Use continuous TPM-based procedure to calculate aCompCor masks (#3589)

Cloes #3587.

## Changes proposed in this pull request

- Assume that anatomical TPMs come from FAST and not Freesurfer's aseg
output, since sMRIPrep always uses FAST. This means setting `freesurfer`
to False in `init_bold_confs_wf`.
- Update boilerplate in `init_bold_confs_wf` to reflect changes and what
is actually happening.

Author: mgxd

fmriprep/interfaces/confounds.py

@@ -83,9 +83,9 @@ def _run_interface(self, runtime):
         from ..utils.confounds import acompcor_masks
 
         self._results['out_masks'] = acompcor_masks(
-            self.inputs.in_vfs,
-            self.inputs.is_aseg,
-            self.inputs.bold_zooms,
+            in_files=self.inputs.in_vfs,
+            is_aseg=self.inputs.is_aseg,
+            zooms=self.inputs.bold_zooms,
         )
         return runtime
 

fmriprep/utils/confounds.py

@@ -94,8 +94,8 @@ def acompcor_masks(in_files, is_aseg=False, zooms=None):
 
       * the masks are prepared in high-resolution, anatomical space and then
         projected into BOLD space; and,
-      * instead of using binary erosion, a dilated GM map is generated -- thresholding
-        the corresponding PV map at 0.05 (i.e., pixels containing at least 5% of GM tissue)
+      * instead of using binary erosion, a dilated GM map is generated after thresholding
+        the corresponding PV map at >0.05 (i.e., pixels containing more than 5% of GM tissue)
         and then subtracting that map from the CSF, WM and CSF+WM (combined) masks.
         This should be equivalent to eroding the masks, except that the erosion
         only happens at direct interfaces with GM.

fmriprep/workflows/bold/base.py

@@ -775,7 +775,7 @@ def init_bold_wf(
     bold_confounds_wf = init_bold_confs_wf(
         mem_gb=mem_gb['largemem'],
         metadata=all_metadata[0],
-        freesurfer=config.workflow.run_reconall,
+        freesurfer=False,  # sMRIPrep always uses FAST for TPMs
         regressors_all_comps=config.workflow.regressors_all_comps,
         regressors_fd_th=config.workflow.regressors_fd_th,
         regressors_dvars_th=config.workflow.regressors_dvars_th,

fmriprep/workflows/bold/confounds.py

@@ -116,6 +116,7 @@ def init_bold_confs_wf(
     freesurfer : :obj:`bool`
         Set to ``True`` if the input volume fractions for the anatomical
         component-based noise correction maps come from FreeSurfer's ``aseg``.
+        sMRIPrep always uses FAST for tissue probability maps, so we always set this to False.
     name : :obj:`str`
         Name of workflow (default: ``bold_confs_wf``)
 
@@ -171,7 +172,10 @@ def init_bold_confs_wf(
     gm_desc = (
         "dilating a GM mask extracted from the FreeSurfer's *aseg* segmentation"
         if freesurfer
-        else 'thresholding the corresponding partial volume map at 0.05'
+        else (
+            'thresholding the corresponding partial volume map at 0.05 and then '
+            'dilating the mask by 3 voxels'
+        )
     )
 
     workflow = Workflow(name=name)
@@ -196,10 +200,10 @@ def init_bold_confs_wf(
 voxels within the brain mask.
 For aCompCor, three probabilistic masks (CSF, WM and combined CSF+WM)
 are generated in anatomical space.
-The implementation differs from that of Behzadi et al. in that instead
+The implementation differs from that of Behzadi et al. in that, instead
 of eroding the masks by 2 pixels on BOLD space, a mask of pixels that
 likely contain a volume fraction of GM is subtracted from the aCompCor masks.
-This mask is obtained by {gm_desc}, and it ensures components are not extracted
+This mask is obtained by {gm_desc}, and it ensures that components are not extracted
 from voxels containing a minimal fraction of GM.
 Finally, these masks are resampled into BOLD space and binarized by
 thresholding at 0.99 (as in the original implementation).