14 cropping of regions for big scans (#15)

* Bump up version

* Add cropping to relevant region

* Add cropping also for liver vessels

* The min/max was computed incorrectly

boa_contrast/features/feature_builder.py

@@ -1,7 +1,7 @@
 import logging
 import traceback
 from pathlib import Path
-from typing import Any, Dict, Generator, Optional
+from typing import Any, Dict, Generator, Optional, Tuple
 
 import cc3d
 import cv2
@@ -54,6 +54,8 @@ def compute_features(
         samples: Dict[str, Any] = {}
         ct_image = sitk.ReadImage(str(ct_data_path))
         ct_data = sitk.GetArrayViewFromImage(ct_image)
+        assert len(ct_data.shape) == 3, "The data should be a 3D CT scan."
+
         body_region_all = None
         if (
             not self.one_mask_per_file
@@ -84,7 +86,7 @@ def compute_features(
                     region_mask = np.isin(body_region_all, self.label_map[region])
             if np.sum(region_mask) == 0:
                 continue
-
+            region_mask, ct_data_region = crop_mask(region_mask, ct_data)
             if region in ORGANS:
                 remove_small_connected_components(region_mask)
 
@@ -93,7 +95,7 @@ def compute_features(
                     new_region = get_pelvis_region(region_mask)
                     compute_statistics(
                         samples,
-                        ct_data[new_region],
+                        ct_data_region[new_region],
                         region_name=f"{region}_pelvis",
                     )
                 except Exception:
@@ -108,13 +110,13 @@ def compute_features(
                 ):
                     compute_statistics(
                         samples,
-                        ct_data[partial_region_mask],
+                        ct_data_region[partial_region_mask],
                         region_name=f"{region}_part{i}",
                     )
             else:
                 compute_statistics(
                     samples,
-                    ct_data[region_mask],
+                    ct_data_region[region_mask],
                     region_name=region,
                 )
 
@@ -123,15 +125,27 @@ def compute_features(
             body_regions = sitk.ReadImage(str(liver_vessels_path))
             region_mask = sitk.GetArrayViewFromImage(body_regions)
             region_mask = region_mask.astype(bool)
+            region_mask, ct_data_region = crop_mask(region_mask, ct_data)
             compute_statistics(
                 samples,
-                ct_data[region_mask],
+                ct_data_region[region_mask],
                 region_name="liver_vessels",
             )
         assert len(samples) > 0, f"No regions were found in {segmentation_path.name}."
         return samples
 
 
+def crop_mask(mask: np.ndarray, ct_data: np.ndarray) -> Tuple[np.ndarray, np.ndarray]:
+    coords = np.argwhere(mask)
+    x_min, y_min, z_min = [coords[:, i].min() for i in range(3)]
+    x_max, y_max, z_max = [coords[:, i].max() + 1 for i in range(3)]
+
+    return (
+        mask[x_min:x_max, y_min:y_max, z_min:z_max],
+        ct_data[x_min:x_max, y_min:y_max, z_min:z_max],
+    )
+
+
 def create_split_regions(
     mask: np.ndarray, n_regions: int = False
 ) -> Generator[np.ndarray, None, None]:

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "boa-contrast"
-version = "0.1.3"
+version = "0.1.4"
 description = ""
 authors = ["Giulia Baldini <[email protected]>"]
 readme = "README.md"