Hack for loading radiological volumes with uint8 data range

utils/utilities.py

@@ -47,10 +47,12 @@ def input_file_category_disambiguation(input_filename: str) -> str:
     image_type = image.GetPixelIDTypeAsString()
     array = sitk.GetArrayFromImage(image)
 
-    # @TODO. Have encountered FLAIR images with integer values in [0, 255], from DICOM conversion, which will be
-    # failed associated as annotations...
     if len(np.unique(array)) > 255 or np.max(array) > 255 or np.min(array) < -1:
         category = "MRI"
+    # If the input radiological volume has values within [0, 255] only. Empirical solution for now, since less than
+    # 10 classes are usually handle at any given time.
+    elif len(np.unique(array)) >= 25:
+        category = "MRI"
     else:
         category = "Annotation"
     return category