SPM fails to properly segment when pre-registerign the brain to MNI space

I have run SPM segmentation module in two scenarios: first one with no pre-registration to MNI space, and second one pre-registering the image with MNI_1mm_T1.nii template available from HCP github.

The pre-registration step was done as suggested here

To my surprise, when I performed the pre-registration step the segmentation completely failed. The reason I want to pre-register the scans is because according to some post like this one, this step can be helpful if there is some failure with the segmentation (I have also found in some papers that when they described the preprocessing pipeline they mention a step of ACPC correction prior to SPM/FreeSurfer analysis and I would have liked to implement that myself).

Further analysis of the outputs in the pre-registration case showed that the bias field correction was not adequate, thus I followed the procedure in the recommendation 2 when segmentation fails as suggested in the previous shared link. This did not improved the output.

I would like to know if this can be solved in some way, and whether it is useful to spend time trying to solve it. If the pre-registration step is only recommended when and only when the segmentation fails with no-preregistration (not the case in my scenario), is there an equivalent tool like VisualQC but for SPM outputs?

Many thanks in advance.