[FIX] Fix t1-freesurfer pipeline handler (#1240)

* fix bad kwargs passing to pipeline handlers

* use t1-freesurfer for name

* fix handling of ignore_groups

* fix special case for segmentationVolumes

* add unit tests

* trigger CI

clinica/iotools/utils/data_handling.py

@@ -269,6 +269,8 @@ def _add_data_to_merge_file_from_caps(
         "dwi-dti": dwi_dti_pipeline,
     }
     merged_summary_df = pd.DataFrame()
+    if "group_selection" in kwargs and kwargs["group_selection"] is None:
+        kwargs.pop("group_selection")
     if not pipelines:
         for pipeline_name, pipeline_fn in pipeline_options.items():
             cprint(f"Extracting from CAPS pipeline output: {pipeline_name}...")

clinica/iotools/utils/pipeline_handling.py

@@ -12,7 +12,7 @@ def _get_atlas_name(atlas_path: Path, pipeline: str) -> str:
     """Helper function for _extract_metrics_from_pipeline."""
     if pipeline == "dwi_dti":
         splitter = "_dwi_space-"
-    elif pipeline in ("t1_freesurfer_longitudinal", "t1_freesurfer"):
+    elif pipeline in ("t1_freesurfer_longitudinal", "t1-freesurfer"):
         splitter = "_parcellation-"
     elif pipeline in ("t1-volume", "pet-volume"):
         splitter = "_space-"
@@ -46,7 +46,7 @@ def _get_mod_path(ses_path: Path, pipeline: str) -> Optional[Path]:
             / "freesurfer_longitudinal"
             / "regional_measures"
         )
-    if pipeline == "t1_freesurfer":
+    if pipeline == "t1-freesurfer":
         return ses_path / "t1" / "freesurfer_cross_sectional" / "regional_measures"
     if pipeline == "t1-volume":
         return ses_path / "t1" / "spm" / "dartel"
@@ -189,15 +189,15 @@ def _extract_metrics_from_pipeline(
         except KeyError:
             raise KeyError("Fields `participant_id` and `session_id` are required.")
 
-    if group_selection is None:
-        try:
-            group_selection = [f.name for f in (caps_dir / "groups").iterdir()]
-        except FileNotFoundError:
-            return df, None
-    else:
-        group_selection = [f"group-{group}" for group in group_selection]
     ignore_groups = group_selection == [""]
-
+    if not ignore_groups:
+        if group_selection is None:
+            try:
+                group_selection = [f.name for f in (caps_dir / "groups").iterdir()]
+            except FileNotFoundError:
+                return df, None
+        else:
+            group_selection = [f"group-{group}" for group in group_selection]
     subjects_dir = caps_dir / "subjects"
     records = []
     for participant_id, session_id in df.index.values:
@@ -227,6 +227,20 @@ def _extract_metrics_from_pipeline(
                                 )
                             )
                         for atlas_path in atlas_paths:
+                            if metric == "segmentationVolumes":
+                                from clinica.iotools.converters.adni_to_bids.adni_utils import (
+                                    replace_sequence_chars,
+                                )
+
+                                atlas_df = pd.read_csv(atlas_path, sep="\t")
+                                label_list = [
+                                    f"t1-freesurfer_segmentation-volumes_ROI-{replace_sequence_chars(roi_name)}_volume"
+                                    for roi_name in atlas_df.label_name.values
+                                ]
+                                values = atlas_df["label_value"].to_numpy()
+                                for label, value in zip(label_list, values):
+                                    records[-1][label] = value
+                                continue
                             if not _skip_atlas(
                                 atlas_path, pipeline, pvc_restriction, tracers_selection
                             ):
@@ -279,7 +293,7 @@ def _extract_metrics_from_pipeline(
 t1_freesurfer_pipeline = functools.partial(
     _extract_metrics_from_pipeline,
     metrics=["thickness", "segmentationVolumes"],
-    pipeline="t1_freesurfer",
+    pipeline="t1-freesurfer",
     group_selection=[""],
 )
 

test/unittests/iotools/utils/test_pipeline_handling.py

@@ -1,4 +1,5 @@
 import operator
+from pathlib import Path
 
 import pandas as pd
 import pytest
@@ -46,7 +47,7 @@ def test_get_mod_path_errors(tmp_path):
     "pipeline,expected_path",
     [
         ("dwi_dti", ["dwi", "dti_based_processing", "atlas_statistics"]),
-        ("t1_freesurfer", ["t1", "freesurfer_cross_sectional", "regional_measures"]),
+        ("t1-freesurfer", ["t1", "freesurfer_cross_sectional", "regional_measures"]),
         ("t1-volume", ["t1", "spm", "dartel"]),
         ("pet-volume", ["pet", "preprocessing"]),
     ],
@@ -210,3 +211,139 @@ def test_extract_metrics_from_pipeline(tmp_path):
         "group_id",
         "regions_number",
     }
+
+
+def test_t1_freesurfer_pipeline_nothing_found(tmp_path):
+    from pandas.testing import assert_frame_equal
+
+    from clinica.iotools.utils.pipeline_handling import t1_freesurfer_pipeline
+
+    caps = tmp_path / "caps"
+    caps.mkdir()
+    merged_df = pd.DataFrame(
+        {
+            "participant_id": ["sub-01"],
+            "session_id": ["ses-M000"],
+            "age": [85],
+        }
+    )
+    merged_df.set_index(
+        ["participant_id", "session_id"], inplace=True, verify_integrity=True
+    )
+    merged_df_with_t1_freesurfer_metrics, summary = t1_freesurfer_pipeline(
+        caps, merged_df
+    )
+    merged_df_with_t1_freesurfer_metrics.set_index(
+        ["participant_id", "session_id"], inplace=True, verify_integrity=True
+    )
+
+    assert_frame_equal(merged_df_with_t1_freesurfer_metrics, merged_df)
+    assert summary.empty
+
+
+def test_t1_freesurfer_longitudinal_pipeline_nothing_found(tmp_path):
+    from pandas.testing import assert_frame_equal
+
+    from clinica.iotools.utils.pipeline_handling import (
+        t1_freesurfer_longitudinal_pipeline,
+    )
+
+    caps = tmp_path / "caps"
+    caps.mkdir()
+    merged_df = pd.DataFrame(
+        {
+            "participant_id": ["sub-01"],
+            "session_id": ["ses-M000"],
+            "age": [85],
+        }
+    )
+    merged_df.set_index(
+        ["participant_id", "session_id"], inplace=True, verify_integrity=True
+    )
+    merged_df_with_t1_freesurfer_metrics, summary = t1_freesurfer_longitudinal_pipeline(
+        caps, merged_df
+    )
+    merged_df_with_t1_freesurfer_metrics.set_index(
+        ["participant_id", "session_id"], inplace=True, verify_integrity=True
+    )
+
+    assert_frame_equal(merged_df_with_t1_freesurfer_metrics, merged_df)
+    assert summary.empty
+
+
+def write_fake_statistics(tsv_file: Path):
+    fake = pd.DataFrame(
+        {
+            "label_name": ["foo", "bar", "baz"],
+            "label_value": [1.2, 2.3, 3.4],
+        }
+    )
+    fake.to_csv(tsv_file, sep="\t")
+
+
+def test_t1_freesurfer_pipeline(tmp_path):
+    from clinica.iotools.utils.pipeline_handling import t1_freesurfer_pipeline
+
+    caps = tmp_path / "caps"
+    regional_measures_folder = (
+        caps
+        / "subjects"
+        / "sub-01"
+        / "ses-M000"
+        / "t1"
+        / "freesurfer_cross_sectional"
+        / "regional_measures"
+    )
+    regional_measures_folder.mkdir(parents=True)
+    for file in (
+        "sub-01_ses-M000_parcellation-desikan_thickness.tsv",
+        "sub-01_ses-M000_parcellation-desikan_area.tsv",
+        "sub-01_ses-M000_parcellation-destrieux_thickness.tsv",
+        "sub-01_ses-M000_segmentationVolumes.tsv",
+    ):
+        write_fake_statistics(regional_measures_folder / file)
+
+    assert len([f for f in regional_measures_folder.iterdir()]) == 4
+    merged_df = pd.DataFrame(
+        {
+            "participant_id": ["sub-01"],
+            "session_id": ["ses-M000"],
+            "age": [85],
+        }
+    )
+    merged_df.set_index(
+        ["participant_id", "session_id"], inplace=True, verify_integrity=True
+    )
+    merged_df_with_t1_freesurfer_metrics, summary = t1_freesurfer_pipeline(
+        caps, merged_df
+    )
+    merged_df_with_t1_freesurfer_metrics.set_index(
+        ["participant_id", "session_id"], inplace=True, verify_integrity=True
+    )
+
+    assert len(merged_df_with_t1_freesurfer_metrics) == 1
+    assert set(merged_df_with_t1_freesurfer_metrics.columns) == {
+        "age",
+        "t1-freesurfer_atlas-desikan_ROI-bar_thickness",
+        "t1-freesurfer_atlas-desikan_ROI-baz_thickness",
+        "t1-freesurfer_atlas-desikan_ROI-foo_thickness",
+        "t1-freesurfer_atlas-destrieux_ROI-bar_thickness",
+        "t1-freesurfer_atlas-destrieux_ROI-baz_thickness",
+        "t1-freesurfer_atlas-destrieux_ROI-foo_thickness",
+        "t1-freesurfer_segmentation-volumes_ROI-bar_volume",
+        "t1-freesurfer_segmentation-volumes_ROI-baz_volume",
+        "t1-freesurfer_segmentation-volumes_ROI-foo_volume",
+    }
+    assert summary.pipeline_name.to_list() == ["t1-freesurfer"] * 3
+    assert summary.atlas_id.to_list() == ["desikan", "destrieux", "volumes"]
+    assert summary.regions_number.to_list() == [3, 3, 3]
+    assert summary.first_column_name.to_list() == [
+        "t1-freesurfer_atlas-desikan_ROI-foo_thickness",
+        "t1-freesurfer_atlas-destrieux_ROI-foo_thickness",
+        "t1-freesurfer_segmentation-volumes_ROI-foo_volume",
+    ]
+    assert summary.last_column_name.to_list() == [
+        "t1-freesurfer_atlas-desikan_ROI-baz_thickness",
+        "t1-freesurfer_atlas-destrieux_ROI-baz_thickness",
+        "t1-freesurfer_segmentation-volumes_ROI-baz_volume",
+    ]