Correct reg aladin values

macapype/nodes/prepare.py

@@ -294,8 +294,17 @@ def average_align(list_img, reorient=False):
             img_0 = nib.load(list_img[0])
             path, fname, ext = split_f(list_img[0])
 
-            list_data = [img_0.get_data()]
+            data_0 = img_0.get_fdata()
+
+            data_0_shape = data_0.shape
+            list_data = [data_0]
+
             for i, img in enumerate(list_img[1:]):
+                data_orig_shape = nib.load(img).get_fdata().shape
+                if data_orig_shape != data_0_shape:
+                    print(f"**** Warning , original image shape \
+                          {data_orig_shape} \
+                          have a different shape than ref {data_0_shape}")
 
                 print("running flirt on {}".format(img))
                 flirt = fsl.FLIRT(dof=6)
@@ -306,11 +315,21 @@ def average_align(list_img, reorient=False):
                 out_file = flirt.run().outputs.out_file
                 print(out_file)
 
-                data = nib.load(out_file).get_data()
+                data = nib.load(out_file).get_fdata()
+                assert data.shape == data_0_shape, \
+                    f"**** Error, flirt results {out_file} \
+                    have a different shape {data.shape} \
+                    than ref {data_0_shape}"
+
                 list_data.append(data)
 
             avg_data = np.mean(np.array(list_data), axis=0)
-            print(avg_data.shape)
+
+            assert data_0_shape == avg_data.shape, \
+                f"Error with average shape {avg_data.shape} != {data_0_shape}"
+
+            # after casting type
+            avg_data = avg_data.astype(np.int16)
 
             av_img_file = os.path.abspath("avg_" + fname + ext)
             nib.save(nib.Nifti1Image(avg_data, header=img_0.header,

macapype/nodes/register.py

@@ -190,15 +190,14 @@ class IterREGBETInputSpec(CommandLineInputSpec):
         mandatory=False, argstr="-refw %s")
 
     k = traits.Bool(
-        False, usedefault=True,
-        position=3, argstr="-k",
+        False, usedefault=True, argstr="-k",
         desc="Will keep temporary files",
         mandatory=False)
 
     p = traits.String(
         desc="Prefix for running FSL functions\
             (can be a path or just a prefix)",
-        position=3, argstr="-p %s")
+        argstr="-p %s")
 
 
 class IterREGBETOutputSpec(TraitedSpec):
@@ -749,11 +748,26 @@ def pad_zero_mri(img_file, pad_val=10, const=0):
         img_arr = np.array(img.dataobj)
         img_arr_copy = np.copy(img_arr)
 
-        img_arr_copy_padded = np.pad(
-            img_arr_copy,
-            pad_width=pad_val,
-            mode='constant',
-            constant_values=const)
+        if isinstance(pad_val, int):
+            print("running with pad_val")
+
+            img_arr_copy_padded = np.pad(
+                img_arr_copy,
+                pad_width=pad_val,
+                mode='constant',
+                constant_values=const)
+
+        elif pad_val == 'double':
+
+            pad_val = np.max(img_arr.shape)
+
+            print(f"running with double pad_val = {pad_val}")
+
+            img_arr_copy_padded = np.pad(
+                img_arr_copy,
+                pad_width=pad_val,
+                mode='constant',
+                constant_values=const)
 
         img_padded = nib.Nifti1Image(img_arr_copy_padded,
                                      header=img.header,

macapype/pipelines/full_pipelines.py

@@ -514,111 +514,118 @@ def create_brain_segment_from_mask_pipe(
                     "prob_csf", "gen_5tt"]),
         name='outputnode')
 
-    if "register_NMT_pipe" in params:
+    if "use_priors" in params["segment_atropos_pipe"].keys():
 
-        # register NMT template, template mask and priors to subject T1
-        register_NMT_pipe = create_register_NMT_pipe(
-            params_template=params_template,
-            params=parse_key(params, "register_NMT_pipe"))
-
-        brain_segment_pipe.connect(
-            inputnode, 'masked_debiased_T1',
-            register_NMT_pipe, "inputnode.T1")
-
-        brain_segment_pipe.connect(
-            inputnode, 'indiv_params',
-            register_NMT_pipe, "inputnode.indiv_params")
+        if "register_NMT_pipe" in params:
 
-    elif "reg" in params:
-        # Iterative registration to the INIA19 template
-        reg = NodeParams(
-            IterREGBET(),
-            params=parse_key(params, "reg"),
-            name='reg')
+            # register NMT template, template mask and priors to subject T1
+            register_NMT_pipe = create_register_NMT_pipe(
+                params_template=params_template,
+                params=parse_key(params, "register_NMT_pipe"))
 
-        reg.inputs.refb_file = params_template["template_brain"]
+            brain_segment_pipe.connect(
+                inputnode, 'masked_debiased_T1',
+                register_NMT_pipe, "inputnode.T1")
 
-        brain_segment_pipe.connect(
-            inputnode, 'debiased_T1',
-            reg, 'inw_file')
+            brain_segment_pipe.connect(
+                inputnode, 'indiv_params',
+                register_NMT_pipe, "inputnode.indiv_params")
 
-        brain_segment_pipe.connect(
-            inputnode, 'masked_debiased_T1',
-            reg, 'inb_file')
+        elif "reg" in params:
+            # Iterative registration to the template
+            reg = NodeParams(
+                IterREGBET(),
+                params=parse_key(params, "reg"),
+                name='reg')
 
-        brain_segment_pipe.connect(
-            inputnode, ('indiv_params', parse_key, "reg"),
-            reg, 'indiv_params')
+            reg.inputs.refb_file = params_template["template_brain"]
 
-        if "template_seg" in params_template.keys():
+            reg.inputs.refw_file = params_template["template_head"]
+            reg.inputs.k = True
 
-            # seg
-            register_seg_to_nat = pe.Node(
-                fsl.ApplyXFM(), name="register_seg_to_nat")
-            register_seg_to_nat.inputs.interp = "nearestneighbour"
+            brain_segment_pipe.connect(
+                inputnode, 'debiased_T1',
+                reg, 'inw_file')
 
-            register_seg_to_nat.inputs.in_file = params_template[
-                "template_seg"]
             brain_segment_pipe.connect(
                 inputnode, 'masked_debiased_T1',
-                register_seg_to_nat, 'reference')
+                reg, 'inb_file')
 
             brain_segment_pipe.connect(
-                reg, 'inv_transfo_file',
-                register_seg_to_nat, "in_matrix_file")
+                inputnode, ('indiv_params', parse_key, "reg"),
+                reg, 'indiv_params')
 
-        else:
-            # gm
-            register_gm_to_nat = pe.Node(
-                fsl.ApplyXFM(), name="register_gm_to_nat")
-            register_gm_to_nat.inputs.output_type = "NIFTI"  # for SPM segment
-            register_gm_to_nat.inputs.interp = "nearestneighbour"
+            if "template_seg" in params_template.keys():
 
-            register_gm_to_nat.inputs.in_file = params_template["template_gm"]
+                # seg
+                register_seg_to_nat = pe.Node(
+                    fsl.ApplyXFM(), name="register_seg_to_nat")
+                register_seg_to_nat.inputs.interp = "nearestneighbour"
 
-            brain_segment_pipe.connect(
-                inputnode, 'masked_debiased_T1',
-                register_gm_to_nat, 'reference')
+                register_seg_to_nat.inputs.in_file = params_template[
+                    "template_seg"]
+                brain_segment_pipe.connect(
+                    inputnode, 'masked_debiased_T1',
+                    register_seg_to_nat, 'reference')
 
-            brain_segment_pipe.connect(
-                reg, 'inv_transfo_file',
-                register_gm_to_nat, "in_matrix_file")
+                brain_segment_pipe.connect(
+                    reg, 'inv_transfo_file',
+                    register_seg_to_nat, "in_matrix_file")
 
-            # wm
-            register_wm_to_nat = pe.Node(
-                fsl.ApplyXFM(), name="register_wm_to_nat")
-            register_wm_to_nat.inputs.output_type = "NIFTI"  # for SPM segment
-            register_wm_to_nat.inputs.interp = "nearestneighbour"
+            else:
+                # gm
+                register_gm_to_nat = pe.Node(
+                    fsl.ApplyXFM(), name="register_gm_to_nat")
+                register_gm_to_nat.inputs.output_type = "NIFTI"  # for SPM
+                register_gm_to_nat.inputs.interp = "nearestneighbour"
 
-            register_wm_to_nat.inputs.in_file = params_template["template_wm"]
+                register_gm_to_nat.inputs.in_file = \
+                    params_template["template_gm"]
 
-            brain_segment_pipe.connect(
-                inputnode, 'masked_debiased_T1',
-                register_wm_to_nat, 'reference')
+                brain_segment_pipe.connect(
+                    inputnode, 'masked_debiased_T1',
+                    register_gm_to_nat, 'reference')
 
-            brain_segment_pipe.connect(
-                reg, 'inv_transfo_file',
-                register_wm_to_nat, "in_matrix_file")
+                brain_segment_pipe.connect(
+                    reg, 'inv_transfo_file',
+                    register_gm_to_nat, "in_matrix_file")
 
-            # csf
-            register_csf_to_nat = pe.Node(
-                fsl.ApplyXFM(), name="register_csf_to_nat")
-            register_csf_to_nat.inputs.output_type = "NIFTI"  # for SPM segment
-            register_csf_to_nat.inputs.interp = "nearestneighbour"
+                # wm
+                register_wm_to_nat = pe.Node(
+                    fsl.ApplyXFM(), name="register_wm_to_nat")
+                register_wm_to_nat.inputs.output_type = "NIFTI"  # for SPM
+                register_wm_to_nat.inputs.interp = "nearestneighbour"
 
-            register_csf_to_nat.inputs.in_file = params_template[
-                "template_csf"]
+                register_wm_to_nat.inputs.in_file = \
+                    params_template["template_wm"]
 
-            brain_segment_pipe.connect(
-                inputnode, 'masked_debiased_T1',
-                register_csf_to_nat, 'reference')
+                brain_segment_pipe.connect(
+                    inputnode, 'masked_debiased_T1',
+                    register_wm_to_nat, 'reference')
 
-            brain_segment_pipe.connect(
-                reg, 'inv_transfo_file',
-                register_csf_to_nat, "in_matrix_file")
-    else:
-        print("##### Error, no coregistration method is defined")
-        return brain_segment_pipe
+                brain_segment_pipe.connect(
+                    reg, 'inv_transfo_file',
+                    register_wm_to_nat, "in_matrix_file")
+
+                # csf
+                register_csf_to_nat = pe.Node(
+                    fsl.ApplyXFM(), name="register_csf_to_nat")
+                register_csf_to_nat.inputs.output_type = "NIFTI"  # for SPM
+                register_csf_to_nat.inputs.interp = "nearestneighbour"
+
+                register_csf_to_nat.inputs.in_file = \
+                    params_template["template_csf"]
+
+                brain_segment_pipe.connect(
+                    inputnode, 'masked_debiased_T1',
+                    register_csf_to_nat, 'reference')
+
+                brain_segment_pipe.connect(
+                    reg, 'inv_transfo_file',
+                    register_csf_to_nat, "in_matrix_file")
+        else:
+            print("##### Error, no coregistration method is defined")
+            return brain_segment_pipe
 
     # ants Atropos
     if "template_seg" in params_template.keys():