Spatial Patterns for Discriminative Estimation (SPADE)

Abstract

Functional connectivity between brain regions is modulated by cognitive states or experimental conditions. A multivariate methodology that can capture fMRI connectivity maps in light of different experimental conditions would be of primary importance to learn about the specific roles of the different brain areas involved in the observed connectivity variations. Here we detail, adapt, optimize and evaluate a supervised dimensionality reduction model to fMRI timeseries. We demonstrate the strength of such an approach for fMRI data using data from the Human Connectome Project to show that the model provides close to perfect discrimination between different fMRI tasks at low dimensionality. The straightforward interpretability and relevance of the model results is demonstrated by the obtained linear filters relating to anatomical areas well known to be involved on each considered task, and its robustness by testing discriminatory generalization and spatial reproducibility with respect to the number of subjects and fMRI time-points acquired. We additionally suggest how such approach can provide a complementary view to traditional task fMRI analyses by looking at changes in the covariance structure as a substitute to changes in the mean signal. We conclude that the presented methodology provides a robust tool to investigate brain connectivity alterations across induced cognitive changes and has the potential to be used in pathological or pharmacological cohort studies. A publicly available toolbox is provided to facilitate the end use and further development of this methodology to extract Spatial Patterns for Discriminative Estimation (SP{spadesuit}DE).