# MASiVar: Multisite, Multiscanner, and Multisubject Acquisitions for Studying Variability in Diffusion Weighted Magnetic Resonance Imaging

## Authors and Reference

[Leon Y. Cai](mailto:leon.y.cai@vanderbilt.edu), Qi Yang, Praitayini Kanakaraj, Vishwesh Nath, Allen T. Newton, Heidi A. Edmonson, Jeffrey Luci, Benjamin N. Conrad, Gavin R. Price, Colin B. Hansen, Cailey I. Kerley, Karthik Ramadass, Fang-Cheng Yeh, Hakmook Kang, Eleftherios Garyfallidis, Maxime Descoteaux, Francois Rheault, Kurt G. Schilling, and Bennett A. Landman. MASiVar: Multisite, Multiscanner, and Multisubject Acquisitions for Studying Variability in Diffusion Weighted Magnetic Resonance Imaging. [bioRxiv](https://www.biorxiv.org/content/10.1101/2020.12.03.408567v3), 2021. Preprint.

[Medical-image Analysis and Statistical Interpretation (MASI) Lab](https://my.vanderbilt.edu/masi), Vanderbilt University, Nashville, TN, USA

## Overview

MASiVar is a dataset consisting of 319 diffusion scans acquired at 3T from b = 1000 to 3000 s/mm<sup>2</sup> across 14 healthy adults, 83 healthy children (5 to 8 years), three sites, and four scanners curated to promote investigation of diffusion MRI variability. Cohort I consists of one adult subject scanned repeatedly on one scanner. This subject underwent three separate imaging sessions and acquired 3-4 scans per session. Cohort II consists of 5 adult subjects each scanned on 3-4 different scanners across 3 institutions. Each subject underwent 1-2 sessions on each scanner and had one scan acquired per session. Cohort III consists of 8 adult subjects all scanned on one scanner. Each subject underwent 1-6 sessions on the scanner and had two scans acquired per session. Cohort IV consists of 83 child subjects all scanned on one scanner. Each subject underwent 1-2 sessions on the scanner and had two scans acquired per session.

The acquisitions acquired per scan are as follows:

| Cohort | Shell (b-value) | Number of Directions|
|--------|-----------------|---------------------|
| I | 1000 | 96 |
| | 1500 | 96 |
| | 2000 | 96 |
| | 2500 | 96 |
| | 3000 | 96 |
| II | 1000 | 30 to 33 |
| | 1000 | 96 |
| | 1500 | 96 |
| | 2000 | 96 |
| | 2465 or 2500 | 96 |
| III | 1000 | 40 |
| | 2000 | 56 |
| IV | 1000 | 40 |
| | 2000 | 56 |

## Naming Scheme

MASiVar is in BIDS format with the following naming scheme:

* Subject: __sub-c__\<cohort\>__s__\<subject\>
* Session: __ses-s__\<site (and scanner if applicable)\>__x__\<session number\>
* Acquisition: __acq-b__\<shell\>__n__\<number of directions\>__r__\<resolution\>__pe__\<phase encoding direction\>
    * Shells are indicated by b-value (s/mm<sup>2</sup>)
    * Resolutions are presented in 10<sup>-1</sup> mm to maintain BIDS compliance.
    * Phase encoding direction APP indicates posterior-to-anterior and APA indicates anterior-to-posterior.
* Run: __run-__<scan number (first digit)><acquisition number (last two digits)>

Example: `sub-cIs1_ses-s1Ax2_acq-b3000n96r25x25x25peAPP_run-105_dwi.nii.gz` is the fifth acquisition in the first scan of session 2 at site 1A for subject 1 of cohort I acquired with 96 directions at b = 3000 s/mm<sup>2</sup> and a resolution of 2.5mm isotropic in the posterior-to-anterior phase encoding direction.

Note: Most of the subjects and sessions are named sequentially, however, some are not due to missed or truncated imaging sessions.

## Derivatives

Both raw and preprocessed MASiVar data are available de-faced and de-identified. Diffusion images were preprocessed with PreQual v1.0.0 under default settings. More information about PreQual can be found here: https://github.com/MASILab/PreQual. In short, all acquisitions per scan were denoised with the Marchenko-Pastur PCA technique, intensity normalized, and distortion corrected. Distortion correction included susceptibility-induced distortion correction using APA b = 0 s/mm<sup>2</sup> volumes when available and the Synb0-DisCo deep learning framework and T1 images when not, eddy current-induced distortion correction, intervolume motion correction, and slice-wise signal drop out imputation.