Fix value error with 4D images

Fixes #95

Add handling for 4D images in `transform_seg2image` function.

* Add a check for 4D images at the beginning of the `transform_seg2image` function.
* If the input image is 4D, take the first image from the last axis for resampling.
* Add a comment explaining the purpose of the check for 4D images.

---

For more details, open the [Copilot Workspace session](https://copilot-workspace.githubnext.com/neuropoly/totalspineseg/issues/95?shareId=XXXX-XXXX-XXXX-XXXX).

totalspineseg/utils/transform_seg2image.py

@@ -236,6 +236,10 @@ def transform_seg2image(
     nibabel.Nifti1Image
         Output segmentation.
     '''
+    # Check if the input image is 4D and take the first image from the last axis for resampling
+    if len(np.asanyarray(image.dataobj).shape) == 4:
+        image = image.slicer[..., 0]
+
     image_data = np.asanyarray(image.dataobj).astype(np.float64)
     image_affine = image.affine.copy()
     seg_data = np.asanyarray(seg.dataobj)
@@ -307,4 +311,4 @@ def transform_seg2image(
     return output_seg
 
 if __name__ == '__main__':
-    main()
+    main()