WIP: CSF normalization

Author: mgxd

nibabies/cli/parser.py

@@ -753,6 +753,11 @@ def _str_none(val):
         default=16,
         help='Frame to start head motion estimation on BOLD.',
     )
+    g_baby.add_argument(
+        '--norm-csf',
+        action='store_true',
+        help='Replace low intensity voxels in CSF mask with average',
+    )
     return parser
 
 

nibabies/config.py

@@ -578,6 +578,8 @@ class workflow(_Config):
     """Run FreeSurfer ``recon-all`` with the ``-logitudinal`` flag."""
     medial_surface_nan = None
     """Fill medial surface with :abbr:`NaNs (not-a-number)` when sampling."""
+    norm_csf = False
+    """Replace low intensity voxels in CSF mask with average."""
     project_goodvoxels = False
     """Exclude voxels with locally high coefficient of variation from sampling."""
     regressors_all_comps = None

nibabies/workflows/anatomical/fit.py

@@ -184,6 +184,12 @@ def init_infant_anat_fit_wf(
         name='anat_buffer',
     )
 
+    # Additional CSF normalization, if necessary
+    anat_norm_buffer = pe.Node(
+        niu.IdentityInterface(fields=['anat_preproc']),
+        name='anat_norm_buffer',
+    )
+
     if reference_anat == 'T1w':
         LOGGER.info('ANAT: Using T1w as the reference anatomical')
         workflow.connect([
@@ -637,24 +643,6 @@ def init_infant_anat_fit_wf(
                     (binarize_t2w, t2w_buffer, [('out_file', 't2w_mask')]),
                 ])  # fmt:skip
         else:
-            # Check whether we can convert a previously computed T2w mask
-            # or need to run the atlas based brain extraction
-
-            # if t1w_mask:
-            #     LOGGER.info('ANAT T1w mask will be transformed into T2w space')
-            #     transform_t1w_mask = pe.Node(
-            #         ApplyTransforms(interpolation='MultiLabel'),
-            #         name='transform_t1w_mask',
-            #     )
-
-            #     workflow.connect([
-            #         (t1w_buffer, transform_t1w_mask, [('t1w_mask', 'input_image')]),
-            #         (coreg_buffer, transform_t1w_mask, [('t1w2t2w_xfm', 'transforms')]),
-            #         (transform_t1w_mask, apply_t2w_mask, [('output_image', 'in_mask')]),
-            #         (t2w_buffer, apply_t1w_mask, [('t2w_preproc', 'in_file')]),
-            #         # TODO: Unsure about this connection^
-            #     ])  # fmt:skip
-            # else:
             LOGGER.info('ANAT Atlas-based brain mask will be calculated on the T2w')
             brain_extraction_wf = init_infant_brain_extraction_wf(
                 omp_nthreads=omp_nthreads,

nibabies/workflows/anatomical/preproc.py

@@ -66,3 +66,29 @@ def init_anat_preproc_wf(
         (final_clip, outputnode, [('out_file', 'anat_preproc')]),
     ])  # fmt:skip
     return wf
+
+
+def init_anat_csf_norm_wf(name='anat_csf_norm_wf') -> LiterateWorkflow:
+    """Replace low intensity voxels within the CSF mask with the median value."""
+
+    workflow = LiterateWorkflow(name=name)
+    inputnode = niu.IdentityInterface(fields=['anat_preproc', 'anat_dseg'], name='inputnode')
+    outputnode = niu.IdentityInterface(fields=['anat_preproc'], name='outputnode')
+
+    applymask = pe.Node(ApplyMask(), name='applymask')
+
+    norm2median = pe.Node(niu.Function(function=_normalize_csf), name='norm2median')
+    # 1. mask brain with CSF mask
+    # fslmaths input.nii.gz -mas aseg_label-CSF_mask.nii.gz input_CSF.nii.gz
+    # 2. get median intensity of nonzero voxels in mask
+    # fslstats input_CSF.nii.gz -P 50
+    # 3. normalize CSF-masked T2w to the median
+    # fslmaths input_CSF.nii.gz -bin -mul <median intensity from (> median_CSF.nii.gz
+    # 4. make the modified T2w, setting voxel intensity to be the max between the original T2w's,
+    # and the normalized mask from (3)'s:
+    # fslmaths input.nii.gz -max median_CSF.nii.gz input_floorCSF.nii.gz
+
+    return workflow
+
+
+def _normalize_csf(in_file): ...