The QSMxT Ecosystem

A family of open tools for Quantitative Susceptibility Mapping — turning MRI phase into maps of tissue magnetic susceptibility. The QSMxT pipeline and the QSMbly browser app share one Rust engine (QSM.rs), so a result you get in the browser is the same one you'd get on a cluster. Alongside them, the QSM-CI benchmark scores QSM reconstruction methods of any kind against ground truth.

QSM.rs— the engine — powers QSMxT& QSMbly ·  QSM-CIbenchmarks QSM methods
QSM.rs
The engine

A Rust library implementing 40+ QSM algorithms — brain extraction, phase unwrapping, background field removal, dipole inversion, and source separation. QSMxT and QSMbly both call into it, so they produce identical results.

Explore the library →
QSMxT
The pipeline

A fast, BIDS-native command-line pipeline and interactive TUI. Takes raw scanner DICOMs to susceptibility maps in a single self-contained binary, built for batch processing and clusters.

Read the docs →
QSMbly
The browser app

The whole pipeline compiled to WebAssembly, running entirely in your browser. No install, no upload — your data never leaves your machine. The fastest way to experiment with QSM methods interactively.

Launch the app →
QSM-CI
The benchmark

An open challenge and live leaderboard for QSM reconstruction. Submit an algorithm in any language; it runs against phantom ground truth and is scored on the same metrics, stage by stage and end to end.

See the leaderboard →

Why one engine matters

In QSM, the same data can give very different maps depending on which unwrapping, background-removal, and inversion methods you chain together. When every tool reimplements those methods, results stop being comparable. Here they don't: QSMbly in the browser and QSMxT on a cluster both call the same QSM.rs functions, so you can experiment in the browser, scale up on the command line, and trust that the numbers line up. And when you want to know how any method actually performs, QSM-CI scores QSM reconstructions — in any language, QSM.rs's among them — against ground truth.