BrainSpace: a toolbox for the analysis of macroscale gradients in neuroimaging and connectomics datasets

Reinder Vos De Wael,Ting Xu,Seok-Jun Hong,Georg Langs,Daniel Margulies,Jessica Royer,Oualid Benkarim,Shahin Tavakol,Bratislav Misic,Sara Lariviere,Sofie Valk,Jonathan Smallwood,Casey Paquola,Boris C Bernhardt,Michael Milham

doi:10.1038/s42003-020-0794-7

Abstract

Understanding how cognitive functions emerge from brain structure depends on quantifying how discrete regions are integrated within the broader cortical landscape. Recent work established that macroscale brain organization and function can be described in a compact manner with multivariate machine learning approaches that identify manifolds often described as cortical gradients. By quantifying topographic principles of macroscale organization, cortical gradients lend an analytical framework to study structural and functional brain organization across species, throughout development and aging, and its perturbations in disease. Here, we present BrainSpace, a Python/Matlab toolbox for (i) the identification of gradients, (ii) their alignment, and (iii) their visualization. Our toolbox furthermore allows for controlled association studies between gradients with other brain-level features, adjusted with respect to null models that account for spatial autocorrelation. Validation experiments demonstrate the usage and consistency of our tools for the analysis of functional and microstructural gradients across different spatial scales.

Highlights

Understanding how cognitive functions emerge from brain structure depends on quantifying how discrete regions are integrated within the broader cortical landscape
Much of the more recent work linking measures of neural processing to cognition has focused on identifying discrete regions and modules and their specific functional roles[10], recent conceptual and methodological developments have provided the data and methods that allow macroscale brain features mapped to low dimensional manifold representations, described as gradients[11]
This section illustrates the usage of BrainSpace for gradient mapping and null model generation

Summary

Introduction

Understanding how cognitive functions emerge from brain structure depends on quantifying how discrete regions are integrated within the broader cortical landscape. Gradients have been successfully derived from non-imaging data that were registered to stereotaxic space, including hippocampal post mortem gene expression information[26] and 3D histology data[22], to explore cellular and molecular signatures of neuroimaging and connectome measures Core to these techniques is the computation of an affinity matrix that captures inter-area similarity of a given feature followed by the application of dimensionality reduction techniques to identify a gradual ordering of the input matrix in a lower dimensional manifold space (Fig. 1). The goal of this paper is to present an open-access set of easy-to-use tools that allow the identification, visualization, and analysis of macroscale gradients of brain organization We hope this will provide a method for calculating cortical manifolds that facilitates their use in future empirical work, allows comparison between studies, and allows for result replicability. We anticipate that our toolbox will assist researchers interested in studying gradients of cortical organization, and propel further work that establishes the overarching principles through which structural and functional organization of human and nonhuman brains gives rise to key aspects of cognition

Objectives

Methods

Results

Conclusion