Skip to content

niimath

View as Markdown

niimath is a fast image calculator for neuroimaging, a clone of fslmaths. @niivue/niimath runs it as WebAssembly in a Web Worker, so pages can process images without a server. It is a separate package: NiiVue loads what it produces like any other file.

Pick a volume, or drop a NIfTI on the viewer; each pipeline runs as soon as you pick it.

Install

Terminal window
npm install @niivue/niimath

Process a volume

import { Niimath } from '@niivue/niimath'
const niimath = new Niimath()
await niimath.init() // loads the WASM once
const smoothed = await niimath.image(file).s(3).run('smoothed.nii.gz')
await nv.loadVolumes([{ url: smoothed }])

image() takes a File, a Blob or raw bytes. run() resolves to a File named after its argument, so it loads straight into NiiVue, downloads as is, or feeds the next pipeline.

Options as an array

A pipeline held as data, such as a menu or a saved recipe, passes its options as they would follow the input on the command line:

const steps = ['-s', 3, '-thrp', 50, '-bin']
const mask = await niimath.image(file).args(steps).run('mask.nii.gz')

Meshes

mesh() runs marching cubes and writes a mesh format NiiVue reads:

const surface = await niimath.image(file).mesh({ i: 'm', b: 1 }).run('brain.mz3')
await nv.loadMeshes([{ url: surface }])

Use this when you need geometry: to export, edit or measure a surface. To only show a thresholded structure, a colormap that is transparent below the threshold renders it from the volume, with no mesh.

One Niimath runs one pipeline at a time: wait for each run(), or create one per concurrent job. See the niimath package for every operator.