Migrating from nirs-toolbox to MNE

This tutorial will demonstrate how to do convert a nirs-toolbox script to MNE and highlight differences and common issues you may encounter.

nirs-toolbox is a MATLAB based software analysis package. huppertt/nirs-toolbox

Basic nirs-toolbox script to be reproduced in MNE

Below is a common example analysis performed in nirs-toolbox. The NIRx data is loaded and trigger names are modified. Then the intensity signal is convert to optical density then converted to haemoglobin concentration. After which, a GLM analysis is performed with iteratively reweighted autoregressive noise model. Finally a mixed effects model is run on the group level dat.

# Load data
raw = nirs.io.loadDirectory()

# Preprocess data
pipeline_preprocess = nirs.modules.RenameStims(pipeline_preprocess);
pipeline_preprocess.listOfChanges = {
 'channel_1', 'Control';
 'channel_2', 'Condition1';
 'channel_3', 'Condition2'};
pipeline_preprocess = nirs.modules.DiscardStims(pipeline_preprocess);
raw = pipeline_preprocess.run(raw);

# Analyse data
pipeline = nirs.modules.Resample();                     pipeline.Fs = 3;
pipeline = nirs.modules.OpticalDensity(pipeline);
pipeline = nirs.modules.BeerLambertLaw(pipeline);
pipeline = nirs.modules.AR_IRLS(pipeline);
subj_stats = pipeline.run(raw);

# Group level analysis
pipeline = nirs.modules.MixedEffects();
pipeline.formula = 'beta ~ -1 + cond + (1|Name)';
group_stats = pipeline.run(subj_stats);

MNE equivalent of nirs-toolbox script

The MNE equivalent of this analysis is well described in the group level tutorial.

So this page will focus on the differences between the two approaches.

Differences between toolboxes

MNE does not have iterative reweighting of the noise estimate in the GLM analysis.

Further analysis details

If there is useful processing in the nirs-toolbox that is not available in MNE please let us know by creating an issue at mne-tools/mne-nirs#issues

Estimated memory usage: 462 MB