diff --git a/xcp_d/interfaces/regression.py b/xcp_d/interfaces/regression.py
index e0f892e93..54e1bce68 100644
--- a/xcp_d/interfaces/regression.py
+++ b/xcp_d/interfaces/regression.py
@@ -73,7 +73,6 @@ def _run_interface(self, runtime):
         if self.inputs.custom_confounds and exists(self.inputs.custom_confounds):
             confound = load_confound_matrix(
                 original_file=self.inputs.original_file,
-                datafile=self.inputs.in_file,
                 custom_confounds=self.inputs.custom_confounds,
                 confound_tsv=self.inputs.confounds,
                 params=self.inputs.params,
@@ -81,10 +80,10 @@ def _run_interface(self, runtime):
         else:  # No custom confounds
             confound = load_confound_matrix(
                 original_file=self.inputs.original_file,
-                datafile=self.inputs.in_file,
                 confound_tsv=self.inputs.confounds,
                 params=self.inputs.params,
             )
+
         # for testing, let's write out the confounds file:
         confounds_file_output_name = fname_presuffix(
             self.inputs.confounds,
diff --git a/xcp_d/utils/confounds.py b/xcp_d/utils/confounds.py
index bd07922b9..ee5167a0f 100644
--- a/xcp_d/utils/confounds.py
+++ b/xcp_d/utils/confounds.py
@@ -300,14 +300,12 @@ def square_confound(confound):
 
 @fill_doc
 def load_confound_matrix(
-    datafile, original_file, params, custom_confounds=None, confound_tsv=None
+    original_file, params, custom_confounds=None, confound_tsv=None
 ):
     """Load a subset of the confounds associated with a given file.
 
     Parameters
     ----------
-    datafile : str
-        BOLD file whose confounds we want.
     original_file :
        File used to find confounds json.
     %(params)s
@@ -386,13 +384,13 @@ def load_confound_matrix(
         confound = pd.concat([derivative_rot_trans, acompcor, cosine], axis=1)
     elif params == "aroma":  # Get the WM, CSF, and aroma values
         whitematter_csf = load_wm_csf(confoundtsv)
-        aroma = load_aroma(datafile=datafile)
+        aroma = load_aroma(datafile=original_file)
         confound = pd.concat([whitematter_csf, aroma], axis=1)
     elif (
         params == "aroma_gsr"
     ):  # Get the WM, CSF, and aroma values, as well as global signal
         whitematter_csf = load_wm_csf(confoundtsv)
-        aroma = load_aroma(datafile=datafile)
+        aroma = load_aroma(datafile=original_file)
         global_signal = load_global_signal(confoundtsv)
         confound = pd.concat([whitematter_csf, aroma, global_signal], axis=1)
     elif (