Skip to content
QSMxT

QSMxT

Quantitative Susceptibility Mapping, in Rust — a fast, BIDS-native pipeline and interactive TUI.
Get startedTry in Neurodesk
Terminal window
curl -fsSL https://raw.githubusercontent.com/QSMxT/QSMxT/main/install.sh | sh

See the installation guide for other options, including building from source.

Why QSMxT

Section titled “Why QSMxT”

End-to-end pipeline

Masking, phase unwrapping, echo combination, background field removal, dipole inversion, and referencing — orchestrated for you by a single self-contained binary, with defaults that work on human brain GRE data as they come.

50+ algorithms

Swap the method at any stage: six masking recipes, ten background-field removers, 20 classical and 11 deep-learning dipole inversions, and eight source-separation methods — all provided by QSM.rs.

Interactive TUI

Convert, configure, and run from a polished terminal interface — the recommended way to drive QSMxT, no config file required.

BIDS-native

Finds phase and magnitude, reads JSON sidecars, and writes compliant derivatives to derivatives/qsmxt/. Starting from DICOMs, dicom-convert classifies the series for you.

Built for scale

Disk caching skips completed steps on re-runs, memory-aware parallelism fills your cores, and SLURM scripts fan out across a cluster.

Get going

Section titled “Get going”
InstallationOne-line install for Linux, macOS, and Windows.
Quick startFrom raw data to QSM maps in minutes.
Running interactivelyConvert, configure, and run from the TUI — the recommended path.
Coming from QSMxT 8.xWhat changed in the Rust rewrite, and where to find 8.x.

How do I cite QSMxT?

Section titled “How do I cite QSMxT?”

If QSMxT is useful in your research, please cite:

Stewart AW, Robinson SD, O’Brien K, et al. “QSMxT: Robust masking and artifact reduction for quantitative susceptibility mapping.” Magnetic Resonance in Medicine 87.3 (2022): 1289–1300. doi.org/10.1002/mrm.29048

Every run also writes a methods.md to the derivatives directory: a methods paragraph describing what ran, with the citations for the specific methods your data and settings exercised, and pipeline.svg, a picture of the pipeline. A run that ended up processed differently from the rest (a supplied mask it could not use, say) gets its own methods text in changes/.