TFI is single-step: it inverts the total field directly (its own background removal). Larger preconditioner suits data with strong background fields (e.g. air near sinuses).
⚠ QSMART needs a tight brain mask. Build the mask with BET in the masking step — a loose threshold mask leaks non-brain signal into the dipole inversion and causes heavy streaking artifacts.
Parameters in mm - auto-scaled to voxels
BET's surface model is tuned for the human brain in millimetres, so on a mouse brain its search distances span the whole head. Mouse BET multiplies the voxel sizes by the scale factor below before running BET, which brings the brain to roughly human size; the mask comes back on the original grid.
RS2-Net is a deep-learning brain extractor trained on rat and mouse MRI from 89 centres.
First use downloads 63 MB of model weights under GPL-3.0, then caches them in your browser.
Patch size is fixed at 128 × 96 × 128 to fit browser memory.
Scaled BET runs BET on voxel sizes × 10: instant, but may leak into the skull base.
First use downloads 123 MB of model weights under CC-BY-NC-4.0, then caches them in your browser.
Smaller steps overlap more patches to reduce seams.
Patch size is fixed at 128 × 128 × 64 to fit browser memory.
Run this command with QSMxT to process a full BIDS dataset with your current settings:
Auto-generated methods section with citations based on your pipeline settings:
Quantitative Susceptibility Mapping in the browser
QSMbly is a browser-based application for Quantitative Susceptibility Mapping (QSM). It runs entirely in your web browser using WebAssembly, providing near-native performance without requiring any software installation or data upload.
QSMbly runs the same reconstruction code as QSMxT, the command-line QSM pipeline: both are built on QSM.rs. Try methods here, then scale up with QSMxT when you have a whole study to process.
QSMbly runs entirely in your web browser using WebAssembly. All processing occurs locally on your machine.
QSMbly compiles its processing algorithms from Rust to WebAssembly (WASM), which runs natively in the browser at near-native speed. Your NIfTI files are read directly into browser memory, processed locally, and results are displayed without any network transfer.
QSMbly uses Cloudflare Web Analytics to collect anonymous page-view statistics (e.g. number of visitors, country, referrer). This does not use cookies, does not track individual users, and does not have access to any of your data or processing results. It helps us understand how many people use QSMbly and where they come from.
QSMbly does not use cookies, fingerprinting, or any other tracking. Besides the application itself and the analytics beacon, the page downloads code and data from these services; none of them receives your images:
Because of these downloads, deep-learning methods and dicompare need an internet connection the first time they are used.