Niivue

An interactive WebGL based neuroimaging viewer built on top of Bonito.jl and NiiVue.

Installation

This package is not registered yet. Installation can be performed via

import Pkg
Pkg.add(url="https://github.com/korbinian90/Niivue.jl")

Quickstart

This should work in almost any julia environment (REPL, vscode, Pluto, Jupyter)

using Niivue

arr = rand(50,50,20)
nv = niivue(arr)

Use settings

using Niivue

# initialization settings
volumes = [
    Dict(
        :url => "https://niivue.github.io/niivue-demo-images/mni152.nii.gz",
        :colormap => "gray",
    ),
    Dict(
        :url => "https://niivue.github.io/niivue-demo-images/spmMotor.nii.gz", 
        :colormap => "redyell",
        :opacity => 0.5,
    )
]

nv = niivue(volumes)

# interactively adapt settings
nv.setCrosshairWidth(5)
nv.setCrosshairColor([0,1,1,0.5])
nv.isColorbar = true

The methods (e.g. nv.setCrosshairWidth(5)) and options (e.g. nv.isColorbar = true) can be found in the niivue javascript documentation.

Electron Display

To use an Electron display instead of the browser, add in the beginning

using Electron
use_electron_display(devtools=false)

In vscode, by default the plot pane is used. To use the Electron display, deactivate the plotting pane via the command palette (ctrl+shift+p) -> Julia: Disable Plot Pane

Examples

Have a look at the 'examples' subfolder.

Hacking

Javascript can be directly executed from Julia and the nv object is available as window.nv

using Niivue

nv = niivue()
width = 10
js = Niivue.Bonito.js"""
window.nv.setCrosshairWidth($(width))
"""

Niivue.Bonito.evaljs(nv.app.session.x, js)

Return a value from javascript to julia:

using Niivue

nv = niivue()
js_return = Niivue.Bonito.js"""
window.nv.colormaps()
"""

cmaps = Niivue.Bonito.evaljs_value(nv.app.session.x, js_return)

Limitations

• some functions don't work with the Julian Syntax yet and need to be called via the javascript "hacking" way

Future plans

• use niivue-vscode as alternative pre-configured viewer
• Support access to volumes (e.g. nv.volumes[1].opacity = 0.3) and functions with return values (e.g. cmaps = nv.colormaps()) directly from julia
• observable in julia that stores the crosshair location

Acknowledgements

Development of an initial prototype was supported by Chris Rorden (University of South Carolina)