Merge pull request #1120 from nipreps/fix/skip_short_bold

FIX: Skip short BOLD runs that break outlier detection

mriqc/cli/parser.py

@@ -398,6 +398,14 @@ def _bids_filter(value):
 
     # Functional workflow settings
     g_func = parser.add_argument_group('Functional MRI workflow configuration')
+    g_func.add_argument(
+        '--min-bold-length',
+        action='store',
+        default=config.workflow.min_len_bold,
+        dest='min_len_bold',
+        help='Drop BOLD runs with fewer time points than this threshold.',
+        type=int,
+    )
     g_func.add_argument(
         '--fft-spikes-detector',
         action='store_true',

mriqc/config.py

@@ -554,6 +554,8 @@ class workflow(_Config):
     Minimum DWI length to be considered a "processable" dataset
     (default: 7, assuming one low-b and six gradients for diffusion tensor imaging).
     """
+    min_len_bold = 5
+    """Minimum BOLD length to be considered a "processable" dataset."""
     species = 'human'
     """Subject species to choose most appropriate template"""
     template_id = 'MNI152NLin2009cAsym'

mriqc/workflows/functional/base.py

@@ -42,6 +42,7 @@
 
 This workflow is orchestrated by :py:func:`fmri_qc_workflow`.
 """
+import nibabel as nb
 from nipype.interfaces import utility as niu
 from nipype.pipeline import engine as pe
 from niworkflows.utils.connections import pop_file as _pop
@@ -77,21 +78,40 @@ def fmri_qc_workflow(name='funcMRIQC'):
     mem_gb = config.workflow.biggest_file_gb
 
     dataset = config.workflow.inputs.get('bold', [])
+    full_files = []
+    for bold_path in dataset:
+        try:
+            bold_len = nb.load(bold_path).shape[3]
+        except nb.filebasedimages.ImageFileError:
+            bold_len = config.workflow.min_len_bold
+        except IndexError:    # shape has only 3 elements
+            bold_len = 0
+        if bold_len >= config.workflow.min_len_bold:
+            full_files.append(bold_path)
+        else:
+            config.loggers.workflow.warn(
+                f'Dismissing {bold_path} for processing: insufficient number of '
+                f'timepoints ({bold_len}) to execute the workflow.'
+            )
 
     message = BUILDING_WORKFLOW.format(
         modality='functional',
         detail=(
-            f'for {len(dataset)} BOLD runs.'
-            if len(dataset) > 2
+            f'for {len(full_files)} BOLD runs.'
+            if len(full_files) > 2
             else f"({' and '.join('<%s>' % v for v in dataset)})."
         ),
     )
     config.loggers.workflow.info(message)
 
+    if set(dataset) - set(full_files):
+        config.workflow.inputs['bold'] = full_files
+        config.to_filename()
+
     # Define workflow, inputs and outputs
     # 0. Get data, put it in RAS orientation
     inputnode = pe.Node(niu.IdentityInterface(fields=['in_file']), name='inputnode')
-    inputnode.iterables = [('in_file', dataset)]
+    inputnode.iterables = [('in_file', full_files)]
 
     datalad_get = pe.MapNode(
         DataladIdentityInterface(fields=['in_file'], dataset_path=config.execution.bids_dir),