def validation_on_dilated_lesions(normal_validation_dir,
dilated_validation_dir,
gt_dir,
evaluation_labels,
recompute=True):
utils.mkdir(dilated_validation_dir)
list_validation_subdir = utils.list_subfolders(normal_validation_dir)
loop_info = utils.LoopInfo(len(list_validation_subdir), 5, 'validating',
True)
for val_idx, validation_subdir in enumerate(list_validation_subdir):
loop_info.update(val_idx)
# dilate lesion
dilated_validation_subdir = os.path.join(
dilated_validation_dir, os.path.basename(validation_subdir))
dilate_lesions(validation_subdir,
dilated_validation_subdir,
recompute=recompute)
# compute new dice scores
path_dice = os.path.join(dilated_validation_subdir, 'dice.npy')
dice_evaluation(gt_dir,
dilated_validation_subdir,
evaluation_labels,
path_dice=path_dice,
recompute=recompute)
def estimate_t2_cropping(image_dir, result_dir=None, dilation=5):
"""This function takes all the hippocampus images (with 2 channels) within the specified directory, and estimates
the cropping dimensions around the hippocampus in the t2 channel.
It returns the mean and sts deviation for the minimal and maximal croppings, proportional to image size.
:param image_dir: path of the folder containing hippocampus images
:param result_dir: if not None, path of the folder where to write the computed statistics.
:param dilation: dilation coefficient used to extract full brain mask. Default is 5.
:returns t2_cropping_stats: numpy vector of size 4 [mean min crop, std min crop, mean max crop, std max crop]
"""
# create result dir
if result_dir is not None:
utils.mkdir(result_dir)
# loop through images
list_image_paths = utils.list_images_in_folder(image_dir)
max_cropping_proportions = np.zeros(len(list_image_paths))
min_cropping_proportions = np.zeros(len(list_image_paths))
loop_info = utils.LoopInfo(len(list_image_paths), 10, 'processing')
for im_idx, image_path in enumerate(list_image_paths):
loop_info.update(im_idx)
# load t2 channel
im = utils.load_volume(image_path)
t2 = im[..., 1]
shape = t2.shape
hdim = int(np.argmax(shape))
# mask image
_, mask = edit_volumes.mask_volume(t2,
threshold=0,
dilate=dilation,
return_mask=True)
# find cropping indices
indices = np.nonzero(mask)[hdim]
min_cropping_proportions[im_idx] = np.maximum(
np.min(indices) + int(dilation / 2), 0) / shape[hdim]
max_cropping_proportions[im_idx] = np.minimum(
np.max(indices) - int(dilation / 2), shape[hdim]) / shape[hdim]
# compute and save stats
t2_cropping_stats = np.array([
np.mean(min_cropping_proportions),
np.std(min_cropping_proportions),
np.mean(max_cropping_proportions),
np.std(max_cropping_proportions)
])
# save stats if necessary
if result_dir is not None:
np.save(os.path.join(result_dir, 't2_cropping_stats.npy'),
t2_cropping_stats)
return t2_cropping_stats
def paste_lesions_on_buckner(lesion_dir,
buckner_dir,
result_dir,
dilate=2,
recompute=False):
path_lesions = utils.list_images_in_folder(lesion_dir)
path_buckners = utils.list_images_in_folder(buckner_dir)
utils.mkdir(result_dir)
# loop over buckner label maps
loop_info = utils.LoopInfo(len(path_buckners), 1, 'processing', True)
for idx_buckner, path_buckner in enumerate(path_buckners):
loop_info.update(idx_buckner)
buckner_name = os.path.basename(path_buckner).replace(
'_seg', '').replace('.nii.gz', '')
buckner = utils.load_volume(path_buckner)
WM = (buckner == 2) | (buckner == 7) | (buckner == 16) | (
buckner == 41) | (buckner == 46)
# loop over challenge data
for path_lesion in path_lesions:
lesion_name = os.path.basename(path_lesion).replace(
'.samseg_and_lesions.nii.gz', '')
path_result = os.path.join(
result_dir, buckner_name + '_' + lesion_name + '.nii.gz')
if (not os.path.isfile(path_result)) | recompute:
lesion = utils.load_volume(path_lesion)
assert lesion.shape == buckner.shape, 'lesions should have same shape as buckner labels'
# define lesion, WM, and LV masks
lesion = (lesion == 77) & WM
LV_and_lesion = (buckner == 4) | lesion
# morphological operations to bridge the gaps between lesions and LV
morph_struct = utils.build_binary_structure(
dilate, len(lesion.shape))
lesion = binary_dilation(LV_and_lesion, morph_struct)
lesion = binary_erosion(lesion, morph_struct)
lesion = lesion & WM
buckner_lesions = np.where(lesion, 77, buckner)
# save map
utils.save_volume(buckner_lesions, None, None, path_result)
def evaluation(gt_dir,
seg_dir,
label_list,
mask_dir=None,
compute_score_whole_structure=False,
path_dice=None,
path_hausdorff=None,
path_hausdorff_99=None,
path_hausdorff_95=None,
path_mean_distance=None,
crop_margin_around_gt=10,
list_incorrect_labels=None,
list_correct_labels=None,
use_nearest_label=False,
recompute=True,
verbose=True):
"""This function computes Dice scores, as well as surface distances, between two sets of labels maps in gt_dir
(ground truth) and seg_dir (typically predictions). Labels maps in both folders are matched by sorting order.
The resulting scores are saved at the specified locations.
:param gt_dir: path of directory with gt label maps
:param seg_dir: path of directory with label maps to compare to gt_dir. Matched to gt label maps by sorting order.
:param label_list: list of label values for which to compute evaluation metrics. Can be a sequence, a 1d numpy
array, or the path to such array.
:param mask_dir: (optional) path of directory with masks of areas to ignore for each evaluated segmentation.
Matched to gt label maps by sorting order. Default is None, where nothing is masked.
:param compute_score_whole_structure: (optional) whether to also compute the selected scores for the whole segmented
structure (i.e. scores are computed for a single structure obtained by regrouping all non-zero values). If True, the
resulting scores are added as an extra row to the result matrices. Default is False.
:param path_dice: path where the resulting Dice will be writen as numpy array.
Default is None, where the array is not saved.
:param path_hausdorff: path where the resulting Hausdorff distances will be writen as numpy array (only if
compute_distances is True). Default is None, where the array is not saved.
:param path_hausdorff_99: same as for path_hausdorff but for the 99th percentile of the boundary distance.
:param path_hausdorff_95: same as for path_hausdorff but for the 95th percentile of the boundary distance.
:param path_mean_distance: path where the resulting mean distances will be writen as numpy array (only if
compute_distances is True). Default is None, where the array is not saved.
:param crop_margin_around_gt: (optional) margin by which to crop around the gt volumes, in order to copute the
scores more efficiently. If None, no cropping is performed.
:param list_incorrect_labels: (optional) this option enables to replace some label values in the maps in seg_dir by
other label values. Can be a list, a 1d numpy array, or the path to such an array.
The incorrect labels can then be replaced either by specified values, or by the nearest value (see below).
:param list_correct_labels: (optional) list of values to correct the labels specified in list_incorrect_labels.
Correct values must have the same order as their corresponding value in list_incorrect_labels.
:param use_nearest_label: (optional) whether to correct the incorrect lavel values with the nearest labels.
:param recompute: (optional) whether to recompute the already existing results. Default is True.
:param verbose: (optional) whether to print out info about the remaining number of cases.
"""
# check whether to recompute
compute_dice = not os.path.isfile(path_dice) if (path_dice
is not None) else True
compute_hausdorff = not os.path.isfile(path_hausdorff) if (
path_hausdorff is not None) else False
compute_hausdorff_99 = not os.path.isfile(path_hausdorff_99) if (
path_hausdorff_99 is not None) else False
compute_hausdorff_95 = not os.path.isfile(path_hausdorff_95) if (
path_hausdorff_95 is not None) else False
compute_mean_dist = not os.path.isfile(path_mean_distance) if (
path_mean_distance is not None) else False
compute_hd = [
compute_hausdorff, compute_hausdorff_99, compute_hausdorff_95
]
if compute_dice | any(compute_hd) | compute_mean_dist | recompute:
# get list label maps to compare
path_gt_labels = utils.list_images_in_folder(gt_dir)
path_segs = utils.list_images_in_folder(seg_dir)
path_gt_labels = utils.reformat_to_list(path_gt_labels,
length=len(path_segs))
if len(path_gt_labels) != len(path_segs):
print(
'gt and segmentation folders must have the same amount of label maps.'
)
if mask_dir is not None:
path_masks = utils.list_images_in_folder(mask_dir)
if len(path_masks) != len(path_segs):
print('not the same amount of masks and segmentations.')
else:
path_masks = [None] * len(path_segs)
# load labels list
label_list, _ = utils.get_list_labels(label_list=label_list,
FS_sort=True,
labels_dir=gt_dir)
n_labels = len(label_list)
max_label = np.max(label_list) + 1
# initialise result matrices
if compute_score_whole_structure:
max_dists = np.zeros((n_labels + 1, len(path_segs), 3))
mean_dists = np.zeros((n_labels + 1, len(path_segs)))
dice_coefs = np.zeros((n_labels + 1, len(path_segs)))
else:
max_dists = np.zeros((n_labels, len(path_segs), 3))
mean_dists = np.zeros((n_labels, len(path_segs)))
dice_coefs = np.zeros((n_labels, len(path_segs)))
# loop over segmentations
loop_info = utils.LoopInfo(len(path_segs),
10,
'evaluating',
print_time=True)
for idx, (path_gt, path_seg, path_mask) in enumerate(
zip(path_gt_labels, path_segs, path_masks)):
if verbose:
loop_info.update(idx)
# load gt labels and segmentation
gt_labels = utils.load_volume(path_gt, dtype='int')
seg = utils.load_volume(path_seg, dtype='int')
if path_mask is not None:
mask = utils.load_volume(path_mask, dtype='bool')
gt_labels[mask] = max_label
seg[mask] = max_label
# crop images
if crop_margin_around_gt is not None:
gt_labels, cropping = edit_volumes.crop_volume_around_region(
gt_labels, margin=crop_margin_around_gt)
seg = edit_volumes.crop_volume_with_idx(seg, cropping)
if list_incorrect_labels is not None:
seg = edit_volumes.correct_label_map(seg,
list_incorrect_labels,
list_correct_labels,
use_nearest_label)
# compute Dice scores
dice_coefs[:n_labels, idx] = fast_dice(gt_labels, seg, label_list)
# compute Dice scores for whole structures
if compute_score_whole_structure:
temp_gt = (gt_labels > 0) * 1
temp_seg = (seg > 0) * 1
dice_coefs[-1, idx] = dice(temp_gt, temp_seg)
else:
temp_gt = temp_seg = None
# compute average and Hausdorff distances
if any(compute_hd) | compute_mean_dist:
# compute unique label values
unique_gt_labels = np.unique(gt_labels)
unique_seg_labels = np.unique(seg)
# compute max/mean surface distances for all labels
for index, label in enumerate(label_list):
if (label in unique_gt_labels) & (label
in unique_seg_labels):
mask_gt = np.where(gt_labels == label, True, False)
mask_seg = np.where(seg == label, True, False)
tmp_max_dists, mean_dists[index,
idx] = surface_distances(
mask_gt, mask_seg,
[100, 99, 95])
max_dists[index, idx, :] = np.array(tmp_max_dists)
else:
mean_dists[index, idx] = max(gt_labels.shape)
max_dists[index, idx, :] = np.array(
[max(gt_labels.shape)] * 3)
# compute max/mean distances for whole structure
if compute_score_whole_structure:
tmp_max_dists, mean_dists[-1, idx] = surface_distances(
temp_gt, temp_seg, [100, 99, 95])
max_dists[-1, idx, :] = np.array(tmp_max_dists)
# write results
if path_dice is not None:
utils.mkdir(os.path.dirname(path_dice))
np.save(path_dice, dice_coefs)
if path_hausdorff is not None:
utils.mkdir(os.path.dirname(path_hausdorff))
np.save(path_hausdorff, max_dists[..., 0])
if path_hausdorff_99 is not None:
utils.mkdir(os.path.dirname(path_hausdorff_99))
np.save(path_hausdorff_99, max_dists[..., 1])
if path_hausdorff_95 is not None:
utils.mkdir(os.path.dirname(path_hausdorff_95))
np.save(path_hausdorff_95, max_dists[..., 2])
if path_mean_distance is not None:
utils.mkdir(os.path.dirname(path_mean_distance))
np.save(path_mean_distance, max_dists[..., 2])
def validate_training(image_dir,
gt_dir,
models_dir,
validation_main_dir,
segmentation_label_list,
evaluation_label_list=None,
dist_map=False,
step_eval=1,
aff_ref='FS',
sigma_smoothing=0,
keep_biggest_component=False,
padding=None,
cropping=None,
conv_size=3,
n_levels=5,
nb_conv_per_level=2,
unet_feat_count=24,
feat_multiplier=2,
activation='elu',
compute_distances=False,
recompute=True):
"""This function validates models saved at different epochs of the same training.
All models are assumed to be in the same folder.contained in models_dir.
The results of each model are saved in a subfolder in validation_main_dir.
:param image_dir: path of the folder with validation images.
:param gt_dir: path of the folder with ground truth label maps.
These are matched to the validation images by sorting order.
:param models_dir: path of the folder with the models to validate.
:param validation_main_dir: path of the folder where all the models validation subfolders will be saved.
:param segmentation_label_list: path of the numpy array containing all the segmentation labels used during training.
:param evaluation_label_list: (optional) label values to validate on. Must be a subset of the segmentation labels.
Can be a sequence, a 1d numpy array, or the path to a numpy 1d array. Default is the same as segmentation_label_list
:param dist_map: (optional) whether the input will contain distance maps channels (between each intenisty channels)
Default is False.
:param step_eval: (optional) If step_eval > 1 skips models when validating, by validating on models step_eval apart.
:param aff_ref: (optional) affine matrix with which the models were trained. Can be 'FS' (default), or 'identity.
:param sigma_smoothing: (optional) If not None, the posteriors are smoothed with a gaussian kernel of the specified
standard deviation.
:param keep_biggest_component: (optional) whether to only keep the biggest component in the predicted segmentation.
:param padding: (optional) pad the images to the specified shape before predicting the segmentation maps.
Can be an int, a sequence or a 1d numpy array.
:param cropping: (optional) whether to crop the input to smaller size while being run through the network.
The result is then given in the original image space. Can be an int, a sequence, or a 1d numpy array.
:param n_levels: (optional) number of level for the Unet. Default is 5.
:param nb_conv_per_level: (optional) number of convolutional layers per level. Default is 2.
:param conv_size: (optional) size of the convolution kernels. Default is 2.
:param unet_feat_count: (optional) number of feature maps for the first level. Default is 24.
:param feat_multiplier: (optional) multiply the number of feature by this nummber at each new level. Default is 1.
:param activation: (optional) activation function. Can be 'elu', 'relu'.
:param compute_distances: (optional) whether to compute the Haussdorf and mean surface distance.
:param recompute: (optional) whether to recompute result files even if they already exists."""
# create result folder
utils.mkdir(validation_main_dir)
# loop over models
list_models = utils.list_files(models_dir,
expr=['dice', 'h5'],
cond_type='and')[::step_eval]
loop_info = utils.LoopInfo(len(list_models), 1, 'validating', True)
for model_idx, path_model in enumerate(list_models):
# build names and create folders
model_val_dir = os.path.join(
validation_main_dir,
os.path.basename(path_model).replace('.h5', ''))
dice_path = os.path.join(model_val_dir, 'dice.npy')
utils.mkdir(model_val_dir)
if (not os.path.isfile(dice_path)) | recompute:
loop_info.update(model_idx)
predict(path_images=image_dir,
path_model=path_model,
segmentation_label_list=segmentation_label_list,
dist_map=dist_map,
path_segmentations=model_val_dir,
padding=padding,
cropping=cropping,
aff_ref=aff_ref,
sigma_smoothing=sigma_smoothing,
keep_biggest_component=keep_biggest_component,
conv_size=conv_size,
n_levels=n_levels,
nb_conv_per_level=nb_conv_per_level,
unet_feat_count=unet_feat_count,
feat_multiplier=feat_multiplier,
activation=activation,
gt_folder=gt_dir,
evaluation_label_list=evaluation_label_list,
compute_distances=compute_distances,
recompute=recompute,
verbose=False)
def predict(path_images,
path_model,
segmentation_label_list,
dist_map=False,
path_segmentations=None,
path_posteriors=None,
path_volumes=None,
segmentation_names_list=None,
padding=None,
cropping=None,
resample=None,
aff_ref='FS',
sigma_smoothing=0,
keep_biggest_component=False,
conv_size=3,
n_levels=5,
nb_conv_per_level=2,
unet_feat_count=24,
feat_multiplier=2,
activation='elu',
gt_folder=None,
evaluation_label_list=None,
compute_distances=False,
recompute=True,
verbose=True):
"""
This function uses trained models to segment images.
It is crucial that the inputs match the architecture parameters of the trained model.
:param path_images: path of the images to segment. Can be the path to a directory or the path to a single image.
:param path_model: path ot the trained model.
:param segmentation_label_list: List of labels for which to compute Dice scores. It should contain the same values
as the segmentation label list used for training the network.
Can be a sequence, a 1d numpy array, or the path to a numpy 1d array.
:param dist_map: (optional) whether the input will contain distance maps channels (between each intenisty channels)
Default is False.
:param path_segmentations: (optional) path where segmentations will be writen.
Should be a dir, if path_images is a dir, and afile if path_images is a file.
Should not be None, if path_posteriors is None.
:param path_posteriors: (optional) path where posteriors will be writen.
Should be a dir, if path_images is a dir, and afile if path_images is a file.
Should not be None, if path_segmentations is None.
:param path_volumes: (optional) path of a csv file where the soft volumes of all segmented regions will be writen.
The rows of the csv file correspond to subjects, and the columns correspond to segmentation labels.
The soft volume of a structure corresponds to the sum of its predicted probability map.
:param segmentation_names_list: (optional) List of names correponding to the names of the segmentation labels.
Only used when path_volumes is provided. Must be of the same size as segmentation_label_list. Can be given as a
list, a numpy array of strings, or the path to such a numpy array. Default is None.
:param padding: (optional) pad the images to the specified shape before predicting the segmentation maps.
Can be an int, a sequence or a 1d numpy array.
:param cropping: (optional) crop the images to the specified shape before predicting the segmentation maps.
If padding and cropping are specified, images are padded before being cropped.
Can be an int, a sequence or a 1d numpy array.
:param resample: (optional) resample the images to the specified resolution before predicting the segmentation maps.
Can be an int, a sequence or a 1d numpy array.
:param aff_ref: (optional) type of affine matrix of the images used for training. By default this is set to the
FreeSurfer orientation ('FS'), as it was the configuration in which SynthSeg was trained. However, the new models
are now trained on data aligned with identity vox2ras matrix, so you need to change aff_ref to 'identity'.
:param sigma_smoothing: (optional) If not None, the posteriors are smoothed with a gaussian kernel of the specified
standard deviation.
:param keep_biggest_component: (optional) whether to only keep the biggest component in the predicted segmentation.
:param conv_size: (optional) size of unet's convolution masks. Default is 3.
:param n_levels: (optional) number of levels for unet. Default is 5.
:param nb_conv_per_level: (optional) number of convolution layers per level. Default is 2.
:param unet_feat_count: (optional) number of features for the first layer of the unet. Default is 24.
:param feat_multiplier: (optional) multiplicative factor for the number of feature for each new level. Default is 2.
:param activation: (optional) activation function. Can be 'elu', 'relu'.
:param gt_folder: (optional) folder containing ground truth files for evaluation.
A numpy array containing all dice scores (labels in rows, subjects in columns) will be writen either at
segmentations_dir (if not None), or posteriors_dir.
:param evaluation_label_list: (optional) if gt_folder is True you can evaluate the Dice scores on a subset of the
segmentation labels, by providing another label list here. Can be a sequence, a 1d numpy array, or the path to a
numpy 1d array. Default is the same as segmentation_label_list.
:param recompute: (optional) whether to recompute segmentations that were already computed. This also applies to
Dice scores, if gt_folder is not None. Default is True.
:param verbose: (optional) whether to print out info about the remaining number of cases.
"""
# prepare output filepaths
images_to_segment, path_segmentations, path_posteriors, path_volumes, compute = \
prepare_output_files(path_images, path_segmentations, path_posteriors, path_volumes, recompute)
# get label and classes lists
label_list, n_neutral_labels = utils.get_list_labels(label_list=segmentation_label_list, FS_sort=True)
if evaluation_label_list is None:
evaluation_label_list = segmentation_label_list
# prepare volume file if needed
if path_volumes is not None:
if segmentation_names_list is not None:
csv_header = [[''] + utils.reformat_to_list(segmentation_names_list, load_as_numpy=True)]
csv_header += [[''] + [str(lab) for lab in label_list[1:]]]
else:
csv_header = [['subjects'] + [str(lab) for lab in label_list[1:]]]
with open(path_volumes, 'w') as csvFile:
writer = csv.writer(csvFile)
writer.writerows(csv_header)
csvFile.close()
# perform segmentation
net = None
previous_model_input_shape = None
loop_info = utils.LoopInfo(len(images_to_segment), 10, 'predicting', True)
for idx, (path_image, path_segmentation, path_posterior, tmp_compute) in enumerate(zip(images_to_segment,
path_segmentations,
path_posteriors,
compute)):
# compute segmentation only if needed
if tmp_compute:
# preprocess image and get information
image, aff, h, im_res, n_channels, n_dims, shape, pad_shape, crop_idx = \
preprocess_image(path_image, n_levels, cropping, padding, aff_ref=aff_ref, dist_map=dist_map)
model_input_shape = list(image.shape[1:])
# prepare net for first image or if input's size has changed
if (net is None) | (previous_model_input_shape != model_input_shape):
# check for image size compatibility
if (net is not None) & (previous_model_input_shape != model_input_shape) & verbose:
print('image of different shape as previous ones, redefining network')
previous_model_input_shape = model_input_shape
# build network
net = build_model(path_model, model_input_shape, resample, im_res, n_levels, len(label_list), conv_size,
nb_conv_per_level, unet_feat_count, feat_multiplier, activation, sigma_smoothing)
if verbose:
loop_info.update(idx)
# predict posteriors
prediction_patch = net.predict(image)
# get posteriors and segmentation
seg, posteriors = postprocess(prediction_patch, pad_shape, shape, crop_idx, n_dims, label_list,
keep_biggest_component, aff, aff_ref=aff_ref,
keep_biggest_of_each_group=keep_biggest_component,
n_neutral_labels=n_neutral_labels)
# write results to disk
if path_segmentation is not None:
utils.save_volume(seg.astype('int'), aff, h, path_segmentation)
if path_posterior is not None:
if n_channels > 1:
posteriors = utils.add_axis(posteriors, axis=[0, -1])
utils.save_volume(posteriors.astype('float'), aff, h, path_posterior)
else:
if path_volumes is not None:
posteriors, _, _, _, _, _, im_res = utils.get_volume_info(path_posterior, True, aff_ref=np.eye(4))
else:
posteriors = im_res = None
# compute volumes
if path_volumes is not None:
volumes = np.sum(posteriors[..., 1:], axis=tuple(range(0, len(posteriors.shape) - 1)))
volumes = np.around(volumes * np.prod(im_res), 3)
row = [os.path.basename(path_image).replace('.nii.gz', '')] + [str(vol) for vol in volumes]
with open(path_volumes, 'a') as csvFile:
writer = csv.writer(csvFile)
writer.writerow(row)
csvFile.close()
# evaluate
if gt_folder is not None:
# find path evaluation folder
path_first_result = path_segmentations[0] if (path_segmentations[0] is not None) else path_posteriors[0]
eval_folder = os.path.dirname(path_first_result)
# compute evaluation metrics
evaluate.dice_evaluation(gt_folder,
eval_folder,
evaluation_label_list,
compute_distances=compute_distances,
compute_score_whole_structure=False,
path_dice=os.path.join(eval_folder, 'dice.npy'),
path_hausdorff=os.path.join(eval_folder, 'hausdorff.npy'),
path_mean_distance=os.path.join(eval_folder, 'mean_distance.npy'),
recompute=recompute,
verbose=verbose)
def dice_evaluation(gt_dir,
seg_dir,
label_list,
compute_distances=False,
compute_score_whole_structure=False,
path_dice=None,
path_hausdorff=None,
path_mean_distance=None,
crop_margin_around_gt=10,
recompute=True,
verbose=True):
"""This function computes Dice scores between two sets of labels maps in gt_dir (ground truth) and seg_dir
(typically predictions). Labels maps in both folders are matched by sorting order.
:param gt_dir: path of directory with gt label maps
:param seg_dir: path of directory with label maps to compare to gt_dir. Matched to gt label maps by sorting order.
:param label_list: list of label values for which to compute evaluation metrics. Can be a sequence, a 1d numpy
array, or the path to such array.
:param compute_distances: (optional) whether to compute distances (Hausdorff and mean distance) between the surfaces
of GT and predicted labels. Default is False.
:param compute_score_whole_structure: (optional) whether to also compute the selected scores for the whole segmented
structure (i.e. scores are computed for a single structure obtained by regrouping all non-zero values). If True, the
resulting scores are added as an extra row to the result matrices. Default is False.
:param path_dice: path where the resulting Dice will be writen as numpy array.
Default is None, where the array is not saved.
:param path_hausdorff: path where the resulting Hausdorff distances will be writen as numpy array (only if
compute_distances is True). Default is None, where the array is not saved.
:param path_mean_distance: path where the resulting mean distances will be writen as numpy array (only if
compute_distances is True). Default is None, where the array is not saved.
:param crop_margin_around_gt: (optional) margin by which to crop around the gt volumes, in order to copute the
scores more efficiently. If None, no cropping is performed.
:param recompute: (optional) whether to recompute the already existing results. Default is True.
:param verbose: (optional) whether to print out info about the remaining number of cases.
:return: numpy array containing all Dice scores (labels in rows, subjects in columns). Also returns numpy arrays
with the same structures for Hausdorff and mean distances if compute_distances is True.
"""
# check whether to recompute
compute_dice = not os.path.isfile(path_dice) if (path_dice
is not None) else True
if compute_distances:
compute_hausdorff = not os.path.isfile(path_hausdorff) if (
path_hausdorff is not None) else True
compute_mean_dist = not os.path.isfile(path_mean_distance) if (
path_mean_distance is not None) else True
else:
compute_hausdorff = compute_mean_dist = False
if compute_dice | compute_hausdorff | compute_mean_dist | recompute:
# get list label maps to compare
path_gt_labels = utils.list_images_in_folder(gt_dir)
path_segs = utils.list_images_in_folder(seg_dir)
if len(path_gt_labels) != len(path_segs):
print(
'gt and segmentation folders must have the same amount of label maps.'
)
# load labels list
label_list, _ = utils.get_list_labels(label_list=label_list,
FS_sort=True,
labels_dir=gt_dir)
n_labels = len(label_list)
# initialise result matrices
if compute_score_whole_structure:
max_dists = np.zeros((n_labels + 1, len(path_segs)))
mean_dists = np.zeros((n_labels + 1, len(path_segs)))
dice_coefs = np.zeros((n_labels + 1, len(path_segs)))
else:
max_dists = np.zeros((n_labels, len(path_segs)))
mean_dists = np.zeros((n_labels, len(path_segs)))
dice_coefs = np.zeros((n_labels, len(path_segs)))
# loop over segmentations
loop_info = utils.LoopInfo(len(path_segs), 10, 'evaluating')
for idx, (path_gt,
path_seg) in enumerate(zip(path_gt_labels, path_segs)):
if verbose:
loop_info.update(idx)
# load gt labels and segmentation
gt_labels = utils.load_volume(path_gt, dtype='int')
seg = utils.load_volume(path_seg, dtype='int')
# crop images
if crop_margin_around_gt is not None:
gt_labels, cropping = edit_volumes.crop_volume_around_region(
gt_labels, margin=crop_margin_around_gt)
seg = edit_volumes.crop_volume_with_idx(seg, cropping)
# compute Dice scores
dice_coefs[:n_labels, idx] = fast_dice(gt_labels, seg, label_list)
# compute Dice scores for whole structures
if compute_score_whole_structure:
temp_gt = (gt_labels > 0) * 1
temp_seg = (seg > 0) * 1
dice_coefs[-1, idx] = dice(temp_gt, temp_seg)
else:
temp_gt = temp_seg = None
# compute average and Hausdorff distances
if compute_distances:
# compute unique label values
unique_gt_labels = np.unique(gt_labels)
unique_seg_labels = np.unique(seg)
# compute max/mean surface distances for all labels
for index, label in enumerate(label_list):
if (label in unique_gt_labels) & (label
in unique_seg_labels):
mask_gt = np.where(gt_labels == label, True, False)
mask_seg = np.where(seg == label, True, False)
max_dists[index,
idx], mean_dists[index,
idx] = surface_distances(
mask_gt, mask_seg)
else:
max_dists[index, idx] = max(gt_labels.shape)
mean_dists[index, idx] = max(gt_labels.shape)
# compute max/mean distances for whole structure
if compute_score_whole_structure:
max_dists[-1, idx], mean_dists[-1,
idx] = surface_distances(
temp_gt, temp_seg)
# write results
if path_dice is not None:
utils.mkdir(os.path.dirname(path_dice))
np.save(path_dice, dice_coefs)
if compute_distances and path_hausdorff is not None:
utils.mkdir(os.path.dirname(path_hausdorff))
np.save(path_hausdorff, max_dists)
if compute_distances and path_mean_distance is not None:
utils.mkdir(os.path.dirname(path_mean_distance))
np.save(path_mean_distance, mean_dists)
else:
dice_coefs = np.load(path_dice)
if compute_distances:
max_dists = np.load(path_hausdorff)
mean_dists = np.load(path_mean_distance)
else:
max_dists = mean_dists = None
if compute_distances:
return dice_coefs, max_dists, mean_dists
else:
return dice_coefs, None, None
def prepare_hippo_training_atlases(labels_dir,
result_dir,
image_dir=None,
image_result_dir=None,
smooth=True,
crop_margin=50,
recompute=True):
"""This function prepares training label maps from CobraLab. It first crops each atlas around the right and left
hippocampi, with a margin. It then equalises the shape of these atlases by croppping them to the size of the
smallest hippocampus. Finally it realigns the obtained atlases to FS orientation axes.
:param labels_dir: path of directory with label maps to prepare
:param result_dir: path of directory where prepared atlases will be writen
:param image_dir: (optional) path of directory with images corresponding to the label maps to prepare.
This can be sued to prepare a dataset of real images for supervised training.
:param image_result_dir: (optional) path of directory where images corresponding to prepared atlases will be writen
:param smooth: (optional) whether to smooth the final cropped label maps
:param crop_margin: (optional) margin to add around hippocampi when cropping
:param recompute: (optional) whether to recompute result files even if they already exists"""
# create results dir
utils.mkdir(result_dir)
tmp_result_dir = os.path.join(result_dir, 'first_cropping')
utils.mkdir(tmp_result_dir)
if image_dir is not None:
assert image_result_dir is not None, 'image_result_dir should not be None if image_dir is specified'
utils.mkdir(image_result_dir)
tmp_image_result_dir = os.path.join(image_result_dir, 'first_cropping')
utils.mkdir(tmp_image_result_dir)
else:
tmp_image_result_dir = None
# list labels and images
labels_paths = utils.list_images_in_folder(labels_dir)
if image_dir is not None:
path_images = utils.list_images_in_folder(image_dir)
else:
path_images = [None] * len(labels_paths)
# crop all atlases around hippo
print('\ncropping around hippo')
shape_array = np.zeros((len(labels_paths)*2, 3))
loop_info = utils.LoopInfo(len(labels_paths), 1)
for idx, (path_label, path_image) in enumerate(zip(labels_paths, path_images)):
loop_info.update(idx)
# crop left hippo first
path_label_first_crop_l = os.path.join(tmp_result_dir,
os.path.basename(path_label).replace('.nii', '_left.nii'))
lab, aff, h = utils.load_volume(path_label, im_only=False)
lab_l, croppping_idx, aff_l = edit_volumes.crop_volume_around_region(lab, crop_margin,
list(range(20101, 20109)), aff=aff)
if (not os.path.exists(path_label_first_crop_l)) | recompute:
utils.save_volume(lab_l, aff_l, h, path_label_first_crop_l)
else:
lab_l = utils.load_volume(path_label_first_crop_l)
if path_image is not None:
path_image_first_crop_l = os.path.join(tmp_image_result_dir,
os.path.basename(path_image).replace('.nii', '_left.nii'))
if (not os.path.exists(path_image_first_crop_l)) | recompute:
im, aff, h = utils.load_volume(path_image, im_only=False)
im, aff = edit_volumes.crop_volume_with_idx(im, croppping_idx, aff)
utils.save_volume(im, aff, h, path_image_first_crop_l)
shape_array[2*idx, :] = np.array(lab_l.shape)
# crop right hippo and flip them
path_label_first_crop_r = os.path.join(tmp_result_dir,
os.path.basename(path_label).replace('.nii', '_right_flipped.nii'))
lab, aff, h = utils.load_volume(path_label, im_only=False)
lab_r, croppping_idx, aff_r = edit_volumes.crop_volume_around_region(lab, crop_margin,
list(range(20001, 20009)), aff=aff)
if (not os.path.exists(path_label_first_crop_r)) | recompute:
lab_r = edit_volumes.flip_volume(lab_r, direction='rl', aff=aff_r)
utils.save_volume(lab_r, aff_r, h, path_label_first_crop_r)
else:
lab_r = utils.load_volume(path_label_first_crop_r)
if path_image is not None:
path_image_first_crop_r = os.path.join(tmp_image_result_dir,
os.path.basename(path_image).replace('.nii', '_right.nii'))
if (not os.path.exists(path_image_first_crop_r)) | recompute:
im, aff, h = utils.load_volume(path_image, im_only=False)
im, aff = edit_volumes.crop_volume_with_idx(im, croppping_idx, aff)
im = edit_volumes.flip_volume(im, direction='rl', aff=aff)
utils.save_volume(im, aff, h, path_image_first_crop_r)
shape_array[2*idx+1, :] = np.array(lab_r.shape)
# list croppped files
path_labels_first_cropped = utils.list_images_in_folder(tmp_result_dir)
if tmp_image_result_dir is not None:
path_images_first_cropped = utils.list_images_in_folder(tmp_image_result_dir)
else:
path_images_first_cropped = [None] * len(path_labels_first_cropped)
# crop all label maps to same size
print('\nequalising shapes')
new_shape = np.min(shape_array, axis=0).astype('int32')
loop_info = utils.LoopInfo(len(path_labels_first_cropped), 1)
for i, (path_label, path_image) in enumerate(zip(path_labels_first_cropped, path_images_first_cropped)):
loop_info.update(i)
# get cropping indices
path_lab_cropped = os.path.join(result_dir, os.path.basename(path_label))
lab, aff, h = utils.load_volume(path_label, im_only=False)
lab_shape = lab.shape
min_cropping = np.array([np.maximum(int((lab_shape[i]-new_shape[i])/2), 0) for i in range(3)])
max_cropping = np.array([min_cropping[i] + new_shape[i] for i in range(3)])
# crop labels and realign on adni format
if (not os.path.exists(path_lab_cropped)) | recompute:
lab, aff = edit_volumes.crop_volume_with_idx(lab, np.concatenate([min_cropping, max_cropping]), aff)
# realign on adni format
lab = np.flip(lab, axis=2)
aff[0:3, 0:3] = np.array([[-0.6, 0, 0], [0, 0, -0.6], [0, -0.6, 0]])
utils.save_volume(lab, aff, h, path_lab_cropped)
# crop image and realign on adni format
if path_image is not None:
path_im_cropped = os.path.join(image_result_dir, os.path.basename(path_image))
if (not os.path.exists(path_im_cropped)) | recompute:
im, aff, h = utils.load_volume(path_image, im_only=False)
im, aff = edit_volumes.crop_volume_with_idx(im, np.concatenate([min_cropping, max_cropping]), aff)
im = np.flip(im, axis=2)
aff[0:3, 0:3] = np.array([[-0.6, 0, 0], [0, 0, -0.6], [0, -0.6, 0]])
im = edit_volumes.mask_volume(im, lab)
utils.save_volume(im, aff, h, path_im_cropped)
# correct all labels to left values
print('\ncorrecting labels')
list_incorrect_labels = [77, 80, 251, 252, 253, 254, 255, 29, 41, 42, 43, 44, 46, 47, 49, 50, 51, 52, 54, 58, 60,
61, 62, 63, 7012, 20001, 20002, 20004, 20005, 20006, 20007, 20008]
list_correct_labels = [2, 3, 2, 2, 2, 2, 2, 2, 2, 3, 4, 5, 7, 8, 10, 11, 12, 13, 18, 26, 28, 2, 30, 31, 20108,
20101, 20102, 20104, 20105, 20106, 20107, 20108]
edit_volumes.correct_labels_in_dir(result_dir, list_incorrect_labels, list_correct_labels, result_dir)
# smooth labels
if smooth:
print('\nsmoothing labels')
edit_volumes.smooth_labels_in_dir(result_dir, result_dir)
def sample_intensity_stats_from_single_dataset(image_dir, labels_dir, labels_list, classes_list=None, max_channel=3,
rescale=True):
"""This function aims at estimating the intensity distributions of K different structure types from a set of images.
The distribution of each structure type is modelled as a Gaussian, parametrised by a mean and a standard deviation.
Because the intensity distribution of structures can vary accross images, we additionally use Gausian priors for the
parameters of each Gaussian distribution. Therefore, the intensity distribution of each structure type is described
by 4 parameters: a mean/std for the mean intensity, and a mean/std for the std deviation.
This function uses a set of images along with corresponding segmentations to estimate the 4*K parameters.
Structures can share the same statistics by being regrouped into classes of similar structure types.
Images can be multi-modal (n_channels), in which case different statistics are estimated for each modality.
:param image_dir: path of directory with images to estimate the intensity distribution
:param labels_dir: path of directory with segmentation of input images.
They are matched with images by sorting order.
:param labels_list: list of labels for which to evaluate mean and std intensity.
Can be a sequence, a 1d numpy array, or the path to a 1d numpy array.
:param classes_list: (optional) enables to regroup structures into classes of similar intensity statistics.
Intenstites associated to regrouped labels will thus contribute to the same Gaussian during statistics estimation.
Can be a sequence, a 1d numpy array, or the path to a 1d numpy array.
It should have the same length as labels_list, and contain values between 0 and K-1, where K is the total number of
classes. Default is all labels have different classes (K=len(labels_list)).
:param max_channel: (optional) maximum number of channels to consider if the data is multispectral. Default is 3.
:param rescale: (optional) whether to rescale images between 0 and 255 before intensity estimation
:return: 2 numpy arrays of size (2*n_channels, K), one with the evaluated means/std for the mean
intensity, and one for the mean/std for the standard deviation.
Each block of two rows correspond to a different modality (channel). For each block of two rows, the first row
represents the mean, and the second represents the std.
"""
# list files
path_images = utils.list_images_in_folder(image_dir)
path_labels = utils.list_images_in_folder(labels_dir)
assert len(path_images) == len(path_labels), 'image and labels folders do not have the same number of files'
# reformat list labels and classes
labels_list = np.array(utils.reformat_to_list(labels_list, load_as_numpy=True, dtype='int'))
if classes_list is not None:
classes_list = np.array(utils.reformat_to_list(classes_list, load_as_numpy=True, dtype='int'))
else:
classes_list = np.arange(labels_list.shape[0])
assert len(classes_list) == len(labels_list), 'labels and classes lists should have the same length'
# get unique classes
unique_classes, unique_indices = np.unique(classes_list, return_index=True)
n_classes = len(unique_classes)
if not np.array_equal(unique_classes, np.arange(n_classes)):
raise ValueError('classes_list should only contain values between 0 and K-1, '
'where K is the total number of classes. Here K = %d' % n_classes)
# initialise result arrays
n_dims, n_channels = utils.get_dims(utils.load_volume(path_images[0]).shape, max_channels=max_channel)
means = np.zeros((len(path_images), n_classes, n_channels))
stds = np.zeros((len(path_images), n_classes, n_channels))
# loop over images
loop_info = utils.LoopInfo(len(path_images), 10, 'estimating', print_time=True)
for idx, (path_im, path_la) in enumerate(zip(path_images, path_labels)):
loop_info.update(idx)
# load image and label map
image = utils.load_volume(path_im)
la = utils.load_volume(path_la)
if n_channels == 1:
image = utils.add_axis(image, -1)
# loop over channels
for channel in range(n_channels):
im = image[..., channel]
if rescale:
im = edit_volumes.rescale_volume(im)
stats = sample_intensity_stats_from_image(im, la, labels_list, classes_list=classes_list)
means[idx, :, channel] = stats[0, :]
stds[idx, :, channel] = stats[1, :]
# compute prior parameters for mean/std
mean_means = np.mean(means, axis=0)
std_means = np.std(means, axis=0)
mean_stds = np.mean(stds, axis=0)
std_stds = np.std(stds, axis=0)
# regroup prior parameters in two different arrays: one for the mean and one for the std
prior_means = np.zeros((2 * n_channels, n_classes))
prior_stds = np.zeros((2 * n_channels, n_classes))
for channel in range(n_channels):
prior_means[2 * channel, :] = mean_means[:, channel]
prior_means[2 * channel + 1, :] = std_means[:, channel]
prior_stds[2 * channel, :] = mean_stds[:, channel]
prior_stds[2 * channel + 1, :] = std_stds[:, channel]
return prior_means, prior_stds
def predict(path_images,
path_segmentations,
path_model,
segmentation_labels,
n_neutral_labels=None,
path_posteriors=None,
path_resampled=None,
path_volumes=None,
segmentation_label_names=None,
padding=None,
cropping=None,
target_res=1.,
gradients=False,
flip=True,
topology_classes=None,
sigma_smoothing=0.5,
keep_biggest_component=True,
conv_size=3,
n_levels=5,
nb_conv_per_level=2,
unet_feat_count=24,
feat_multiplier=2,
activation='elu',
gt_folder=None,
evaluation_labels=None,
mask_folder=None,
list_incorrect_labels=None,
list_correct_labels=None,
compute_distances=False,
recompute=True,
verbose=True):
"""
This function uses trained models to segment images.
It is crucial that the inputs match the architecture parameters of the trained model.
:param path_images: path of the images to segment. Can be the path to a directory or the path to a single image.
:param path_segmentations: path where segmentations will be writen.
Should be a dir, if path_images is a dir, and a file if path_images is a file.
:param path_model: path ot the trained model.
:param segmentation_labels: List of labels for which to compute Dice scores. It should be the same list as the
segmentation_labels used in training.
:param n_neutral_labels: (optional) if the label maps contain some right/left specific labels and if test-time
flipping is applied (see parameter 'flip'), please provide the number of non-sided labels (including background).
It should be the same value as for training. Default is None.
:param path_posteriors: (optional) path where posteriors will be writen.
Should be a dir, if path_images is a dir, and a file if path_images is a file.
:param path_resampled: (optional) path where images resampled to 1mm isotropic will be writen.
We emphasise that images are resampled as soon as the resolution in one of the axes is not in the range [0.9; 1.1].
Should be a dir, if path_images is a dir, and a file if path_images is a file. Default is None, where resampled
images are not saved.
:param path_volumes: (optional) path of a csv file where the soft volumes of all segmented regions will be writen.
The rows of the csv file correspond to subjects, and the columns correspond to segmentation labels.
The soft volume of a structure corresponds to the sum of its predicted probability map.
:param segmentation_label_names: (optional) List of names correponding to the names of the segmentation labels.
Only used when path_volumes is provided. Must be of the same size as segmentation_labels. Can be given as a
list, a numpy array of strings, or the path to such a numpy array. Default is None.
:param padding: (optional) pad the images to the specified shape before predicting the segmentation maps.
Can be an int, a sequence or a 1d numpy array.
:param cropping: (optional) crop the images to the specified shape before predicting the segmentation maps.
If padding and cropping are specified, images are padded before being cropped.
Can be an int, a sequence or a 1d numpy array.
:param target_res: (optional) target resolution at which the network operates (and thus resolution of the output
segmentations). This must match the resolution of the training data ! target_res is used to automatically resampled
the images with resolutions outside [target_res-0.05, target_res+0.05].
Can be a sequence, a 1d numpy array. Set to None to disable the automatic resampling. Default is 1mm.
:param flip: (optional) whether to perform test-time augmentation, where the input image is segmented along with
a right/left flipped version on it. If set to True (default), be careful because this requires more memory.
:param topology_classes: List of classes corresponding to all segmentation labels, in order to group them into
classes, for each of which we will operate a smooth version of biggest connected component.
Can be a sequence, a 1d numpy array, or the path to a numpy 1d array in the same order as segmentation_labels.
Default is None, where no topological analysis is performed.
:param sigma_smoothing: (optional) If not None, the posteriors are smoothed with a gaussian kernel of the specified
standard deviation.
:param keep_biggest_component: (optional) whether to only keep the biggest component in the predicted segmentation.
This is applied independently of topology_classes, and it is applied to the whole segmentation
:param conv_size: (optional) size of unet's convolution masks. Default is 3.
:param n_levels: (optional) number of levels for unet. Default is 5.
:param nb_conv_per_level: (optional) number of convolution layers per level. Default is 2.
:param unet_feat_count: (optional) number of features for the first layer of the unet. Default is 24.
:param feat_multiplier: (optional) multiplicative factor for the number of feature for each new level. Default is 2.
:param activation: (optional) activation function. Can be 'elu', 'relu'.
:param gt_folder: (optional) path of the ground truth label maps corresponding to the input images. Should be a dir,
if path_images is a dir, or a file if path_images is a file.
Providing a gt_folder will trigger a Dice evaluation, where scores will be writen along with the path_segmentations.
Specifically, the scores are contained in a numpy array, where labels are in rows, and subjects in columns.
:param evaluation_labels: (optional) if gt_folder is True you can evaluate the Dice scores on a subset of the
segmentation labels, by providing another label list here. Can be a sequence, a 1d numpy array, or the path to a
numpy 1d array. Default is np.unique(segmentation_labels).
:param mask_folder: (optional) path of masks that will be used to mask out some parts of the obtained segmentations
during the evaluation. Default is None, where nothing is masked.
:param list_incorrect_labels: (optional) this option enables to replace some label values in the obtained
segmentations by other label values. Can be a list, a 1d numpy array, or the path to such an array.
:param list_correct_labels: (optional) list of values to correct the labels specified in list_incorrect_labels.
Correct values must have the same order as their corresponding value in list_incorrect_labels.
:param compute_distances: (optional) whether to add Hausdorff and mean surface distance evaluations to the default
Dice evaluation. Default is True.
:param recompute: (optional) whether to recompute segmentations that were already computed. This also applies to
Dice scores, if gt_folder is not None. Default is True.
:param verbose: (optional) whether to print out info about the remaining number of cases.
"""
# prepare input/output filepaths
path_images, path_segmentations, path_posteriors, path_resampled, path_volumes, compute = \
prepare_output_files(path_images, path_segmentations, path_posteriors, path_resampled, path_volumes, recompute)
# get label list
segmentation_labels, _ = utils.get_list_labels(
label_list=segmentation_labels)
n_labels = len(segmentation_labels)
# get unique label values, and build correspondance table between contralateral structures if necessary
if (n_neutral_labels is not None) & flip:
n_sided_labels = int((n_labels - n_neutral_labels) / 2)
lr_corresp = np.stack([
segmentation_labels[n_neutral_labels:n_neutral_labels +
n_sided_labels],
segmentation_labels[n_neutral_labels + n_sided_labels:]
])
segmentation_labels, indices = np.unique(segmentation_labels,
return_index=True)
lr_corresp_unique, lr_corresp_indices = np.unique(lr_corresp[0, :],
return_index=True)
lr_corresp_unique = np.stack(
[lr_corresp_unique, lr_corresp[1, lr_corresp_indices]])
lr_corresp_unique = lr_corresp_unique[:, 1:] if not np.all(
lr_corresp_unique[:, 0]) else lr_corresp_unique
lr_indices = np.zeros_like(lr_corresp_unique)
for i in range(lr_corresp_unique.shape[0]):
for j, lab in enumerate(lr_corresp_unique[i]):
lr_indices[i, j] = np.where(segmentation_labels == lab)[0]
else:
segmentation_labels, indices = np.unique(segmentation_labels,
return_index=True)
lr_indices = None
# prepare topology classes
if topology_classes is not None:
topology_classes = utils.load_array_if_path(
topology_classes, load_as_numpy=True)[indices]
# prepare volume file if needed
if path_volumes is not None:
if segmentation_label_names is not None:
segmentation_label_names = utils.load_array_if_path(
segmentation_label_names)[indices]
csv_header = [[''] + segmentation_label_names[1:].tolist()]
csv_header += [[''] +
[str(lab) for lab in segmentation_labels[1:]]]
else:
csv_header = [['subjects'] +
[str(lab) for lab in segmentation_labels[1:]]]
with open(path_volumes, 'w') as csvFile:
writer = csv.writer(csvFile)
writer.writerows(csv_header)
csvFile.close()
# build network
_, _, n_dims, n_channels, _, _ = utils.get_volume_info(path_images[0])
model_input_shape = [None] * n_dims + [n_channels]
net = build_model(path_model, model_input_shape, n_levels,
len(segmentation_labels), conv_size, nb_conv_per_level,
unet_feat_count, feat_multiplier, activation,
sigma_smoothing, gradients)
# perform segmentation
loop_info = utils.LoopInfo(len(path_images), 10, 'predicting', True)
for idx, (path_image, path_segmentation, path_posterior, path_resample, tmp_compute) in \
enumerate(zip(path_images, path_segmentations, path_posteriors, path_resampled, compute)):
# compute segmentation only if needed
if tmp_compute:
if verbose:
loop_info.update(idx)
# preprocessing
image, aff, h, im_res, _, _, shape, pad_shape, crop_idx, im_flipped = \
preprocess_image(path_image, n_levels, target_res, cropping, padding, flip, path_resample)
# prediction
prediction_patch = net.predict(image)
prediction_patch_flip = net.predict(im_flipped) if flip else None
# postprocessing
seg, posteriors = postprocess(
prediction_patch,
pad_shape,
shape,
crop_idx,
n_dims,
segmentation_labels,
lr_indices,
keep_biggest_component,
aff,
topology_classes=topology_classes,
post_patch_flip=prediction_patch_flip)
# write results to disk
if path_segmentation is not None:
utils.save_volume(seg,
aff,
h,
path_segmentation,
dtype='int32')
if path_posterior is not None:
if n_channels > 1:
posteriors = utils.add_axis(posteriors, axis=[0, -1])
utils.save_volume(posteriors,
aff,
h,
path_posterior,
dtype='float32')
else:
if path_volumes is not None:
posteriors, _, _, _, _, _, im_res = utils.get_volume_info(
path_posterior, True, aff_ref=np.eye(4))
else:
posteriors = im_res = None
# compute volumes
if path_volumes is not None:
volumes = np.sum(posteriors[..., 1:],
axis=tuple(range(0,
len(posteriors.shape) - 1)))
volumes = np.around(volumes * np.prod(im_res), 3)
row = [os.path.basename(path_image).replace('.nii.gz', '')
] + [str(vol) for vol in volumes]
with open(path_volumes, 'a') as csvFile:
writer = csv.writer(csvFile)
writer.writerow(row)
csvFile.close()
# evaluate
if gt_folder is not None:
# find path where segmentations are saved evaluation folder, and get labels on which to evaluate
eval_folder = os.path.dirname(path_segmentations[0])
if evaluation_labels is None:
evaluation_labels = segmentation_labels
# set path of result arrays for surface distance if necessary
if compute_distances:
path_hausdorff = os.path.join(eval_folder, 'hausdorff.npy')
path_hausdorff_99 = os.path.join(eval_folder, 'hausdorff_99.npy')
path_hausdorff_95 = os.path.join(eval_folder, 'hausdorff_95.npy')
path_mean_distance = os.path.join(eval_folder, 'mean_distance.npy')
else:
path_hausdorff = path_hausdorff_99 = path_hausdorff_95 = path_mean_distance = None
# compute evaluation metrics
evaluate.evaluation(gt_folder,
eval_folder,
evaluation_labels,
mask_dir=mask_folder,
path_dice=os.path.join(eval_folder, 'dice.npy'),
path_hausdorff=path_hausdorff,
path_hausdorff_99=path_hausdorff_99,
path_hausdorff_95=path_hausdorff_95,
path_mean_distance=path_mean_distance,
list_incorrect_labels=list_incorrect_labels,
list_correct_labels=list_correct_labels,
recompute=recompute,
verbose=verbose)
def validate_training(image_dir,
gt_dir,
models_dir,
validation_main_dir,
segmentation_labels,
n_neutral_labels=None,
evaluation_labels=None,
step_eval=1,
padding=None,
cropping=None,
target_res=1.,
gradients=False,
flip=False,
topology_classes=None,
sigma_smoothing=0,
keep_biggest_component=False,
conv_size=3,
n_levels=5,
nb_conv_per_level=2,
unet_feat_count=24,
feat_multiplier=2,
activation='elu',
mask_dir=None,
compute_distances=False,
recompute=True):
"""This function validates models saved at different epochs of the same training.
All models are assumed to be in the same folder.contained in models_dir.
The results of each model are saved in a subfolder in validation_main_dir.
:param image_dir: path of the folder with validation images.
:param gt_dir: path of the folder with ground truth label maps.
These are matched to the validation images by sorting order.
:param models_dir: path of the folder with the models to validate.
:param validation_main_dir: path of the folder where all the models validation subfolders will be saved.
:param segmentation_labels: path of the numpy array containing all the segmentation labels used during training.
:param n_neutral_labels: (optional) value of n_neutral_labels used during training. Used only if flip is True.
:param evaluation_labels: (optional) label values to validate on. Must be a subset of the segmentation labels.
Can be a sequence, a 1d numpy array, or the path to a numpy 1d array. Default is the same as segmentation_label_list
:param step_eval: (optional) If step_eval > 1 skips models when validating, by validating on models step_eval apart.
:param padding: (optional) pad the images to the specified shape before predicting the segmentation maps.
Can be an int, a sequence or a 1d numpy array.
:param cropping: (optional) whether to crop the input to smaller size while being run through the network.
The result is then given in the original image space. Can be an int, a sequence, or a 1d numpy array.
:param target_res: (optional) target resolution at which the network operates (and thus resolution of the output
segmentations). This must match the resolution of the training data ! target_res is used to automatically resampled
the images with resolutions outside [target_res-0.05, target_res+0.05].
Can be a sequence, a 1d numpy array. Set to None to disable the automatic resampling. Default is 1mm.
:param flip: (optional) whether to perform test-time augmentation, where the input image is segmented along with
a right/left flipped version on it. If set to True (default), be careful because this requires more memory.
:param topology_classes: List of classes corresponding to all segmentation labels, in order to group them into
classes, for each of which we will operate a smooth version of biggest connected component.
Can be a sequence, a 1d numpy array, or the path to a numpy 1d array in the same order as segmentation_label_list.
Default is None, where no topological analysis is performed.
:param sigma_smoothing: (optional) If not None, the posteriors are smoothed with a gaussian kernel of the specified
standard deviation.
:param keep_biggest_component: (optional) whether to only keep the biggest component in the predicted segmentation.
This is applied independently of topology_classes, and it is applied to the whole segmentation
:param conv_size: (optional) size of the convolution kernels. Default is 2.
:param n_levels: (optional) number of level for the Unet. Default is 5.
:param nb_conv_per_level: (optional) number of convolutional layers per level. Default is 2.
:param unet_feat_count: (optional) number of feature maps for the first level. Default is 24.
:param feat_multiplier: (optional) multiply the number of feature by this nummber at each new level. Default is 1.
:param activation: (optional) activation function. Can be 'elu', 'relu'.
:param mask_dir: (optional) path of masks that will be used to mask out some parts of the obtained segmentations
during the evaluation. Default is None, where nothing is masked.
:param compute_distances: (optional) whether to add Hausdorff and mean surface distance evaluations to the default
Dice evaluation. Default is True.
:param recompute: (optional) whether to recompute result files even if they already exists."""
# create result folder
utils.mkdir(validation_main_dir)
# loop over models
list_models = utils.list_files(models_dir,
expr=['dice', '.h5'],
cond_type='and')[::step_eval]
# list_models = [p for p in list_models if int(os.path.basename(p)[-6:-3]) % 10 == 0]
loop_info = utils.LoopInfo(len(list_models), 1, 'validating', True)
for model_idx, path_model in enumerate(list_models):
# build names and create folders
model_val_dir = os.path.join(
validation_main_dir,
os.path.basename(path_model).replace('.h5', ''))
dice_path = os.path.join(model_val_dir, 'dice.npy')
utils.mkdir(model_val_dir)
if (not os.path.isfile(dice_path)) | recompute:
loop_info.update(model_idx)
predict(path_images=image_dir,
path_model=path_model,
segmentation_labels=segmentation_labels,
n_neutral_labels=n_neutral_labels,
path_segmentations=model_val_dir,
padding=padding,
cropping=cropping,
target_res=target_res,
gradients=gradients,
flip=flip,
topology_classes=topology_classes,
sigma_smoothing=sigma_smoothing,
keep_biggest_component=keep_biggest_component,
conv_size=conv_size,
n_levels=n_levels,
nb_conv_per_level=nb_conv_per_level,
unet_feat_count=unet_feat_count,
feat_multiplier=feat_multiplier,
activation=activation,
gt_folder=gt_dir,
mask_folder=mask_dir,
evaluation_labels=evaluation_labels,
compute_distances=compute_distances,
recompute=recompute,
verbose=False)