Anatomically constraint tractography

1

Hi,

I have a question regarding the possibility of using of the ACT method for performing tractography for DWI data obtaining without reverse phase encoding direction (my DWI data were acquired with one and only one b0 volume and 64 volumes with b=1000) . I read these topics (Anatomically-Constrained Tractography (ACT) — MRtrix 3.0 documentation , GitHub - bids-apps/MRtrix3_connectome: Generate subject connectomes from raw image data and perform inter-subject connection density normalisation, using tools provided in the MRtrix3 software package.) and this question arose , can ACT be preformed without correction of EPI susceptibility distortions?

Any help would be much appreciated,

cheers,

-Milad

2

Hello,

It is not recommended. For ACT to work, it is assumed that the GM-WM boundaries derived from the T1 images are also valid for the DWI image. Without susceptibility distortion correction, that assumption is likely not met, even if the DWI and T1 are rigidly aligned.

That being said, QSIPrep has implemented fieldmap-less distortion using symmetric normalization, so you can preprocess your data with QSIPrep and then use ACT, as long as the phase encoding direction of the original image is available.

Best,
Steven

3

Hello @smeisler

Thanks so much for the detailed response. My DWI data have no fieldmap and I can’t use this method for modifying the EPI distortion. But what is your idea regarding the non-linear registration between the T1 images and DWI image ANTs as described here (Frontiers | Evaluation of Field Map and Nonlinear Registration Methods for Correction of Susceptibility Artifacts in Diffusion MRI | Frontiers in Neuroinformatics) ? In this paper, the authors recommended that nonlinear
registration with ANTs can be used to correct this distortion in the absence of field map and reverse phase encoding gradients.

cheers,
-Milad

4

Hi,

Yes that is the paper I was trying to link to in my last response, and the method implemented in QSIPrep.

Best,
Steven

5

Hi @milad,

If your data is adult human data, you can use this method:

They also provide the full code to make it work.

Best regards,

Manuel

6

Hi @smeisler @mblesac

Many thanks for your helps. I try to correct this distortion using both methods and share my results.

cheers,
-Milad