Skip to content

Multi-Site Traveling Subjects Diffusion MRI Dataset

Latest
Latest

Choose a tag to compare

View all tags
@data-hcp data-hcp released this 26 Mar 18:49
urmc-travel
3f46f7f
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Reproducibility Dataset Across 10 Imaging Centers

This dataset provides a multi-shell diffusion MRI benchmark collected from three traveling human subjects scanned across ten imaging centers, with tightly controlled acquisition protocols. It is designed to evaluate scanner variability, site effects, and reproducibility in diffusion MRI studies.

Preprocess note

Quality control (QC) analysis identified a significant issue with the b-table data across the majority of the original dataset, which would lead to incorrect diffusion tensor reconstruction results. Specifically, only two scans, sub-3_ses-c02r1 and sub-2_ses-c02r1, were found to have valid b-values (bval) and b-vectors (bvec)

Dataset Overview

  • Subjects: 3 (traveling subjects scanned repeatedly)
  • Sites: 10 imaging centers
  • Design: Same subjects scanned across multiple sites
  • Purpose:
    • Quantify inter-site variability
    • Evaluate reproducibility of diffusion MRI metrics
    • Benchmark harmonization and processing pipelines

Imaging Protocol

  • Diffusion MRI:

    • Multi-shell acquisition
    • Identical protocol across all sites

  • Structural MRI:

    • T1-weighted imaging
    • Consistent acquisition parameters

  • Key feature:

    • Strict control of scanner settings and protocols
    • Minimizes confounds unrelated to site/scanner differences

Preprocessing Pipeline

The dataset includes a standardized preprocessing workflow implemented on:

  • System: CentOS 6.7

Software Dependencies

  • FSL 5.0.11
  • FreeSurfer v1.379.2.73
  • MRtrix3 0.3.15

Processing Steps

Diffusion Processing

  1. Denoising (MRtrix3)
  2. Gibbs ringing removal (unring)
  3. Distortion correction
    • TOPUP (susceptibility distortion)
    • EDDY (motion and eddy current correction)

Structural & Skull Processing

  • Skull stripping applied to both:
    • T1-weighted images
    • Diffusion images

Data Organization

  • Scripts included:

    • Prep_diffusion.sh → diffusion preprocessing pipeline
    • Prep_face_removal.sh → skull stripping
    • example.sh → full pipeline execution

  • Designed as a reproducible and automated workflow

Data Privacy

  • All personal identifiers removed
  • Acquisition timestamps anonymized:
    • Synthetic timestamps used
    • Only relative timing preserved

Usage Agreement

Users are free to:

  • Share and adapt the data
  • Use for research and publication

With requirement:

  • Proper attribution

Scientific Value

This dataset is particularly useful for:

  • Multi-site harmonization studies
  • Scanner variability analysis
  • Diffusion metric reliability evaluation
  • Method validation (e.g., tractography, connectomics)
  • Benchmarking preprocessing pipelines

Contact

Assets 148
Loading