The lesion must be on the MNI152 1 mm grid; normalize it with SYNcro first.
Choose an example or drop a lesion map to begin.
Scroll to change slices · drag to adjust the view
Choose an example or drop a lesion map to begin.
A lesion interrupts more of the brain than it destroys: fibres passing through it lose their connection even where the tissue beyond is intact. This app scores that disconnection. It intersects a lesion map with a population tractography atlas and reports, for each white-matter bundle, the fraction of streamlines that pass through the lesion.
The lesion map must already be on the MNI152 1 mm template grid, 182 × 218 × 182; normalize it with SYNcro first if it is not. Without an anatomical scan the MNI152 template is shown behind the lesion.
Two atlases are offered. The ENIGMA Symmetric White Matter Tractography Atlas (65 bundles) is the default: it is built to be left–right symmetric, so a bundle and its contralateral twin are directly comparable. The HCP1065 population-averaged atlas (87 bundles) covers more bundles, including cranial nerves and brainstem pathways. They are different parcellations, not two measurements of one thing, so their bundle names and counts do not line up; report which one you used.
The arithmetic runs locally in WebAssembly (nii2tvx), on the full-resolution atlas, so the numbers match the command-line tool exactly. The streamlines you see are a decimated sample of the same atlas, kept small enough to download; do not count fibres on screen and expect the reported fraction.
The same analysis runs from a terminal: exes/nii2tvx in the Neurodesk webapps repository builds with make, and the dataset README shows the two commands. The browser and the command line produce the same table.
Research use only. A population-averaged atlas is not this patient's anatomy, and disconnection scored this way is a prediction, not a measurement.
Images are processed in this browser and are never uploaded. The tract atlas and the example scans are downloaded from Hugging Face at a pinned revision.