top of page
  • Writer's picturePennLINC

QSI Prep Launched

Huge congrats to Matt Cieslak!! After a year of development and testing, we are thrilled to introduce QSIPrep – a rigorous, automatic, and transparent BIDS app for diffusion MRI preprocessing and reconstruction. Check out it out here

QSIPrep adopts the framework and philosophy of fMRIPrep, and takes BIDS data as input.

QSIPrep conforms your diffusion-weighted images and their bvecs to a common orientation, and takes care to prevent unintentional gradient direction flipping at each step of the pipeline (this is often very difficult when using disparate packages). Distortion correction can be performed with either BIDS-defined fieldmaps, or using experimental fieldmapless SyN-based registration. TOPUP is used for EPI fieldmaps if using eddy. Otherwise, FMRIPREP’s workflows have been adjusted to work with B0 images. Head motion correction is available to all sampling schemes. DTI and multi-shell sampling can be corrected using FSL’s eddy. Additionally, DSI, compressed-sensing DSI, and multi-shell can be corrected by our novel 3dSHORE-based SHORELine method. An unbiased single-subject B0 template can be created using ANTs. This is useful for multi-session longitudinal studies, where the shape of the brain changes due to aging and/or distortion. T1w co-registration, normalization, and resampling is done with ANTs. The final resampling is done in a single-shot that can incorporate head motion, eddy current and distortion correction, warping to the B0 SST, coregistration, and normalization. HTML reports similar to FMRIPREP’s provide a quick sanity-check that the pipeline worked correctly. Check out: Head motion reports, framewise displacement and outlier reports are always provided. Finally, we have a curated set of “Reconstruction Workflows” that use Dipy, MRtrix3 and DSI Studio to take your preprocessed data to connectivity matrices, fixels, ODFs, FODs and many other outputs. Our development model is open and the code is available under the BSD 3-clause license. Each release goes through CI testing. All code here: Please file bug reports – and pull requests are welcome!

320 views0 comments


bottom of page