Fix/anat fast track derivative

giga_auto_qc/assessments.py

@@ -8,7 +8,7 @@
 from nibabel import Nifti1Image
 
 from nilearn.image import load_img, resample_to_img
-from nilearn.masking import intersect_masks, _load_mask_img
+from nilearn.masking import intersect_masks, load_mask_img
 
 from bids import BIDSLayout
 
@@ -169,7 +169,7 @@ def _check_mask_affine(
     header_info = {"affine": []}
     key_to_header = {}
     for this_mask in mask_imgs:
-        _, affine = _load_mask_img(this_mask, allow_empty=True)
+        _, affine = load_mask_img(this_mask, allow_empty=True)
         affine_hashable = str(affine)
         header_info["affine"].append(affine_hashable)
         if affine_hashable not in key_to_header:
@@ -281,6 +281,7 @@ def calculate_functional_metrics(
             "mean_fd_raw": fds_mean_raw,
             "mean_fd_scrubbed": fds_mean_scrub,
             "proportion_kept": proportion_kept,
+            "total_frames": timeseries_length,
         }
 
     func_filter = {
@@ -334,6 +335,12 @@ def calculate_anat_metrics(
     """
     if verbose > 0:
         print("Calculate the anatomical dice score.")
+    # check if the derivative was created with anatomical fast-track
+    check_anat = fmriprep_bids_layout.get(datatype="anat", return_type="file")
+    if not check_anat:
+        print("`anat/` not present in the derivatives. " "Skip anatomical QC.")
+        return pd.DataFrame()
+
     metrics = {}
     for sub in tqdm(subjects):
         anat_filter = {
@@ -397,6 +404,12 @@ def quality_accessments(
         > qulaity_control_standards["functional_dice"]
     )
     functional_metrics["pass_func_qc"] = keep_fd * keep_proportion * keep_func
+    if anatomical_metrics.empty:
+        functional_metrics["pass_anat_qc"] = np.nan
+        functional_metrics["pass_all_qc"] = functional_metrics[
+            "pass_func_qc"
+        ].copy()
+        return functional_metrics
 
     # get the anatomical pass / fail
     pass_anat_qc = {}

giga_auto_qc/tests/test_assessments.py

@@ -31,6 +31,17 @@ def test_quality_accessments():
     )
     assert metrics["pass_all_qc"].astype(int).sum() == 1
 
+    metrics = assessments.quality_accessments(
+        functional_metrics=functional_metrics,
+        anatomical_metrics=pd.DataFrame(),
+        qulaity_control_standards=qc,
+    )
+    assert (
+        metrics["pass_all_qc"].astype(int).sum()
+        == metrics["pass_func_qc"].astype(int).sum()
+    )
+    assert metrics["pass_anat_qc"].isnull().all() == True
+
 
 def test_dice_coefficient():
     """Check the dice coefficient is calculated correctly."""

giga_auto_qc/utils.py

@@ -56,6 +56,7 @@ def parse_scan_information(metrics: pd.DataFrame) -> pd.DataFrame:
     """
     Parse the identifier into BIDS entities: subject, session, task, run.
     If session and run are not present, the information will not be parsed.
+    Parts of the identifier that do not match are ignored.
 
     Parameters
     ----------
@@ -73,20 +74,37 @@ def parse_scan_information(metrics: pd.DataFrame) -> pd.DataFrame:
     # get all unique entities
     headers_members = set()
     for id in metrics.index:
-        examplar = id.split("_")
-        new_headers = set([e.split("-")[0] for e in examplar])
+        exemplar = id.split("_")
+        new_headers = set(
+            [
+                e.split("-")[0]
+                for e in exemplar
+                if e.split("-")[0] in BIDS_ENTITIES
+            ]
+        )
         headers_members.update(new_headers)
     ordered_header = [None] * len(BIDS_ENTITIES)
     for header in headers_members:
-        ordered_header[BIDS_ENTITIES[header]] = header
+        if header in BIDS_ENTITIES:
+            ordered_header[BIDS_ENTITIES[header]] = header
     headers = [h for h in ordered_header if h is not None]  # remove none
 
     identifiers = pd.DataFrame(
         metrics.index.tolist(), index=metrics.index, columns=["identifier"]
     )
-    identifiers[headers] = identifiers["identifier"].str.split(
-        "_", expand=True
-    )
+    split_data = identifiers["identifier"].str.split("_", expand=True)
+
+    # filter split data to only include columns corresponding to recognised headers
+    split_data = split_data.iloc[
+        :,
+        [
+            i
+            for i, part in enumerate(identifiers["identifier"][0].split("_"))
+            if part.split("-")[0] in BIDS_ENTITIES
+        ],
+    ]
+    identifiers[headers] = split_data
+
     identifiers = identifiers.drop("identifier", axis=1)
     for h in headers:
         identifiers[h] = identifiers[h].str.replace(f"{h}-", "")

giga_auto_qc/workflow.py

@@ -51,6 +51,12 @@ def workflow(args):
         derivatives=True,
         reset_database=args.reindex_bids,
     )
+    if fmriprep_bids_layout is None:
+        raise ValueError(
+            f"Cannot index directory in {bids_dir}. "
+            "Please ensure the path is a fMRIPrep output directory."
+        )
+
     # check output path
     output_dir.mkdir(parents=True, exist_ok=True)