Dear MRtrix community,
I am in the process of acquiring DTI on real patients and I am looking for the recommendation about voxel size to use on a 3T scanner, having a 12-channel head coil, which results in a reasonable SNR, acceptable tractography, and reasonable scan time. We are limited to use a single-shell (b=2000) acquisition with 45 directions. Any recommendations or suggestions would be highly appreciated.
It totally depends on the patient population (with focus on the motion/tremor, tolerating long acquisition times, etc.), the measure of interest, whether to use multiband or not, and the time you can keep the subject under the scanner (i.e., center-related scanner acquisition time regulations). Also, I totally recommend it would be really beneficial to acquire another shell.
I suggest running several pilots (with/without multiband, whole sequence in RPE or just a few b0s, different voxel sizes) on the same subject, doing quality control steps including visual inspection for the artifacts and automatic qc (-eddy_qc all and checking for CNR maps), run the whole analytical pipeline on each (e.g., specific ROI-to-ROI tractography, graph theory, etc.) and come to a final decision.