[ENH] add HTML reports for each atlas / masked image (#115)

* add reports

* add matploblib

* spelling

giga_connectome/connectome.py

@@ -114,7 +114,7 @@ def generate_timeseries_connectomes(
     group_mask: str | Path,
     correlation_measure: ConnectivityMeasure,
     calculate_average_correlation: bool,
-) -> tuple[np.ndarray[Any, Any], np.ndarray[Any, Any]]:
+) -> tuple[np.ndarray[Any, Any], np.ndarray[Any, Any], NiftiMasker]:
     """Generate timeseries-based connectomes from functional data.
 
     Parameters
@@ -159,4 +159,4 @@ def generate_timeseries_connectomes(
     # convert to float 32 instead of 64
     time_series_atlas = time_series_atlas.astype(np.float32)
     correlation_matrix = correlation_matrix.astype(np.float32)
-    return correlation_matrix, time_series_atlas
+    return correlation_matrix, time_series_atlas, masker

giga_connectome/postprocess.py

@@ -161,15 +161,14 @@ def run_postprocessing_dataset(
                     continue
 
                 # extract timeseries and connectomes
-                (
-                    correlation_matrix,
-                    time_series_atlas,
-                ) = generate_timeseries_connectomes(
-                    masker,
-                    denoised_img,
-                    group_mask,
-                    correlation_measure,
-                    calculate_average_correlation,
+                (correlation_matrix, time_series_atlas, masker) = (
+                    generate_timeseries_connectomes(
+                        masker,
+                        denoised_img,
+                        group_mask,
+                        correlation_measure,
+                        calculate_average_correlation,
+                    )
                 )
 
                 # dump correlation_matrix to tsv
@@ -198,6 +197,17 @@ def run_postprocessing_dataset(
                 df = pd.DataFrame(time_series_atlas)
                 df.to_csv(timeseries_filename, sep="\t", index=False)
 
+                report = masker.generate_report()
+                report_filename = connectome_path / utils.output_filename(
+                    source_file=Path(img.filename).stem,
+                    atlas=atlas["name"],
+                    suffix="report",
+                    extension="html",
+                    strategy=strategy["name"],
+                    desc=desc,
+                )
+                report.save_as_html(report_filename)
+
             progress.update(task, advance=1)
 
     gc_log.info(f"Saved to:\n{connectome_path}")
@@ -208,8 +218,14 @@ def _get_masker(atlas_path: Path) -> NiftiLabelsMasker | NiftiMapsMasker:
     atlas_type = atlas_path.name.split("_")[-1].split(".nii")[0]
     if atlas_type == "dseg":
         atlas_masker = NiftiLabelsMasker(
-            labels_img=atlas_path, standardize=False
+            labels_img=atlas_path,
+            standardize=False,
+            cmap="gray",
         )
     elif atlas_type == "probseg":
-        atlas_masker = NiftiMapsMasker(maps_img=atlas_path, standardize=False)
+        atlas_masker = NiftiMapsMasker(
+            maps_img=atlas_path,
+            standardize=False,
+            cmap="gray",
+        )
     return atlas_masker

giga_connectome/tests/test_connectome.py

@@ -54,7 +54,7 @@ def test_calculate_intranetwork_correlation():
     correlation_measure = ConnectivityMeasure(
         kind="correlation", vectorize=False, discard_diagonal=False
     )
-    conn, tseries = generate_timeseries_connectomes(
+    conn, _, _ = generate_timeseries_connectomes(
         masker=NiftiLabelsMasker(labels_img=atlas),
         denoised_img=denoised_img,
         group_mask=mask,

pyproject.toml

@@ -21,7 +21,7 @@ classifiers = [
 ]
 dependencies = [
   "h5py",
-  "nilearn >=0.10.2",
+  "nilearn[plotting] >=0.10.2",
   "pybids >=0.15.0, <0.16.0",
   "templateflow < 23.0.0",
   "setuptools",

requirements.txt

@@ -11,7 +11,7 @@ interface-meta==1.3.0
 joblib==1.3.2
 lxml==4.9.3
 nibabel==5.2.0
-nilearn==0.10.2
+nilearn[plotting]==0.10.2
 num2words==0.5.13
 numpy==1.26.2
 packaging==23.2