nltools

Python toolbox for multivariate neuroimaging
Author: Luke Chang (luke.j.chang@dartmouth.edu)

Neurolearn tools (nltools) is a neuroimaging analysis toolbox developed in Python. It is a collection of tools to easily conduct multivariate imaging analyses such as prediction/classification, decomposition, representational similarity analysis (RSA), and functional alignment. The toolbox is developed and maintained by Luke Chang’s computational social and affective neuroscience lab at Dartmouth College. The toolbox is a partial python port of Tor Wager’s object-oriented Matlab toolbox and heavily leverages code from nilearn and sci-kit learn.

The toolbox is based around several data classes (e.g., 4-D neuroimaging objects, 4-D adjacency objects, and 2-D design matrices). There are many functions to help with basic data I/O from files and websites, manipulating data, and analyzing data. See our API for a list of functionality or our tutorials for walkthrough examples in the form of Jupyter notebooks.

Functionality

• Load nifiti files from files or urls
• Univariate Regression
• Multivariate Prediction/Classification
• Representational Similarity Analyses (RSA)
• Functional Alignment (e.g., hyperalignment, shared response model)
• Decomposition (e.g., PCA, factor analysis, NNMF)