Final fixes per CodeRabbit review

- Improve CT docstring with clinical rationale for HU windowing [-1000, 400]
- Improve MRI docstring explaining nonzero=True for background exclusion
- Add explicit test_mr_modality_distinct() for MR modality testing
- Update integration test to cover CT, MR, and MRI modalities
- Update return type annotation and docstring for MetaTensor

Signed-off-by: Hitendrasinh Rathod <[email protected]>

Author: Hitendrasinhdata7

monai/tests/test_clinical_preprocessing.py

@@ -90,6 +90,14 @@ def test_modality_case_insensitivity(mock_load):
         assert result is not None
 
 
+@patch("monai.transforms.clinical_preprocessing.LoadImage")
+def test_mr_modality_distinct(mock_load):
+    """Test MR modality is handled separately from MRI."""
+    mock_load.return_value = Mock(return_value=Mock())
+    result = preprocess_dicom_series("dummy.dcm", "MR")
+    assert result is not None
+
+
 def test_preprocess_dicom_series_integration(tmp_path):
     """Integration test with dummy NIfTI file."""
     # Create a dummy NIfTI file for testing
@@ -98,8 +106,8 @@ def test_preprocess_dicom_series_integration(tmp_path):
 
     write_nifti(dummy_data, test_file)
 
-    # Test with each modality
-    for modality in ["CT", "MRI"]:
+    # Test with each modality (including both MR and MRI)
+    for modality in ["CT", "MR", "MRI"]:
         result = preprocess_dicom_series(str(test_file), modality)
         assert result is not None
         assert hasattr(result, "shape")

monai/transforms/clinical_preprocessing.py

@@ -15,8 +15,7 @@
 This module provides modality-specific preprocessing pipelines for common medical imaging modalities.
 """
 
-from typing import Any
-
+from monai.data import MetaTensor
 from monai.transforms import (
     Compose,
     EnsureChannelFirst,
@@ -39,7 +38,9 @@ def get_ct_preprocessing_pipeline() -> Compose:
     Create a preprocessing pipeline for CT images.
 
     Returns:
-        Compose: Transform composition for CT preprocessing.
+        Compose: Transform composition for CT preprocessing. Applies HU windowing
+            [-1000, 400] scaled to [0, 1] with clipping, suitable for soft tissue
+            and lung visualization.
     """
     return Compose(
         [
@@ -61,7 +62,9 @@ def get_mri_preprocessing_pipeline() -> Compose:
     Create a preprocessing pipeline for MRI images.
 
     Returns:
-        Compose: Transform composition for MRI preprocessing.
+        Compose: Transform composition for MRI preprocessing. Normalizes using
+            mean/std computed over non-zero voxels only, appropriate for MRI
+            data with background regions.
     """
     return Compose(
         [
@@ -72,15 +75,15 @@ def get_mri_preprocessing_pipeline() -> Compose:
     )
 
 
-def preprocess_dicom_series(path: str, modality: str) -> Any:
+def preprocess_dicom_series(path: str, modality: str) -> MetaTensor:
     """Preprocess a DICOM series or file based on imaging modality.
 
     Args:
         path: Path to the DICOM file or directory containing a DICOM series.
         modality: Imaging modality. Supported values are "CT", "MR", and "MRI" (case-insensitive).
 
     Returns:
-        The preprocessed image data.
+        MetaTensor: Preprocessed image tensor with metadata.
 
     Raises:
         ModalityTypeError: If modality is not a string.