_label = 'label.nii.gz' ftmp_template = 'template.nii' ftmp_template_seg = 'template_seg.nii.gz' ftmp_template_label = 'template_label.nii.gz' # copy files to temporary folder sct.printv('\nCopying input data to tmp folder and convert to nii...', verbose) Image(fname_data).save(os.path.join(path_tmp, ftmp_data)) Image(fname_seg).save(os.path.join(path_tmp, ftmp_seg)) Image(fname_landmarks).save(os.path.join(path_tmp, ftmp_label)) Image(fname_template).save(os.path.join(path_tmp, ftmp_template)) Image(fname_template_seg).save(os.path.join(path_tmp, ftmp_template_seg)) Image(fname_template_vertebral_labeling).save(os.path.join(path_tmp, ftmp_template_label)) if label_type == 'disc': Image(fname_template_disc_labeling).save(os.path.join(path_tmp, ftmp_template_label)) # go to tmp folder curdir = os.getcwd() os.chdir(path_tmp) # Generate labels from template vertebral labeling if label_type == 'body': sct.printv('\nGenerate labels from template vertebral labeling', verbose) ftmp_template_label_, ftmp_template_label = ftmp_template_label, sct.add_suffix(ftmp_template_label, "_body") sct_label_utils.main(args=['-i', ftmp_template_label_, '-vert-body', '0', '-o', ftmp_template_label]) # check if provided labels are available in the template sct.printv('\nCheck if provided labels are available in the template', verbose) image_label_template = Image(ftmp_template_label) labels_template = image_label_template.getNonZeroCoordinates(sorting='value') if labels[-1].value > labels_template[-1].value: sct.printv('ERROR: Wrong landmarks input. Labels must have correspondence in template space. \nLabel max ' 'provided: ' + str(labels[-1].value) + '\nLabel max from template: ' + str(labels_template[-1].value), verbose, 'error') # if only one label is present, force affine transformation to be Tx,Ty,Tz only (no scaling) if len(labels) == 1: paramreg.steps['0'].dof = 'Tx_Ty_Tz' sct.printv('WARNING: Only one label is present. Forcing initial transformation to: ' + paramreg.steps['0'].dof, 1, 'warning') # Project labels onto the spinal cord centerline because later, an affine transformation is estimated between the # template's labels (centered in the cord) and the subject's labels (assumed to be centered in the cord). # If labels are not centered, mis-registration errors are observed (see issue #1826) ftmp_label = project_labels_on_spinalcord(ftmp_label, ftmp_seg, param_centerline) # binarize segmentation (in case it has values below 0 caused by manual editing) sct.printv('\nBinarize segmentation', verbose) ftmp_seg_, ftmp_seg = ftmp_seg, sct.add_suffix(ftmp_seg, "_bin") sct_maths.main(['-i', ftmp_seg_, '-bin', '0.5', '-o', ftmp_seg]) # Switch between modes: subject->template or template->subject if ref == 'template': # resample data to 1mm isotropic sct.printv('\nResample data to 1mm isotropic...', verbose) resample_file(ftmp_data, add_suffix(ftmp_data, '_1mm'), '1.0x1.0x1.0', 'mm', 'linear', verbose) ftmp_data = add_suffix(ftmp_data, '_1mm') resample_file(ftmp_seg, add_suffix(ftmp_seg, '_1mm'), '1.0x1.0x1.0', 'mm', 'linear', verbose) ftmp_seg = add_suffix(ftmp_seg, '_1mm') # N.B. resampling of labels is more complicated, because they are single-point labels, therefore resampling # with nearest neighbour can make them disappear. resample_labels(ftmp_label, ftmp_data, add_suffix(ftmp_label, '_1mm')) ftmp_label = add_suffix(ftmp_label, '_1mm') # Change orientation of input images to RPI sct.printv('\nChange orientation of input images to RPI...', verbose) ftmp_data = Image(ftmp_data).change_orientation("RPI", generate_path=True).save().absolutepath ftmp_seg = Image(ftmp_seg).change_orientation("RPI", generate_path=True).save().absolutepath ftmp_label = Image(ftmp_label).change_orientation("RPI", generate_path=True).save().absolutepath ftmp_seg_, ftmp_seg = ftmp_seg, add_suffix(ftmp_seg, '_crop') if vertebral_alignment: # cropping the segmentation based on the label coverage to ensure good registration with vertebral alignment # See https://github.com/neuropoly/spinalcordtoolbox/pull/1669 for details image_labels = Image(ftmp_label) coordinates_labels = image_labels.getNonZeroCoordinates(sorting='z') nx, ny, nz, nt, px, py, pz, pt = image_labels.dim offset_crop = 10.0 * pz # cropping the image 10 mm above and below the highest and lowest label cropping_slices = [coordinates_labels[0].z - offset_crop, coordinates_labels[-1].z + offset_crop] # make sure that the cropping slices do not extend outside of the slice range (issue #1811) if cropping_slices[0] < 0: cropping_slices[0] = 0 if cropping_slices[1] > nz: cropping_slices[1] = nz msct_image.spatial_crop(Image(ftmp_seg_), dict(((2, np.int32(np.round(cropping_slices))),))).save(ftmp_seg) else: # if we do not align the vertebral levels, we crop the segmentation from top to bottom im_seg_rpi = Image(ftmp_seg_) bottom = 0 for data in msct_image.SlicerOneAxis(im_seg_rpi, "IS"): if (data != 0).any(): break bottom += 1 top = im_seg_rpi.data.shape[2] for data in msct_image.SlicerOneAxis(im_seg_rpi, "SI"): if (data != 0).any(): break top -= 1 msct_image.spatial_crop(im_seg_rpi, dict(((2, (bottom, top)),))).save(ftmp_seg) # straighten segmentation sct.printv('\nStraighten the spinal cord using centerline/segmentation...', verbose) # check if warp_curve2straight and warp_straight2curve already exist (i.e. no need to do it another time) fn_warp_curve2straight = os.path.join(curdir, "warp_curve2straight.nii.gz") fn_warp_straight2curve = os.path.join(curdir, "warp_straight2curve.nii.gz") fn_straight_ref = os.path.join(curdir, "straight_ref.nii.gz") cache_input_files=[ftmp_seg] if vertebral_alignment: cache_input_files += [ ftmp_template_seg, ftmp_label, ftmp_template_label, ] cache_sig = sct.cache_signature( input_files=cache_input_files, ) cachefile = os.path.join(curdir, "straightening.cache") if sct.cache_valid(cachefile, cache_sig) and os.path.isfile(fn_warp_curve2straight) and os.path.isfile(fn_warp_straight2curve) and os.path.isfile(fn_straight_ref): sct.printv('Reusing existing warping field which seems to be valid', verbose, 'warning') sct.copy(fn_warp_curve2straight, 'warp_curve2straight.nii.gz') sct.copy(fn_warp_straight2curve, 'warp_straight2curve.nii.gz') sct.copy(fn_straight_ref, 'straight_ref.nii.gz') # apply straightening sct_apply_transfo.main(args=[ '-i', ftmp_seg, '-w', 'warp_curve2straight.nii.gz', '-d', 'straight_ref.nii.gz', '-o', add_suffix(ftmp_seg, '_straight')]) else: from spinalcordtoolbox.straightening import SpinalCordStraightener sc_straight = SpinalCordStraightener(ftmp_seg, ftmp_seg) sc_straight.param_centerline = param_centerline sc_straight.output_filename = add_suffix(ftmp_seg, '_straight') sc_straight.path_output = './' sc_straight.qc = '0' sc_straight.remove_temp_files = param.remove_temp_files sc_straight.verbose = verbose if vertebral_alignment: sc_straight.centerline_reference_filename = ftmp_template_seg sc_straight.use_straight_reference = True sc_straight.discs_input_filename = ftmp_label sc_straight.discs_ref_filename = ftmp_template_label sc_straight.straighten() sct.cache_save(cachefile, cache_sig) # N.B. DO NOT UPDATE VARIABLE ftmp_seg BECAUSE TEMPORARY USED LATER # re-define warping field using non-cropped space (to avoid issue #367) sct_concat_transfo.main(args=[ '-w', 'warp_straight2curve.nii.gz', '-d', ftmp_data, '-o', 'warp_straight2curve.nii.gz']) if vertebral_alignment: sct.copy('warp_curve2straight.nii.gz', 'warp_curve2straightAffine.nii.gz') else: # Label preparation: # -------------------------------------------------------------------------------- # Remove unused label on template. Keep only label present in the input label image sct.printv('\nRemove unused label on template. Keep only label present in the input label image...', verbose) sct.run(['sct_label_utils', '-i', ftmp_template_label, '-o', ftmp_template_label, '-remove-reference', ftmp_label]) # Dilating the input label so they can be straighten without losing them sct.printv('\nDilating input labels using 3vox ball radius') sct_maths.main(['-i', ftmp_label, '-dilate', '3', '-o', add_suffix(ftmp_label, '_dilate')]) ftmp_label = add_suffix(ftmp_label, '_dilate') # Apply straightening to labels sct.printv('\nApply straightening to labels...', verbose) sct_apply_transfo.main(args=[ '-i', ftmp_label, '-o', add_suffix(ftmp_label, '_straight'), '-d', add_suffix(ftmp_seg, '_straight'), '-w', 'warp_curve2straight.nii.gz', '-x', 'nn']) ftmp_label = add_suffix(ftmp_label, '_straight') # Compute rigid transformation straight landmarks --> template landmarks sct.printv('\nEstimate transformation for step #0...', verbose) try: register_landmarks(ftmp_label, ftmp_template_label, paramreg.steps['0'].dof, fname_affine='straight2templateAffine.txt', verbose=verbose) except RuntimeError: raise('Input labels do not seem to be at the right place. Please check the position of the labels. ' 'See documentation for more details: https://www.slideshare.net/neuropoly/sct-course-20190121/42') # Concatenate transformations: curve --> straight --> affine sct.printv('\nConcatenate transformations: curve --> straight --> affine...', verbose) sct_concat_transfo.main(args=[ '-w', ['warp_curve2straight.nii.gz', 'straight2templateAffine.txt'], '-d', 'template.nii', '-o', 'warp_curve2straightAffine.nii.gz']) # Apply transformation sct.printv('\nApply transformation...', verbose) sct_apply_transfo.main(args=[ '-i', ftmp_data, '-o', add_suffix(ftmp_data, '_straightAffine'), '-d', ftmp_template, '-w', 'warp_curve2straightAffine.nii.gz']) ftmp_data = add_suffix(ftmp_data, '_straightAffine') sct_apply_transfo.main(args=[ '-i', ftmp_seg, '-o', add_suffix(ftmp_seg, '_straightAffine'), '-d', ftmp_template, '-w', 'warp_curve2straightAffine.nii.gz', '-x', 'linear']) ftmp_seg = add_suffix(ftmp_seg, '_straightAffine') """ # Benjamin: Issue from Allan Martin, about the z=0 slice that is screwed up, caused by the affine transform. # Solution found: remove slices below and above landmarks to avoid rotation effects points_straight = [] for coord in landmark_template: points_straight.append(coord.z) min_point, max_point = int(np.round(np.min(points_straight))), int(np.round(np.max(points_straight))) ftmp_seg_, ftmp_seg = ftmp_seg, add_suffix(ftmp_seg, '_black') msct_image.spatial_crop(Image(ftmp_seg_), dict(((2, (min_point,max_point)),))).save(ftmp_seg) """ # open segmentation im = Image(ftmp_seg) im_new = msct_image.empty_like(im) # binarize im_new.data = im.data > 0.5 # find min-max of anat2template (for subsequent cropping) zmin_template, zmax_template = msct_image.find_zmin_zmax(im_new, threshold=0.5) # save binarized segmentation im_new.save(add_suffix(ftmp_seg, '_bin')) # unused? # crop template in z-direction (for faster processing) # TODO: refactor to use python module instead of doing i/o sct.printv('\nCrop data in template space (for faster processing)...', verbose) ftmp_template_, ftmp_template = ftmp_template, add_suffix(ftmp_template, '_crop') msct_image.spatial_crop(Image(ftmp_template_), dict(((2, (zmin_template,zmax_template)),))).save(ftmp_template) ftmp_template_seg_, ftmp_template_seg = ftmp_template_seg, add_suffix(ftmp_template_seg, '_crop') msct_image.spatial_crop(Image(ftmp_template_seg_), dict(((2, (zmin_template,zmax_template)),))).save(ftmp_template_seg) ftmp_data_, ftmp_data = ftmp_data, add_suffix(ftmp_data, '_crop') msct_image.spatial_crop(Image(ftmp_data_), dict(((2, (zmin_template,zmax_template)),))).save(ftmp_data) ftmp_seg_, ftmp_seg = ftmp_seg, add_suffix(ftmp_seg, '_crop') msct_image.spatial_crop(Image(ftmp_seg_), dict(((2, (zmin_template,zmax_template)),))).save(ftmp_seg) # sub-sample in z-direction # TODO: refactor to use python module instead of doing i/o sct.printv('\nSub-sample in z-direction (for faster processing)...', verbose) sct.run(['sct_resample', '-i', ftmp_template, '-o', add_suffix(ftmp_template, '_sub'), '-f', '1x1x' + zsubsample], verbose) ftmp_template = add_suffix(ftmp_template, '_sub') sct.run(['sct_resample', '-i', ftmp_template_seg, '-o', add_suffix(ftmp_template_seg, '_sub'), '-f', '1x1x' + zsubsample], verbose) ftmp_template_seg = add_suffix(ftmp_template_seg, '_sub') sct.run(['sct_resample', '-i', ftmp_data, '-o', add_suffix(ftmp_data, '_sub'), '-f', '1x1x' + zsubsample], verbose) ftmp_data = add_suffix(ftmp_data, '_sub') sct.run(['sct_resample', '-i', ftmp_seg, '-o', add_suffix(ftmp_seg, '_sub'), '-f', '1x1x' + zsubsample], verbose) ftmp_seg = add_suffix(ftmp_seg, '_sub') # Registration straight spinal cord to template sct.printv('\nRegister straight spinal cord to template...', verbose) # loop across registration steps warp_forward = [] warp_inverse = [] for i_step in range(1, len(paramreg.steps)): sct.printv('\nEstimate transformation for step #' + str(i_step) + '...', verbose) # identify which is the src and dest if paramreg.steps[str(i_step)].type == 'im': src = ftmp_data dest = ftmp_template interp_step = 'linear' elif paramreg.steps[str(i_step)].type == 'seg': src = ftmp_seg dest = ftmp_template_seg interp_step = 'nn' else: sct.printv('ERROR: Wrong image type.', 1, 'error') if paramreg.steps[str(i_step)].algo == 'centermassrot' and paramreg.steps[str(i_step)].rot_method == 'hog': src_seg = ftmp_seg dest_seg = ftmp_template_seg # if step>1, apply warp_forward_concat to the src image to be used if i_step > 1: # apply transformation from previous step, to use as new src for registration sct_apply_transfo.main(args=[ '-i', src, '-d', dest, '-w', warp_forward, '-o', add_suffix(src, '_regStep' + str(i_step - 1)), '-x', interp_step]) src = add_suffix(src, '_regStep' + str(i_step - 1)) if paramreg.steps[str(i_step)].algo == 'centermassrot' and paramreg.steps[str(i_step)].rot_method == 'hog': # also apply transformation to the seg sct_apply_transfo.main(args=[ '-i', src_seg, '-d', dest_seg, '-w', warp_forward, '-o', add_suffix(src, '_regStep' + str(i_step - 1)), '-x', interp_step]) src_seg = add_suffix(src_seg, '_regStep' + str(i_step - 1)) # register src --> dest # TODO: display param for debugging if paramreg.steps[str(i_step)].algo == 'centermassrot' and paramreg.steps[str(i_step)].rot_method == 'hog': # im_seg case warp_forward_out, warp_inverse_out = register([src, src_seg], [dest, dest_seg], paramreg, param, str(i_step)) else: warp_forward_out, warp_inverse_out = register(src, dest, paramreg, param, str(i_step)) warp_forward.append(warp_forward_out) warp_inverse.append(warp_inverse_out) # Concatenate transformations: anat --> template sct.printv('\nConcatenate transformations: anat --> template...', verbose) warp_forward.insert(0, 'warp_curve2straightAffine.nii.gz') sct_concat_transfo.main(args=[ '-w', warp_forward, '-d', 'template.nii', '-o', 'warp_anat2template.nii.gz']) # Concatenate transformations: template --> anat sct.printv('\nConcatenate transformations: template --> anat...', verbose) warp_inverse.reverse() if vertebral_alignment: warp_inverse.append('warp_straight2curve.nii.gz') sct_concat_transfo.main(args=[ '-w', warp_inverse, '-d', 'data.nii', '-o', 'warp_template2anat.nii.gz']) else: warp_inverse.append('straight2templateAffine.txt') warp_inverse.append('warp_straight2curve.nii.gz') sct_concat_transfo.main(args=[ '-w', warp_inverse, '-winv', ['straight2templateAffine.txt'], '-d', 'data.nii', '-o', 'warp_template2anat.nii.gz']) # register template->subject elif ref == 'subject': # Change orientation of input images to RPI sct.printv('\nChange orientation of input images to RPI...', verbose) ftmp_data = Image(ftmp_data).change_orientation("RPI", generate_path=True).save().absolutepath ftmp_seg = Image(ftmp_seg).change_orientation("RPI", generate_path=True).save().absolutepath ftmp_label = Image(ftmp_label).change_orientation("RPI", generate_path=True).save().absolutepath # Remove unused label on template. Keep only label present in the input label image sct.printv('\nRemove unused label on template. Keep only label present in the input label image...', verbose) sct.run(['sct_label_utils', '-i', ftmp_template_label, '-o', ftmp_template_label, '-remove-reference', ftmp_label]) # Add one label because at least 3 orthogonal labels are required to estimate an affine transformation. This # new label is added at the level of the upper most label (lowest value), at 1cm to the right. for i_file in [ftmp_label, ftmp_template_label]: im_label = Image(i_file) coord_label = im_label.getCoordinatesAveragedByValue() # N.B. landmarks are sorted by value # Create new label from copy import deepcopy new_label = deepcopy(coord_label[0]) # move it 5mm to the left (orientation is RAS) nx, ny, nz, nt, px, py, pz, pt = im_label.dim new_label.x = np.round(coord_label[0].x + 5.0 / px) # assign value 99 new_label.value = 99 # Add to existing image im_label.data[int(new_label.x), int(new_label.y), int(new_label.z)] = new_label.value # Overwrite label file # im_label.absolutepath = 'label_rpi_modif.nii.gz' im_label.save() # Bring template to subject space using landmark-based transformation sct.printv('\nEstimate transformation for step #0...', verbose) warp_forward = ['template2subjectAffine.txt'] warp_inverse = ['template2subjectAffine.txt'] try: register_landmarks(ftmp_template_label, ftmp_label, paramreg.steps['0'].dof, fname_affine=warp_forward[0], verbose=verbose, path_qc="./") except Exception: sct.printv('ERROR: input labels do not seem to be at the right place. Please check the position of the labels. See documentation for more details: https://www.slideshare.net/neuropoly/sct-course-20190121/42', verbose=verbose, type='error') # loop across registration steps for i_step in range(1, len(paramreg.steps)): sct.printv('\nEstimate transformation for step #' + str(i_step) + '...', verbose) # identify which is the src and dest if paramreg.steps[str(i_step)].type == 'im': src = ftmp_template dest = ftmp_data interp_step = 'linear' elif paramreg.steps[str(i_step)].type == 'seg': src = ftmp_template_seg dest = ftmp_seg interp_step = 'nn' else: sct.printv('ERROR: Wrong image type.', 1, 'error') # apply transformation from previous step, to use as new src for registration sct_apply_transfo.main(args=[ '-i', src, '-d', dest, '-w', warp_forward, '-o', add_suffix(src, '_regStep' + str(i_step - 1)), '-x', interp_step]) src = add_suffix(src, '_regStep' + str(i_step - 1)) # register src --> dest # TODO: display param for debugging warp_forward_out, warp_inverse_out = register(src, dest, paramreg, param, str(i_step)) warp_forward.append(warp_forward_out) warp_inverse.insert(0, warp_inverse_out) # Concatenate transformations: sct.printv('\nConcatenate transformations: template --> subject...', verbose) sct_concat_transfo.main(args=[ '-w', warp_forward, '-d', 'data.nii', '-o', 'warp_template2anat.nii.gz']) sct.printv('\nConcatenate transformations: subject --> template...', verbose) sct_concat_transfo.main(args=[ '-w', warp_inverse, '-winv', ['template2subjectAffine.txt'], '-d', 'template.nii', '-o', 'warp_anat2template.nii.gz']) # Apply warping fields to anat and template sct.run(['sct_apply_transfo', '-i', 'template.nii', '-o', 'template2anat.nii.gz', '-d', 'data.nii', '-w', 'warp_template2anat.nii.gz', '-crop', '1'], verbose) sct.run(['sct_apply_transfo', '-i', 'data.nii', '-o', 'anat2template.nii.gz', '-d', 'template.nii', '-w', 'warp_anat2template.nii.gz', '-crop', '1'], verbose) # come back os.chdir(curdir) # Generate output files sct.printv('\nGenerate output files...', verbose) fname_template2anat = os.path.join(path_output, 'template2anat' + ext_data) fname_anat2template = os.path.join(path_output, 'anat2template' + ext_data) sct.generate_output_file(os.path.join(path_tmp, "warp_template2anat.nii.gz"), os.path.join(path_output, "warp_template2anat.nii.gz"), verbose) sct.generate_output_file(os.path.join(path_tmp, "warp_anat2template.nii.gz"), os.path.join(path_output, "warp_anat2template.nii.gz"), verbose) sct.generate_output_file(os.path.join(path_tmp, "template2anat.nii.gz"), fname_template2anat, verbose) sct.generate_output_file(os.path.join(path_tmp, "anat2template.nii.gz"), fname_anat2template, verbose) if ref == 'template': # copy straightening files in case subsequent SCT functions need them sct.generate_output_file(os.path.join(path_tmp, "warp_curve2straight.nii.gz"), os.path.join(path_output, "warp_curve2straight.nii.gz"), verbose) sct.generate_output_file(os.path.join(path_tmp, "warp_straight2curve.nii.gz"), os.path.join(path_output, "warp_straight2curve.nii.gz"), verbose) sct.generate_output_file(os.path.join(path_tmp, "straight_ref.nii.gz"), os.path.join(path_output, "straight_ref.nii.gz"), verbose) # Delete temporary files if param.remove_temp_files: sct.printv('\nDelete temporary files...', verbose) sct.rmtree(path_tmp, verbose=verbose) # display elapsed time elapsed_time = time.time() - start_time sct.printv('\nFinished! Elapsed time: ' + str(int(np.round(elapsed_time))) + 's', verbose) qc_dataset = arguments.get("-qc-dataset", None) qc_subject = arguments.get("-qc-subject", None) if param.path_qc is not None: generate_qc(fname_data, fname_in2=fname_template2anat, fname_seg=fname_seg, args=args, path_qc=os.path.abspath(param.path_qc), dataset=qc_dataset, subject=qc_subject, process='sct_register_to_template') sct.display_viewer_syntax([fname_data, fname_template2anat], verbose=verbose) sct.display_viewer_syntax([fname_template, fname_anat2template], verbose=verbose)
def deep_segmentation_spinalcord(im_image, contrast_type, ctr_algo='cnn', ctr_file=None, brain_bool=True, kernel_size='2d', remove_temp_files=1, verbose=1): """Pipeline""" # create temporary folder with intermediate results tmp_folder = sct.TempFolder(verbose=verbose) tmp_folder_path = tmp_folder.get_path() if ctr_algo == 'file': # if the ctr_file is provided tmp_folder.copy_from(ctr_file) file_ctr = os.path.basename(ctr_file) else: file_ctr = None tmp_folder.chdir() # re-orient image to RPI logger.info("Reorient the image to RPI, if necessary...") original_orientation = im_image.orientation fname_orient = 'image_in_RPI.nii' im_image.change_orientation('RPI').save(fname_orient) input_resolution = im_image.dim[4:7] # find the spinal cord centerline - execute OptiC binary logger.info("Finding the spinal cord centerline...") fname_res, centerline_filename = find_centerline(algo=ctr_algo, image_fname=fname_orient, contrast_type=contrast_type, brain_bool=brain_bool, folder_output=tmp_folder_path, remove_temp_files=remove_temp_files, centerline_fname=file_ctr) im_nii, ctr_nii = Image(fname_res), Image(centerline_filename) # crop image around the spinal cord centerline logger.info("Cropping the image around the spinal cord...") crop_size = 96 if (kernel_size == '3d' and contrast_type == 't2s') else 64 X_CROP_LST, Y_CROP_LST, Z_CROP_LST, im_crop_nii = crop_image_around_centerline(im_in=im_nii, ctr_in=ctr_nii, crop_size=crop_size) del ctr_nii # normalize the intensity of the images logger.info("Normalizing the intensity...") im_norm_in = apply_intensity_normalization(im_in=im_crop_nii) del im_crop_nii if kernel_size == '2d': # segment data using 2D convolutions logger.info("Segmenting the spinal cord using deep learning on 2D patches...") segmentation_model_fname = \ os.path.join(sct.__sct_dir__, 'data', 'deepseg_sc_models', '{}_sc.h5'.format(contrast_type)) seg_crop_data = segment_2d(model_fname=segmentation_model_fname, contrast_type=contrast_type, input_size=(crop_size, crop_size), im_in=im_norm_in) elif kernel_size == '3d': # segment data using 3D convolutions logger.info("Segmenting the spinal cord using deep learning on 3D patches...") segmentation_model_fname = \ os.path.join(sct.__sct_dir__, 'data', 'deepseg_sc_models', '{}_sc_3D.h5'.format(contrast_type)) seg_crop_data = segment_3d(model_fname=segmentation_model_fname, contrast_type=contrast_type, im_in=im_norm_in) del im_norm_in # reconstruct the segmentation from the crop data logger.info("Reassembling the image...") seg_uncrop_nii = uncrop_image(ref_in=im_nii, data_crop=seg_crop_data, x_crop_lst=X_CROP_LST, y_crop_lst=Y_CROP_LST, z_crop_lst=Z_CROP_LST) fname_res_seg = sct.add_suffix(fname_res, '_seg') seg_uncrop_nii.save(fname_res_seg) del seg_crop_data # resample to initial resolution logger.info("Resampling the segmentation to the original image resolution...") initial_resolution = 'x'.join([str(input_resolution[0]), str(input_resolution[1]), str(input_resolution[2])]) fname_res_seg_downsamp = sct.add_suffix(fname_res_seg, '_downsamp') resampling.resample_file(fname_res_seg, fname_res_seg_downsamp, initial_resolution, 'mm', 'linear', verbose=0) im_image_res_seg_downsamp = Image(fname_res_seg_downsamp) if ctr_algo == 'viewer': # resample and reorient the viewer labels fname_res_labels = sct.add_suffix(fname_orient, '_labels-centerline') resampling.resample_file(fname_res_labels, fname_res_labels, initial_resolution, 'mm', 'linear', verbose=0) im_image_res_labels_downsamp = Image(fname_res_labels).change_orientation(original_orientation) else: im_image_res_labels_downsamp = None if verbose == 2: fname_res_ctr = sct.add_suffix(fname_orient, '_ctr') resampling.resample_file(fname_res_ctr, fname_res_ctr, initial_resolution, 'mm', 'linear', verbose=0) im_image_res_ctr_downsamp = Image(fname_res_ctr).change_orientation(original_orientation) else: im_image_res_ctr_downsamp = None # binarize the resampled image to remove interpolation effects logger.info("Binarizing the segmentation to avoid interpolation effects...") thr = 0.0001 if contrast_type in ['t1', 'dwi'] else 0.5 # TODO: optimize speed --> np.where is slow im_image_res_seg_downsamp.data[np.where(im_image_res_seg_downsamp.data >= thr)] = 1 im_image_res_seg_downsamp.data[np.where(im_image_res_seg_downsamp.data < thr)] = 0 # post processing step to z_regularized im_image_res_seg_downsamp_postproc = post_processing_volume_wise(im_in=im_image_res_seg_downsamp) tmp_folder.chdir_undo() # remove temporary files if remove_temp_files: logger.info("Remove temporary files...") tmp_folder.cleanup() # reorient to initial orientation return im_image_res_seg_downsamp_postproc.change_orientation(original_orientation), \ im_nii, \ seg_uncrop_nii.change_orientation('RPI'), \ im_image_res_labels_downsamp, \ im_image_res_ctr_downsamp
def find_centerline(algo, image_fname, contrast_type, brain_bool, folder_output, remove_temp_files, centerline_fname): """ Assumes RPI orientation :param algo: :param image_fname: :param contrast_type: :param brain_bool: :param folder_output: :param remove_temp_files: :param centerline_fname: :return: """ # TODO: remove unnecessary i/o if Image(image_fname).dim[2] == 1: # isct_spine_detect requires nz > 1 im_concat = concat_data([image_fname, image_fname], dim=2) im_concat.save(sct.add_suffix(image_fname, '_concat')) image_fname = sct.add_suffix(image_fname, '_concat') bool_2d = True else: bool_2d = False # TODO: maybe change 'svm' for 'optic', because this is how we call it in sct_get_centerline if algo == 'svm': # run optic on a heatmap computed by a trained SVM+HoG algorithm # optic_models_fname = os.path.join(path_sct, 'data', 'optic_models', '{}_model'.format(contrast_type)) # # TODO: replace with get_centerline(method=optic) img_ctl, arr_ctl, _ = get_centerline(Image(image_fname), algo_fitting='optic', contrast=contrast_type) centerline_filename = sct.add_suffix(image_fname, "_ctr") img_ctl.save(centerline_filename) elif algo == 'cnn': # CNN parameters dct_patch_ctr = {'t2': {'size': (80, 80), 'mean': 51.1417, 'std': 57.4408}, 't2s': {'size': (80, 80), 'mean': 68.8591, 'std': 71.4659}, 't1': {'size': (80, 80), 'mean': 55.7359, 'std': 64.3149}, 'dwi': {'size': (80, 80), 'mean': 55.744, 'std': 45.003}} dct_params_ctr = {'t2': {'features': 16, 'dilation_layers': 2}, 't2s': {'features': 8, 'dilation_layers': 3}, 't1': {'features': 24, 'dilation_layers': 3}, 'dwi': {'features': 8, 'dilation_layers': 2}} # load model ctr_model_fname = os.path.join(sct.__sct_dir__, 'data', 'deepseg_sc_models', '{}_ctr.h5'.format(contrast_type)) ctr_model = nn_architecture_ctr(height=dct_patch_ctr[contrast_type]['size'][0], width=dct_patch_ctr[contrast_type]['size'][1], channels=1, classes=1, features=dct_params_ctr[contrast_type]['features'], depth=2, temperature=1.0, padding='same', batchnorm=True, dropout=0.0, dilation_layers=dct_params_ctr[contrast_type]['dilation_layers']) ctr_model.load_weights(ctr_model_fname) logger.info("Resample the image to 0.5x0.5 mm in-plane resolution...") fname_res = sct.add_suffix(image_fname, '_resampled') input_resolution = Image(image_fname).dim[4:7] new_resolution = 'x'.join(['0.5', '0.5', str(input_resolution[2])]) resampling.resample_file(image_fname, fname_res, new_resolution, 'mm', 'linear', verbose=0) # compute the heatmap fname_heatmap = sct.add_suffix(image_fname, "_heatmap") img_filename = ''.join(sct.extract_fname(fname_heatmap)[:2]) fname_heatmap_nii = img_filename + '.nii' z_max = heatmap(filename_in=fname_res, filename_out=fname_heatmap_nii, model=ctr_model, patch_shape=dct_patch_ctr[contrast_type]['size'], mean_train=dct_patch_ctr[contrast_type]['mean'], std_train=dct_patch_ctr[contrast_type]['std'], brain_bool=brain_bool) # run optic on the heatmap centerline_filename = sct.add_suffix(fname_heatmap, "_ctr") heatmap2optic(fname_heatmap=fname_heatmap_nii, lambda_value=7 if contrast_type == 't2s' else 1, fname_out=centerline_filename, z_max=z_max if brain_bool else None) elif algo == 'viewer': im_labels = _call_viewer_centerline(Image(image_fname)) im_centerline, arr_centerline, _ = get_centerline(im_labels) centerline_filename = sct.add_suffix(image_fname, "_ctr") labels_filename = sct.add_suffix(image_fname, "_labels-centerline") im_centerline.save(centerline_filename) im_labels.save(labels_filename) elif algo == 'file': centerline_filename = sct.add_suffix(image_fname, "_ctr") # Re-orient the manual centerline Image(centerline_fname).change_orientation('RPI').save(centerline_filename) else: logger.error('The parameter "-centerline" is incorrect. Please try again.') sys.exit(1) if algo != 'cnn': logger.info("Resample the image to 0.5x0.5 mm in-plane resolution...") fname_res = sct.add_suffix(image_fname, '_resampled') input_resolution = Image(image_fname).dim[4:7] new_resolution = 'x'.join(['0.5', '0.5', str(input_resolution[2])]) resampling.resample_file(image_fname, fname_res, new_resolution, 'mm', 'linear', verbose=0) resampling.resample_file(centerline_filename, centerline_filename, new_resolution, 'mm', 'linear', verbose=0) if bool_2d: im_split_lst = split_data(Image(centerline_filename), dim=2) im_split_lst[0].save(centerline_filename) return fname_res, centerline_filename
def deep_segmentation_MSlesion(im_image, contrast_type, ctr_algo='svm', ctr_file=None, brain_bool=True, remove_temp_files=1, verbose=1): """ Segment lesions from MRI data. :param im_image: Image() object containing the lesions to segment :param contrast_type: Constrast of the image. Need to use one supported by the CNN models. :param ctr_algo: Algo to find the centerline. See sct_get_centerline :param ctr_file: Centerline or segmentation (optional) :param brain_bool: If brain if present or not in the image. :param remove_temp_files: :return: """ # create temporary folder with intermediate results tmp_folder = sct.TempFolder(verbose=verbose) tmp_folder_path = tmp_folder.get_path() if ctr_algo == 'file': # if the ctr_file is provided tmp_folder.copy_from(ctr_file) file_ctr = os.path.basename(ctr_file) else: file_ctr = None tmp_folder.chdir() # orientation of the image, should be RPI logger.info("\nReorient the image to RPI, if necessary...") fname_in = im_image.absolutepath original_orientation = im_image.orientation fname_orient = 'image_in_RPI.nii' im_image.change_orientation('RPI').save(fname_orient) input_resolution = im_image.dim[4:7] # find the spinal cord centerline - execute OptiC binary logger.info("\nFinding the spinal cord centerline...") contrast_type_ctr = contrast_type.split('_')[0] fname_res, centerline_filename = find_centerline(algo=ctr_algo, image_fname=fname_orient, contrast_type=contrast_type_ctr, brain_bool=brain_bool, folder_output=tmp_folder_path, remove_temp_files=remove_temp_files, centerline_fname=file_ctr) im_nii, ctr_nii = Image(fname_res), Image(centerline_filename) # crop image around the spinal cord centerline logger.info("\nCropping the image around the spinal cord...") crop_size = 48 X_CROP_LST, Y_CROP_LST, Z_CROP_LST, im_crop_nii = crop_image_around_centerline(im_in=im_nii, ctr_in=ctr_nii, crop_size=crop_size) del ctr_nii # normalize the intensity of the images logger.info("Normalizing the intensity...") im_norm_in = apply_intensity_normalization(img=im_crop_nii, contrast=contrast_type) del im_crop_nii # resample to 0.5mm isotropic fname_norm = sct.add_suffix(fname_orient, '_norm') im_norm_in.save(fname_norm) fname_res3d = sct.add_suffix(fname_norm, '_resampled3d') resampling.resample_file(fname_norm, fname_res3d, '0.5x0.5x0.5', 'mm', 'linear', verbose=0) # segment data using 3D convolutions logger.info("\nSegmenting the MS lesions using deep learning on 3D patches...") segmentation_model_fname = os.path.join(sct.__sct_dir__, 'data', 'deepseg_lesion_models', '{}_lesion.h5'.format(contrast_type)) fname_seg_crop_res = sct.add_suffix(fname_res3d, '_lesionseg') im_res3d = Image(fname_res3d) seg_im = segment_3d(model_fname=segmentation_model_fname, contrast_type=contrast_type, im=im_res3d.copy()) seg_im.save(fname_seg_crop_res) del im_res3d, seg_im # resample to the initial pz resolution fname_seg_res2d = sct.add_suffix(fname_seg_crop_res, '_resampled2d') initial_2d_resolution = 'x'.join(['0.5', '0.5', str(input_resolution[2])]) resampling.resample_file(fname_seg_crop_res, fname_seg_res2d, initial_2d_resolution, 'mm', 'linear', verbose=0) seg_crop = Image(fname_seg_res2d) # reconstruct the segmentation from the crop data logger.info("\nReassembling the image...") seg_uncrop_nii = uncrop_image(ref_in=im_nii, data_crop=seg_crop.copy().data, x_crop_lst=X_CROP_LST, y_crop_lst=Y_CROP_LST, z_crop_lst=Z_CROP_LST) fname_seg_res_RPI = sct.add_suffix(fname_in, '_res_RPI_seg') seg_uncrop_nii.save(fname_seg_res_RPI) del seg_crop # resample to initial resolution logger.info("Resampling the segmentation to the original image resolution...") initial_resolution = 'x'.join([str(input_resolution[0]), str(input_resolution[1]), str(input_resolution[2])]) fname_seg_RPI = sct.add_suffix(fname_in, '_RPI_seg') resampling.resample_file(fname_seg_res_RPI, fname_seg_RPI, initial_resolution, 'mm', 'linear', verbose=0) seg_initres_nii = Image(fname_seg_RPI) if ctr_algo == 'viewer': # resample and reorient the viewer labels fname_res_labels = sct.add_suffix(fname_orient, '_labels-centerline') resampling.resample_file(fname_res_labels, fname_res_labels, initial_resolution, 'mm', 'linear', verbose=0) im_image_res_labels_downsamp = Image(fname_res_labels).change_orientation(original_orientation) else: im_image_res_labels_downsamp = None if verbose == 2: fname_res_ctr = sct.add_suffix(fname_orient, '_ctr') resampling.resample_file(fname_res_ctr, fname_res_ctr, initial_resolution, 'mm', 'linear', verbose=0) im_image_res_ctr_downsamp = Image(fname_res_ctr).change_orientation(original_orientation) else: im_image_res_ctr_downsamp = None # binarize the resampled image to remove interpolation effects logger.info("\nBinarizing the segmentation to avoid interpolation effects...") thr = 0.1 seg_initres_nii.data[np.where(seg_initres_nii.data >= thr)] = 1 seg_initres_nii.data[np.where(seg_initres_nii.data < thr)] = 0 # reorient to initial orientation logger.info("\nReorienting the segmentation to the original image orientation...") tmp_folder.chdir_undo() # remove temporary files if remove_temp_files: logger.info("\nRemove temporary files...") tmp_folder.cleanup() # reorient to initial orientation return seg_initres_nii.change_orientation(original_orientation), im_image_res_labels_downsamp, im_image_res_ctr_downsamp
def deep_segmentation_MSlesion(im_image, contrast_type, ctr_algo='svm', ctr_file=None, brain_bool=True, remove_temp_files=1, verbose=1): """ Segment lesions from MRI data. :param im_image: Image() object containing the lesions to segment :param contrast_type: Constrast of the image. Need to use one supported by the CNN models. :param ctr_algo: Algo to find the centerline. See sct_get_centerline :param ctr_file: Centerline or segmentation (optional) :param brain_bool: If brain if present or not in the image. :param remove_temp_files: :return: """ # create temporary folder with intermediate results tmp_folder = sct.TempFolder(verbose=verbose) tmp_folder_path = tmp_folder.get_path() if ctr_algo == 'file': # if the ctr_file is provided tmp_folder.copy_from(ctr_file) file_ctr = os.path.basename(ctr_file) else: file_ctr = None tmp_folder.chdir() fname_in = im_image.absolutepath # re-orient image to RPI logger.info("Reorient the image to RPI, if necessary...") original_orientation = im_image.orientation # fname_orient = 'image_in_RPI.nii' im_image.change_orientation('RPI') input_resolution = im_image.dim[4:7] # Resample image to 0.5mm in plane im_image_res = \ resampling.resample_nib(im_image, new_size=[0.5, 0.5, im_image.dim[6]], new_size_type='mm', interpolation='linear') fname_orient = 'image_in_RPI_res.nii' im_image_res.save(fname_orient) # find the spinal cord centerline - execute OptiC binary logger.info("\nFinding the spinal cord centerline...") contrast_type_ctr = contrast_type.split('_')[0] _, im_ctl, im_labels_viewer = find_centerline( algo=ctr_algo, image_fname=fname_orient, contrast_type=contrast_type_ctr, brain_bool=brain_bool, folder_output=tmp_folder_path, remove_temp_files=remove_temp_files, centerline_fname=file_ctr) if ctr_algo == 'file': im_ctl = \ resampling.resample_nib(im_ctl, new_size=[0.5, 0.5, im_image.dim[6]], new_size_type='mm', interpolation='linear') # crop image around the spinal cord centerline logger.info("\nCropping the image around the spinal cord...") crop_size = 48 X_CROP_LST, Y_CROP_LST, Z_CROP_LST, im_crop_nii = crop_image_around_centerline( im_in=im_image_res, ctr_in=im_ctl, crop_size=crop_size) del im_ctl # normalize the intensity of the images logger.info("Normalizing the intensity...") im_norm_in = apply_intensity_normalization(img=im_crop_nii, contrast=contrast_type) del im_crop_nii # resample to 0.5mm isotropic fname_norm = sct.add_suffix(fname_orient, '_norm') im_norm_in.save(fname_norm) fname_res3d = sct.add_suffix(fname_norm, '_resampled3d') resampling.resample_file(fname_norm, fname_res3d, '0.5x0.5x0.5', 'mm', 'linear', verbose=0) # segment data using 3D convolutions logger.info( "\nSegmenting the MS lesions using deep learning on 3D patches...") segmentation_model_fname = sct_dir_local_path( 'data', 'deepseg_lesion_models', '{}_lesion.h5'.format(contrast_type)) fname_seg_crop_res = sct.add_suffix(fname_res3d, '_lesionseg') im_res3d = Image(fname_res3d) seg_im = segment_3d(model_fname=segmentation_model_fname, contrast_type=contrast_type, im=im_res3d.copy()) seg_im.save(fname_seg_crop_res) del im_res3d, seg_im # resample to the initial pz resolution fname_seg_res2d = sct.add_suffix(fname_seg_crop_res, '_resampled2d') initial_2d_resolution = 'x'.join(['0.5', '0.5', str(input_resolution[2])]) resampling.resample_file(fname_seg_crop_res, fname_seg_res2d, initial_2d_resolution, 'mm', 'linear', verbose=0) seg_crop = Image(fname_seg_res2d) # reconstruct the segmentation from the crop data logger.info("\nReassembling the image...") seg_uncrop_nii = uncrop_image(ref_in=im_image_res, data_crop=seg_crop.copy().data, x_crop_lst=X_CROP_LST, y_crop_lst=Y_CROP_LST, z_crop_lst=Z_CROP_LST) fname_seg_res_RPI = sct.add_suffix(fname_in, '_res_RPI_seg') seg_uncrop_nii.save(fname_seg_res_RPI) del seg_crop # resample to initial resolution logger.info( "Resampling the segmentation to the original image resolution...") initial_resolution = 'x'.join([ str(input_resolution[0]), str(input_resolution[1]), str(input_resolution[2]) ]) fname_seg_RPI = sct.add_suffix(fname_in, '_RPI_seg') resampling.resample_file(fname_seg_res_RPI, fname_seg_RPI, initial_resolution, 'mm', 'linear', verbose=0) seg_initres_nii = Image(fname_seg_RPI) if ctr_algo == 'viewer': # resample and reorient the viewer labels im_labels_viewer_nib = nib.nifti1.Nifti1Image( im_labels_viewer.data, im_labels_viewer.hdr.get_best_affine()) im_viewer_r_nib = resampling.resample_nib(im_labels_viewer_nib, new_size=input_resolution, new_size_type='mm', interpolation='linear') im_viewer = Image( im_viewer_r_nib.get_data(), hdr=im_viewer_r_nib.header, orientation='RPI', dim=im_viewer_r_nib.header.get_data_shape()).change_orientation( original_orientation) else: im_viewer = None if verbose == 2: fname_res_ctr = sct.add_suffix(fname_orient, '_ctr') resampling.resample_file(fname_res_ctr, fname_res_ctr, initial_resolution, 'mm', 'linear', verbose=0) im_image_res_ctr_downsamp = Image(fname_res_ctr).change_orientation( original_orientation) else: im_image_res_ctr_downsamp = None # binarize the resampled image to remove interpolation effects logger.info( "\nBinarizing the segmentation to avoid interpolation effects...") thr = 0.1 seg_initres_nii.data[np.where(seg_initres_nii.data >= thr)] = 1 seg_initres_nii.data[np.where(seg_initres_nii.data < thr)] = 0 # change data type seg_initres_nii.change_type(np.uint8) # reorient to initial orientation logger.info( "\nReorienting the segmentation to the original image orientation...") tmp_folder.chdir_undo() # remove temporary files if remove_temp_files: logger.info("\nRemove temporary files...") tmp_folder.cleanup() # reorient to initial orientation return seg_initres_nii.change_orientation( original_orientation), im_viewer, im_image_res_ctr_downsamp
def deep_segmentation_MSlesion(im_image, contrast_type, ctr_algo='svm', ctr_file=None, brain_bool=True, remove_temp_files=1): """ Segment lesions from MRI data. :param im_image: Image() object containing the lesions to segment :param contrast_type: Constrast of the image. Need to use one supported by the CNN models. :param ctr_algo: Algo to find the centerline. See sct_get_centerline :param ctr_file: Centerline or segmentation (optional) :param brain_bool: If brain if present or not in the image. :param remove_temp_files: :return: """ # create temporary folder with intermediate results sct.log.info("\nCreating temporary folder...") tmp_folder = sct.TempFolder() tmp_folder_path = tmp_folder.get_path() if ctr_algo == 'manual': # if the ctr_file is provided tmp_folder.copy_from(ctr_file) file_ctr = os.path.basename(ctr_file) else: file_ctr = None tmp_folder.chdir() # orientation of the image, should be RPI sct.log.info("\nReorient the image to RPI, if necessary...") fname_in = im_image.absolutepath original_orientation = im_image.orientation fname_orient = 'image_in_RPI.nii' im_image.change_orientation('RPI').save(fname_orient) input_resolution = im_image.dim[4:7] # find the spinal cord centerline - execute OptiC binary sct.log.info("\nFinding the spinal cord centerline...") contrast_type_ctr = contrast_type.split('_')[0] fname_res, centerline_filename = find_centerline( algo=ctr_algo, image_fname=fname_orient, contrast_type=contrast_type_ctr, brain_bool=brain_bool, folder_output=tmp_folder_path, remove_temp_files=remove_temp_files, centerline_fname=file_ctr) im_nii, ctr_nii = Image(fname_res), Image(centerline_filename) # crop image around the spinal cord centerline sct.log.info("\nCropping the image around the spinal cord...") crop_size = 48 X_CROP_LST, Y_CROP_LST, Z_CROP_LST, im_crop_nii = crop_image_around_centerline( im_in=im_nii, ctr_in=ctr_nii, crop_size=crop_size) del ctr_nii # normalize the intensity of the images sct.log.info("Normalizing the intensity...") im_norm_in = apply_intensity_normalization(img=im_crop_nii, contrast=contrast_type) del im_crop_nii # resample to 0.5mm isotropic fname_norm = sct.add_suffix(fname_orient, '_norm') im_norm_in.save(fname_norm) fname_res3d = sct.add_suffix(fname_norm, '_resampled3d') resampling.resample_file(fname_norm, fname_res3d, '0.5x0.5x0.5', 'mm', 'linear', verbose=0) # segment data using 3D convolutions sct.log.info( "\nSegmenting the MS lesions using deep learning on 3D patches...") segmentation_model_fname = os.path.join( sct.__sct_dir__, 'data', 'deepseg_lesion_models', '{}_lesion.h5'.format(contrast_type)) fname_seg_crop_res = sct.add_suffix(fname_res3d, '_lesionseg') im_res3d = Image(fname_res3d) seg_im = segment_3d(model_fname=segmentation_model_fname, contrast_type=contrast_type, im=im_res3d.copy()) seg_im.save(fname_seg_crop_res) del im_res3d, seg_im # resample to the initial pz resolution fname_seg_res2d = sct.add_suffix(fname_seg_crop_res, '_resampled2d') initial_2d_resolution = 'x'.join(['0.5', '0.5', str(input_resolution[2])]) resampling.resample_file(fname_seg_crop_res, fname_seg_res2d, initial_2d_resolution, 'mm', 'linear', verbose=0) seg_crop = Image(fname_seg_res2d) # reconstruct the segmentation from the crop data sct.log.info("\nReassembling the image...") seg_uncrop_nii = uncrop_image(ref_in=im_nii, data_crop=seg_crop.copy().data, x_crop_lst=X_CROP_LST, y_crop_lst=Y_CROP_LST, z_crop_lst=Z_CROP_LST) fname_seg_res_RPI = sct.add_suffix(fname_in, '_res_RPI_seg') seg_uncrop_nii.save(fname_seg_res_RPI) del seg_crop # resample to initial resolution sct.log.info( "Resampling the segmentation to the original image resolution...") initial_resolution = 'x'.join([ str(input_resolution[0]), str(input_resolution[1]), str(input_resolution[2]) ]) fname_seg_RPI = sct.add_suffix(fname_in, '_RPI_seg') resampling.resample_file(fname_seg_res_RPI, fname_seg_RPI, initial_resolution, 'mm', 'linear', verbose=0) seg_initres_nii = Image(fname_seg_RPI) # binarize the resampled image to remove interpolation effects sct.log.info( "\nBinarizing the segmentation to avoid interpolation effects...") thr = 0.1 seg_initres_nii.data[np.where(seg_initres_nii.data >= thr)] = 1 seg_initres_nii.data[np.where(seg_initres_nii.data < thr)] = 0 # reorient to initial orientation sct.log.info( "\nReorienting the segmentation to the original image orientation...") tmp_folder.chdir_undo() # remove temporary files if remove_temp_files: sct.log.info("\nRemove temporary files...") tmp_folder.cleanup() # reorient to initial orientation return seg_initres_nii.change_orientation(original_orientation)
def main(args=None): # initializations param = Param() # check user arguments if not args: args = sys.argv[1:] # Get parser info parser = get_parser() arguments = parser.parse(args) fname_data = arguments['-i'] fname_seg = arguments['-s'] if '-l' in arguments: fname_landmarks = arguments['-l'] label_type = 'body' elif '-ldisc' in arguments: fname_landmarks = arguments['-ldisc'] label_type = 'disc' else: sct.printv('ERROR: Labels should be provided.', 1, 'error') if '-ofolder' in arguments: path_output = arguments['-ofolder'] else: path_output = '' param.path_qc = arguments.get("-qc", None) path_template = arguments['-t'] contrast_template = arguments['-c'] ref = arguments['-ref'] param.remove_temp_files = int(arguments.get('-r')) verbose = int(arguments.get('-v')) sct.init_sct(log_level=verbose, update=True) # Update log level param.verbose = verbose # TODO: not clean, unify verbose or param.verbose in code, but not both param.straighten_fitting = arguments['-straighten-fitting'] # if '-cpu-nb' in arguments: # arg_cpu = ' -cpu-nb '+str(arguments['-cpu-nb']) # else: # arg_cpu = '' # registration parameters if '-param' in arguments: # reset parameters but keep step=0 (might be overwritten if user specified step=0) paramreg = ParamregMultiStep([step0]) if ref == 'subject': paramreg.steps['0'].dof = 'Tx_Ty_Tz_Rx_Ry_Rz_Sz' # add user parameters for paramStep in arguments['-param']: paramreg.addStep(paramStep) else: paramreg = ParamregMultiStep([step0, step1, step2]) # if ref=subject, initialize registration using different affine parameters if ref == 'subject': paramreg.steps['0'].dof = 'Tx_Ty_Tz_Rx_Ry_Rz_Sz' # initialize other parameters zsubsample = param.zsubsample # retrieve template file names file_template_vertebral_labeling = get_file_label(os.path.join(path_template, 'template'), 'vertebral labeling') file_template = get_file_label(os.path.join(path_template, 'template'), contrast_template.upper() + '-weighted template') file_template_seg = get_file_label(os.path.join(path_template, 'template'), 'spinal cord') # start timer start_time = time.time() # get fname of the template + template objects fname_template = os.path.join(path_template, 'template', file_template) fname_template_vertebral_labeling = os.path.join(path_template, 'template', file_template_vertebral_labeling) fname_template_seg = os.path.join(path_template, 'template', file_template_seg) fname_template_disc_labeling = os.path.join(path_template, 'template', 'PAM50_label_disc.nii.gz') # check file existence # TODO: no need to do that! sct.printv('\nCheck template files...') sct.check_file_exist(fname_template, verbose) sct.check_file_exist(fname_template_vertebral_labeling, verbose) sct.check_file_exist(fname_template_seg, verbose) path_data, file_data, ext_data = sct.extract_fname(fname_data) # sct.printv(arguments) sct.printv('\nCheck parameters:', verbose) sct.printv(' Data: ' + fname_data, verbose) sct.printv(' Landmarks: ' + fname_landmarks, verbose) sct.printv(' Segmentation: ' + fname_seg, verbose) sct.printv(' Path template: ' + path_template, verbose) sct.printv(' Remove temp files: ' + str(param.remove_temp_files), verbose) # check input labels labels = check_labels(fname_landmarks, label_type=label_type) vertebral_alignment = False if len(labels) > 2 and label_type == 'disc': vertebral_alignment = True path_tmp = sct.tmp_create(basename="register_to_template", verbose=verbose) # set temporary file names ftmp_data = 'data.nii' ftmp_seg = 'seg.nii.gz' ftmp_label = 'label.nii.gz' ftmp_template = 'template.nii' ftmp_template_seg = 'template_seg.nii.gz' ftmp_template_label = 'template_label.nii.gz' # copy files to temporary folder sct.printv('\nCopying input data to tmp folder and convert to nii...', verbose) Image(fname_data).save(os.path.join(path_tmp, ftmp_data)) Image(fname_seg).save(os.path.join(path_tmp, ftmp_seg)) Image(fname_landmarks).save(os.path.join(path_tmp, ftmp_label)) Image(fname_template).save(os.path.join(path_tmp, ftmp_template)) Image(fname_template_seg).save(os.path.join(path_tmp, ftmp_template_seg)) Image(fname_template_vertebral_labeling).save(os.path.join(path_tmp, ftmp_template_label)) if label_type == 'disc': Image(fname_template_disc_labeling).save(os.path.join(path_tmp, ftmp_template_label)) # go to tmp folder curdir = os.getcwd() os.chdir(path_tmp) # Generate labels from template vertebral labeling if label_type == 'body': sct.printv('\nGenerate labels from template vertebral labeling', verbose) ftmp_template_label_, ftmp_template_label = ftmp_template_label, sct.add_suffix(ftmp_template_label, "_body") sct_label_utils.main(args=['-i', ftmp_template_label_, '-vert-body', '0', '-o', ftmp_template_label]) # check if provided labels are available in the template sct.printv('\nCheck if provided labels are available in the template', verbose) image_label_template = Image(ftmp_template_label) labels_template = image_label_template.getNonZeroCoordinates(sorting='value') if labels[-1].value > labels_template[-1].value: sct.printv('ERROR: Wrong landmarks input. Labels must have correspondence in template space. \nLabel max ' 'provided: ' + str(labels[-1].value) + '\nLabel max from template: ' + str(labels_template[-1].value), verbose, 'error') # if only one label is present, force affine transformation to be Tx,Ty,Tz only (no scaling) if len(labels) == 1: paramreg.steps['0'].dof = 'Tx_Ty_Tz' sct.printv('WARNING: Only one label is present. Forcing initial transformation to: ' + paramreg.steps['0'].dof, 1, 'warning') # Project labels onto the spinal cord centerline because later, an affine transformation is estimated between the # template's labels (centered in the cord) and the subject's labels (assumed to be centered in the cord). # If labels are not centered, mis-registration errors are observed (see issue #1826) ftmp_label = project_labels_on_spinalcord(ftmp_label, ftmp_seg) # binarize segmentation (in case it has values below 0 caused by manual editing) sct.printv('\nBinarize segmentation', verbose) ftmp_seg_, ftmp_seg = ftmp_seg, sct.add_suffix(ftmp_seg, "_bin") sct_maths.main(['-i', ftmp_seg_, '-bin', '0.5', '-o', ftmp_seg]) # Switch between modes: subject->template or template->subject if ref == 'template': # resample data to 1mm isotropic sct.printv('\nResample data to 1mm isotropic...', verbose) resample_file(ftmp_data, add_suffix(ftmp_data, '_1mm'), '1.0x1.0x1.0', 'mm', 'linear', verbose) ftmp_data = add_suffix(ftmp_data, '_1mm') resample_file(ftmp_seg, add_suffix(ftmp_seg, '_1mm'), '1.0x1.0x1.0', 'mm', 'linear', verbose) ftmp_seg = add_suffix(ftmp_seg, '_1mm') # N.B. resampling of labels is more complicated, because they are single-point labels, therefore resampling # with nearest neighbour can make them disappear. resample_labels(ftmp_label, ftmp_data, add_suffix(ftmp_label, '_1mm')) ftmp_label = add_suffix(ftmp_label, '_1mm') # Change orientation of input images to RPI sct.printv('\nChange orientation of input images to RPI...', verbose) ftmp_data = Image(ftmp_data).change_orientation("RPI", generate_path=True).save().absolutepath ftmp_seg = Image(ftmp_seg).change_orientation("RPI", generate_path=True).save().absolutepath ftmp_label = Image(ftmp_label).change_orientation("RPI", generate_path=True).save().absolutepath ftmp_seg_, ftmp_seg = ftmp_seg, add_suffix(ftmp_seg, '_crop') if vertebral_alignment: # cropping the segmentation based on the label coverage to ensure good registration with vertebral alignment # See https://github.com/neuropoly/spinalcordtoolbox/pull/1669 for details image_labels = Image(ftmp_label) coordinates_labels = image_labels.getNonZeroCoordinates(sorting='z') nx, ny, nz, nt, px, py, pz, pt = image_labels.dim offset_crop = 10.0 * pz # cropping the image 10 mm above and below the highest and lowest label cropping_slices = [coordinates_labels[0].z - offset_crop, coordinates_labels[-1].z + offset_crop] # make sure that the cropping slices do not extend outside of the slice range (issue #1811) if cropping_slices[0] < 0: cropping_slices[0] = 0 if cropping_slices[1] > nz: cropping_slices[1] = nz msct_image.spatial_crop(Image(ftmp_seg_), dict(((2, np.int32(np.round(cropping_slices))),))).save(ftmp_seg) else: # if we do not align the vertebral levels, we crop the segmentation from top to bottom im_seg_rpi = Image(ftmp_seg_) bottom = 0 for data in msct_image.SlicerOneAxis(im_seg_rpi, "IS"): if (data != 0).any(): break bottom += 1 top = im_seg_rpi.data.shape[2] for data in msct_image.SlicerOneAxis(im_seg_rpi, "SI"): if (data != 0).any(): break top -= 1 msct_image.spatial_crop(im_seg_rpi, dict(((2, (bottom, top)),))).save(ftmp_seg) # straighten segmentation sct.printv('\nStraighten the spinal cord using centerline/segmentation...', verbose) # check if warp_curve2straight and warp_straight2curve already exist (i.e. no need to do it another time) fn_warp_curve2straight = os.path.join(curdir, "warp_curve2straight.nii.gz") fn_warp_straight2curve = os.path.join(curdir, "warp_straight2curve.nii.gz") fn_straight_ref = os.path.join(curdir, "straight_ref.nii.gz") cache_input_files=[ftmp_seg] if vertebral_alignment: cache_input_files += [ ftmp_template_seg, ftmp_label, ftmp_template_label, ] cache_sig = sct.cache_signature( input_files=cache_input_files, ) cachefile = os.path.join(curdir, "straightening.cache") if sct.cache_valid(cachefile, cache_sig) and os.path.isfile(fn_warp_curve2straight) and os.path.isfile(fn_warp_straight2curve) and os.path.isfile(fn_straight_ref): sct.printv('Reusing existing warping field which seems to be valid', verbose, 'warning') sct.copy(fn_warp_curve2straight, 'warp_curve2straight.nii.gz') sct.copy(fn_warp_straight2curve, 'warp_straight2curve.nii.gz') sct.copy(fn_straight_ref, 'straight_ref.nii.gz') # apply straightening sct.run(['sct_apply_transfo', '-i', ftmp_seg, '-w', 'warp_curve2straight.nii.gz', '-d', 'straight_ref.nii.gz', '-o', add_suffix(ftmp_seg, '_straight')]) else: from spinalcordtoolbox.straightening import SpinalCordStraightener sc_straight = SpinalCordStraightener(ftmp_seg, ftmp_seg) sc_straight.algo_fitting = param.straighten_fitting sc_straight.output_filename = add_suffix(ftmp_seg, '_straight') sc_straight.path_output = './' sc_straight.qc = '0' sc_straight.remove_temp_files = param.remove_temp_files sc_straight.verbose = verbose if vertebral_alignment: sc_straight.centerline_reference_filename = ftmp_template_seg sc_straight.use_straight_reference = True sc_straight.discs_input_filename = ftmp_label sc_straight.discs_ref_filename = ftmp_template_label sc_straight.straighten() sct.cache_save(cachefile, cache_sig) # N.B. DO NOT UPDATE VARIABLE ftmp_seg BECAUSE TEMPORARY USED LATER # re-define warping field using non-cropped space (to avoid issue #367) s, o = sct.run(['sct_concat_transfo', '-w', 'warp_straight2curve.nii.gz', '-d', ftmp_data, '-o', 'warp_straight2curve.nii.gz']) if vertebral_alignment: sct.copy('warp_curve2straight.nii.gz', 'warp_curve2straightAffine.nii.gz') else: # Label preparation: # -------------------------------------------------------------------------------- # Remove unused label on template. Keep only label present in the input label image sct.printv('\nRemove unused label on template. Keep only label present in the input label image...', verbose) sct.run(['sct_label_utils', '-i', ftmp_template_label, '-o', ftmp_template_label, '-remove-reference', ftmp_label]) # Dilating the input label so they can be straighten without losing them sct.printv('\nDilating input labels using 3vox ball radius') sct_maths.main(['-i', ftmp_label, '-dilate', '3', '-o', add_suffix(ftmp_label, '_dilate')]) ftmp_label = add_suffix(ftmp_label, '_dilate') # Apply straightening to labels sct.printv('\nApply straightening to labels...', verbose) sct.run(['sct_apply_transfo', '-i', ftmp_label, '-o', add_suffix(ftmp_label, '_straight'), '-d', add_suffix(ftmp_seg, '_straight'), '-w', 'warp_curve2straight.nii.gz', '-x', 'nn']) ftmp_label = add_suffix(ftmp_label, '_straight') # Compute rigid transformation straight landmarks --> template landmarks sct.printv('\nEstimate transformation for step #0...', verbose) try: register_landmarks(ftmp_label, ftmp_template_label, paramreg.steps['0'].dof, fname_affine='straight2templateAffine.txt', verbose=verbose) except RuntimeError: raise('Input labels do not seem to be at the right place. Please check the position of the labels. ' 'See documentation for more details: https://www.slideshare.net/neuropoly/sct-course-20190121/42') # Concatenate transformations: curve --> straight --> affine sct.printv('\nConcatenate transformations: curve --> straight --> affine...', verbose) sct.run(['sct_concat_transfo', '-w', 'warp_curve2straight.nii.gz,straight2templateAffine.txt', '-d', 'template.nii', '-o', 'warp_curve2straightAffine.nii.gz']) # Apply transformation sct.printv('\nApply transformation...', verbose) sct.run(['sct_apply_transfo', '-i', ftmp_data, '-o', add_suffix(ftmp_data, '_straightAffine'), '-d', ftmp_template, '-w', 'warp_curve2straightAffine.nii.gz']) ftmp_data = add_suffix(ftmp_data, '_straightAffine') sct.run(['sct_apply_transfo', '-i', ftmp_seg, '-o', add_suffix(ftmp_seg, '_straightAffine'), '-d', ftmp_template, '-w', 'warp_curve2straightAffine.nii.gz', '-x', 'linear']) ftmp_seg = add_suffix(ftmp_seg, '_straightAffine') """ # Benjamin: Issue from Allan Martin, about the z=0 slice that is screwed up, caused by the affine transform. # Solution found: remove slices below and above landmarks to avoid rotation effects points_straight = [] for coord in landmark_template: points_straight.append(coord.z) min_point, max_point = int(np.round(np.min(points_straight))), int(np.round(np.max(points_straight))) ftmp_seg_, ftmp_seg = ftmp_seg, add_suffix(ftmp_seg, '_black') msct_image.spatial_crop(Image(ftmp_seg_), dict(((2, (min_point,max_point)),))).save(ftmp_seg) """ # open segmentation im = Image(ftmp_seg) im_new = msct_image.empty_like(im) # binarize im_new.data = im.data > 0.5 # find min-max of anat2template (for subsequent cropping) zmin_template, zmax_template = msct_image.find_zmin_zmax(im_new, threshold=0.5) # save binarized segmentation im_new.save(add_suffix(ftmp_seg, '_bin')) # unused? # crop template in z-direction (for faster processing) # TODO: refactor to use python module instead of doing i/o sct.printv('\nCrop data in template space (for faster processing)...', verbose) ftmp_template_, ftmp_template = ftmp_template, add_suffix(ftmp_template, '_crop') msct_image.spatial_crop(Image(ftmp_template_), dict(((2, (zmin_template,zmax_template)),))).save(ftmp_template) ftmp_template_seg_, ftmp_template_seg = ftmp_template_seg, add_suffix(ftmp_template_seg, '_crop') msct_image.spatial_crop(Image(ftmp_template_seg_), dict(((2, (zmin_template,zmax_template)),))).save(ftmp_template_seg) ftmp_data_, ftmp_data = ftmp_data, add_suffix(ftmp_data, '_crop') msct_image.spatial_crop(Image(ftmp_data_), dict(((2, (zmin_template,zmax_template)),))).save(ftmp_data) ftmp_seg_, ftmp_seg = ftmp_seg, add_suffix(ftmp_seg, '_crop') msct_image.spatial_crop(Image(ftmp_seg_), dict(((2, (zmin_template,zmax_template)),))).save(ftmp_seg) # sub-sample in z-direction # TODO: refactor to use python module instead of doing i/o sct.printv('\nSub-sample in z-direction (for faster processing)...', verbose) sct.run(['sct_resample', '-i', ftmp_template, '-o', add_suffix(ftmp_template, '_sub'), '-f', '1x1x' + zsubsample], verbose) ftmp_template = add_suffix(ftmp_template, '_sub') sct.run(['sct_resample', '-i', ftmp_template_seg, '-o', add_suffix(ftmp_template_seg, '_sub'), '-f', '1x1x' + zsubsample], verbose) ftmp_template_seg = add_suffix(ftmp_template_seg, '_sub') sct.run(['sct_resample', '-i', ftmp_data, '-o', add_suffix(ftmp_data, '_sub'), '-f', '1x1x' + zsubsample], verbose) ftmp_data = add_suffix(ftmp_data, '_sub') sct.run(['sct_resample', '-i', ftmp_seg, '-o', add_suffix(ftmp_seg, '_sub'), '-f', '1x1x' + zsubsample], verbose) ftmp_seg = add_suffix(ftmp_seg, '_sub') # Registration straight spinal cord to template sct.printv('\nRegister straight spinal cord to template...', verbose) # loop across registration steps warp_forward = [] warp_inverse = [] for i_step in range(1, len(paramreg.steps)): sct.printv('\nEstimate transformation for step #' + str(i_step) + '...', verbose) # identify which is the src and dest if paramreg.steps[str(i_step)].type == 'im': src = ftmp_data dest = ftmp_template interp_step = 'linear' elif paramreg.steps[str(i_step)].type == 'seg': src = ftmp_seg dest = ftmp_template_seg interp_step = 'nn' else: sct.printv('ERROR: Wrong image type.', 1, 'error') if paramreg.steps[str(i_step)].algo == 'centermassrot' and paramreg.steps[str(i_step)].rot_method == 'hog': src_seg = ftmp_seg dest_seg = ftmp_template_seg # if step>1, apply warp_forward_concat to the src image to be used if i_step > 1: # apply transformation from previous step, to use as new src for registration sct.run(['sct_apply_transfo', '-i', src, '-d', dest, '-w', ','.join(warp_forward), '-o', add_suffix(src, '_regStep' + str(i_step - 1)), '-x', interp_step], verbose) src = add_suffix(src, '_regStep' + str(i_step - 1)) if paramreg.steps[str(i_step)].algo == 'centermassrot' and paramreg.steps[str(i_step)].rot_method == 'hog': # also apply transformation to the seg sct.run(['sct_apply_transfo', '-i', src_seg, '-d', dest_seg, '-w', ','.join(warp_forward), '-o', add_suffix(src, '_regStep' + str(i_step - 1)), '-x', interp_step], verbose) src_seg = add_suffix(src_seg, '_regStep' + str(i_step - 1)) # register src --> dest # TODO: display param for debugging if paramreg.steps[str(i_step)].algo == 'centermassrot' and paramreg.steps[str(i_step)].rot_method == 'hog': # im_seg case warp_forward_out, warp_inverse_out = register([src, src_seg], [dest, dest_seg], paramreg, param, str(i_step)) else: warp_forward_out, warp_inverse_out = register(src, dest, paramreg, param, str(i_step)) warp_forward.append(warp_forward_out) warp_inverse.append(warp_inverse_out) # Concatenate transformations: sct.printv('\nConcatenate transformations: anat --> template...', verbose) sct.run(['sct_concat_transfo', '-w', 'warp_curve2straightAffine.nii.gz,' + ','.join(warp_forward), '-d', 'template.nii', '-o', 'warp_anat2template.nii.gz'], verbose) # sct.run('sct_concat_transfo -w warp_curve2straight.nii.gz,straight2templateAffine.txt,'+','.join(warp_forward)+' -d template.nii -o warp_anat2template.nii.gz', verbose) sct.printv('\nConcatenate transformations: template --> anat...', verbose) warp_inverse.reverse() if vertebral_alignment: sct.run(['sct_concat_transfo', '-w', ','.join(warp_inverse) + ',warp_straight2curve.nii.gz', '-d', 'data.nii', '-o', 'warp_template2anat.nii.gz'], verbose) else: sct.run(['sct_concat_transfo', '-w', ','.join(warp_inverse) + ',-straight2templateAffine.txt,warp_straight2curve.nii.gz', '-d', 'data.nii', '-o', 'warp_template2anat.nii.gz'], verbose) # register template->subject elif ref == 'subject': # Change orientation of input images to RPI sct.printv('\nChange orientation of input images to RPI...', verbose) ftmp_data = Image(ftmp_data).change_orientation("RPI", generate_path=True).save().absolutepath ftmp_seg = Image(ftmp_seg).change_orientation("RPI", generate_path=True).save().absolutepath ftmp_label = Image(ftmp_label).change_orientation("RPI", generate_path=True).save().absolutepath # Remove unused label on template. Keep only label present in the input label image sct.printv('\nRemove unused label on template. Keep only label present in the input label image...', verbose) sct.run(['sct_label_utils', '-i', ftmp_template_label, '-o', ftmp_template_label, '-remove-reference', ftmp_label]) # Add one label because at least 3 orthogonal labels are required to estimate an affine transformation. This # new label is added at the level of the upper most label (lowest value), at 1cm to the right. for i_file in [ftmp_label, ftmp_template_label]: im_label = Image(i_file) coord_label = im_label.getCoordinatesAveragedByValue() # N.B. landmarks are sorted by value # Create new label from copy import deepcopy new_label = deepcopy(coord_label[0]) # move it 5mm to the left (orientation is RAS) nx, ny, nz, nt, px, py, pz, pt = im_label.dim new_label.x = np.round(coord_label[0].x + 5.0 / px) # assign value 99 new_label.value = 99 # Add to existing image im_label.data[int(new_label.x), int(new_label.y), int(new_label.z)] = new_label.value # Overwrite label file # im_label.absolutepath = 'label_rpi_modif.nii.gz' im_label.save() # Bring template to subject space using landmark-based transformation sct.printv('\nEstimate transformation for step #0...', verbose) warp_forward = ['template2subjectAffine.txt'] warp_inverse = ['-template2subjectAffine.txt'] try: register_landmarks(ftmp_template_label, ftmp_label, paramreg.steps['0'].dof, fname_affine=warp_forward[0], verbose=verbose, path_qc="./") except Exception: sct.printv('ERROR: input labels do not seem to be at the right place. Please check the position of the labels. See documentation for more details: https://www.slideshare.net/neuropoly/sct-course-20190121/42', verbose=verbose, type='error') # loop across registration steps for i_step in range(1, len(paramreg.steps)): sct.printv('\nEstimate transformation for step #' + str(i_step) + '...', verbose) # identify which is the src and dest if paramreg.steps[str(i_step)].type == 'im': src = ftmp_template dest = ftmp_data interp_step = 'linear' elif paramreg.steps[str(i_step)].type == 'seg': src = ftmp_template_seg dest = ftmp_seg interp_step = 'nn' else: sct.printv('ERROR: Wrong image type.', 1, 'error') # apply transformation from previous step, to use as new src for registration sct.run(['sct_apply_transfo', '-i', src, '-d', dest, '-w', ','.join(warp_forward), '-o', add_suffix(src, '_regStep' + str(i_step - 1)), '-x', interp_step], verbose) src = add_suffix(src, '_regStep' + str(i_step - 1)) # register src --> dest # TODO: display param for debugging warp_forward_out, warp_inverse_out = register(src, dest, paramreg, param, str(i_step)) warp_forward.append(warp_forward_out) warp_inverse.insert(0, warp_inverse_out) # Concatenate transformations: sct.printv('\nConcatenate transformations: template --> subject...', verbose) sct.run(['sct_concat_transfo', '-w', ','.join(warp_forward), '-d', 'data.nii', '-o', 'warp_template2anat.nii.gz'], verbose) sct.printv('\nConcatenate transformations: subject --> template...', verbose) sct.run(['sct_concat_transfo', '-w', ','.join(warp_inverse), '-d', 'template.nii', '-o', 'warp_anat2template.nii.gz'], verbose) # Apply warping fields to anat and template sct.run(['sct_apply_transfo', '-i', 'template.nii', '-o', 'template2anat.nii.gz', '-d', 'data.nii', '-w', 'warp_template2anat.nii.gz', '-crop', '1'], verbose) sct.run(['sct_apply_transfo', '-i', 'data.nii', '-o', 'anat2template.nii.gz', '-d', 'template.nii', '-w', 'warp_anat2template.nii.gz', '-crop', '1'], verbose) # come back os.chdir(curdir) # Generate output files sct.printv('\nGenerate output files...', verbose) fname_template2anat = os.path.join(path_output, 'template2anat' + ext_data) fname_anat2template = os.path.join(path_output, 'anat2template' + ext_data) sct.generate_output_file(os.path.join(path_tmp, "warp_template2anat.nii.gz"), os.path.join(path_output, "warp_template2anat.nii.gz"), verbose) sct.generate_output_file(os.path.join(path_tmp, "warp_anat2template.nii.gz"), os.path.join(path_output, "warp_anat2template.nii.gz"), verbose) sct.generate_output_file(os.path.join(path_tmp, "template2anat.nii.gz"), fname_template2anat, verbose) sct.generate_output_file(os.path.join(path_tmp, "anat2template.nii.gz"), fname_anat2template, verbose) if ref == 'template': # copy straightening files in case subsequent SCT functions need them sct.generate_output_file(os.path.join(path_tmp, "warp_curve2straight.nii.gz"), os.path.join(path_output, "warp_curve2straight.nii.gz"), verbose) sct.generate_output_file(os.path.join(path_tmp, "warp_straight2curve.nii.gz"), os.path.join(path_output, "warp_straight2curve.nii.gz"), verbose) sct.generate_output_file(os.path.join(path_tmp, "straight_ref.nii.gz"), os.path.join(path_output, "straight_ref.nii.gz"), verbose) # Delete temporary files if param.remove_temp_files: sct.printv('\nDelete temporary files...', verbose) sct.rmtree(path_tmp, verbose=verbose) # display elapsed time elapsed_time = time.time() - start_time sct.printv('\nFinished! Elapsed time: ' + str(int(np.round(elapsed_time))) + 's', verbose) qc_dataset = arguments.get("-qc-dataset", None) qc_subject = arguments.get("-qc-subject", None) if param.path_qc is not None: generate_qc(fname_data, fname_in2=fname_template2anat, fname_seg=fname_seg, args=args, path_qc=os.path.abspath(param.path_qc), dataset=qc_dataset, subject=qc_subject, process='sct_register_to_template') sct.display_viewer_syntax([fname_data, fname_template2anat], verbose=verbose) sct.display_viewer_syntax([fname_template, fname_anat2template], verbose=verbose)