Hi all,
I am a PhD student exploring analysis of diffusion images.
I am running the HCP diffusion pipeline on a dataset with an asymmetric AP/PA DWI acquisition design:
- AP direction: 3 b0 volumes only (used as field map reference for topup)
- PA direction: full multi-shell DWI dataset (~90 volumes, b=300/1000/2000)
The pipeline runs successfully through eddy:
- Pos_Neg.nii.gz: 187 volumes ✓
- eddy_unwarped_images.nii.gz: 187 volumes ✓
However, the final data.nii.gz contains only 6 volumes with bvals: 0 0 212 212 212 212. It appears that after eddy, the pipeline extracts only the AP volumes (3 b0 from session 1 + 3 b0 from session 2), discarding the full PA DWI dataset. As a result, dtifit produces physically impossible values (FA > 1, negative eigenvalues).
I have been advised that the parameter hcp_dwi_combinedata might be relevant here, but I am not sure how to configure it correctly for this acquisition design.
My questions:
1. Does the HCP diffusion pipeline support asymmetric AP/PA acquisitions where only one direction contains the full DWI dataset?
2. What are the correct HCP Pipelines parameters to handle this scenario (b0-only reference in AP, full DWI in PA)?
3. Specifically, how should hcp_dwi_combinedata be set in this case?
Any advice would be greatly appreciated!
Best regards,
Benjamin Bušelič