Merge pull request #1020 from frheault/lesions_divide

Modify lesions_info to support labels instead of mask

scilpy/image/labels.py

@@ -68,7 +68,8 @@ def get_binary_mask_from_labels(atlas, label_list):
     return mask
 
 
-def get_labels_from_mask(mask_data, labels=None, background_label=0):
+def get_labels_from_mask(mask_data, labels=None, background_label=0,
+                         min_voxel_count=0):
     """
     Get labels from a binary mask which contains multiple blobs. Each blob
     will be assigned a label, by default starting from 1. Background will
@@ -83,6 +84,9 @@ def get_labels_from_mask(mask_data, labels=None, background_label=0):
         label.
     background_label: int
         Label for the background.
+    min_voxel_count: int, optional
+        Minimum number of voxels for a blob to be considered. Blobs with fewer
+        voxels will be ignored.
 
     Returns
     -------
@@ -91,6 +95,19 @@ def get_labels_from_mask(mask_data, labels=None, background_label=0):
     """
     # Get the number of structures and assign labels to each blob
     label_map, nb_structures = ndi.label(mask_data)
+    if min_voxel_count:
+        new_count = 0
+        for label in range(1, nb_structures + 1):
+            if np.count_nonzero(label_map == label) < min_voxel_count:
+                label_map[label_map == label] = 0
+            else:
+                new_count += 1
+                label_map[label_map == label] = new_count
+        logging.debug(
+            f"Ignored blob {nb_structures-new_count} with fewer "
+            "than {min_voxel_count} voxels")
+        nb_structures = new_count
+
     # Assign labels to each blob if provided
     if labels:
         # Only keep the first nb_structures labels if the number of labels

scripts/scil_labels_from_mask.py

@@ -34,6 +34,10 @@ def _build_arg_parser():
     p.add_argument('--background_label', default=0, type=int,
                    help='Label to assign to the background. [%(default)s]')
 
+    p.add_argument('--min_volume', type=float, default=7,
+                   help='Minimum volume in mm3 [%(default)s],'
+                        'Useful for lesions.')
+
     add_verbose_arg(p)
     add_overwrite_arg(p)
 
@@ -50,9 +54,13 @@ def main():
     # Load mask and get data
     mask_img = nib.load(args.in_mask)
     mask_data = get_data_as_mask(mask_img)
+    voxel_volume = np.prod(np.diag(mask_img.affine)[:3])
+    min_voxel_count = args.min_volume // voxel_volume
+
     # Get labels from mask
     label_map = get_labels_from_mask(
-        mask_data, args.labels, args.background_label)
+        mask_data, args.labels, args.background_label,
+        min_voxel_count=min_voxel_count)
     # Save result
     out_img = nib.Nifti1Image(label_map.astype(np.uint16), mask_img.affine)
     nib.save(out_img, args.out_labels)

scripts/scil_lesions_info.py

@@ -3,12 +3,12 @@
 
 """
 This script will output informations about lesion load in bundle(s).
-The input can either be streamlines, binary bundle map, or a bundle voxel
-label map.
+The input lesion file is a labeled volume (.nii.gz) where each lesion is
+represented by a unique label. To label a lesion file use
+scil_labels_from_mask.py
 
-To be considered a valid lesion, the lesion volume must be at least
-min_lesion_vol mm3. This avoid the detection of thousand of single voxel
-lesions if an automatic lesion segmentation tool is used.
+Then, the second input can either be streamlines, binary bundle mask, or a
+bundle voxel label map.
 
 Formerly: scil_analyse_lesions_load.py
 """
@@ -20,7 +20,6 @@
 
 import nibabel as nib
 import numpy as np
-import scipy.ndimage as ndi
 
 from scilpy.image.labels import get_data_as_labels
 from scilpy.io.image import get_data_as_mask
@@ -40,7 +39,7 @@ def _build_arg_parser():
                                 formatter_class=argparse.RawTextHelpFormatter)
 
     p.add_argument('in_lesion',
-                   help='Binary mask of the lesion(s) (.nii.gz).')
+                   help='Lesions file as labels (.nii.gz).')
     p.add_argument('out_json',
                    help='Output file for lesion information (.json).')
     p1 = p.add_mutually_exclusive_group()
@@ -51,8 +50,6 @@ def _build_arg_parser():
     p1.add_argument('--bundle_labels_map',
                     help='Path of the bundle labels map (.nii.gz).')
 
-    p.add_argument('--min_lesion_vol', type=float, default=7,
-                   help='Minimum lesion volume in mm3 [%(default)s].')
     p.add_argument('--out_lesion_atlas', metavar='FILE',
                    help='Save the labelized lesion(s) map (.nii.gz).')
     p.add_argument('--out_lesion_stats', metavar='FILE',
@@ -89,7 +86,8 @@ def main():
                               reference=args.reference)
 
     lesion_img = nib.load(args.in_lesion)
-    lesion_data = get_data_as_mask(lesion_img, dtype=bool)
+    voxel_sizes = lesion_img.header.get_zooms()[0:3]
+    lesion_atlas = get_data_as_labels(lesion_img)
 
     if args.bundle:
         bundle_name, _ = split_name_with_nii(os.path.basename(args.bundle))
@@ -98,7 +96,7 @@ def main():
         sft.to_corner()
         streamlines = sft.get_streamlines_copy()
         map_data = compute_tract_counts_map(streamlines,
-                                            lesion_data.shape)
+                                            lesion_atlas.shape)
         map_data[map_data > 0] = 1
     elif args.bundle_mask:
         bundle_name, _ = split_name_with_nii(
@@ -112,8 +110,6 @@ def main():
         map_data = get_data_as_labels(map_img)
 
     is_single_label = args.bundle_labels_map is None
-    voxel_sizes = lesion_img.header.get_zooms()[0:3]
-    lesion_atlas, _ = ndi.label(lesion_data)
 
     lesion_load_dict = compute_lesion_stats(
         map_data, lesion_atlas, single_label=is_single_label,

scripts/tests/test_labels_from_mask.py

@@ -24,7 +24,7 @@ def test_execution(script_runner, monkeypatch):
                            'bundle_4_head_tail_offset.nii.gz')
     ret = script_runner.run('scil_labels_from_mask.py',
                             in_mask, 'labels_from_mask.nii.gz',
-                            '-f')
+                            '--min_volume', '0', '-f')
     assert ret.success