Clusterwise Independent Component Analysis (C-ICA): Using fMRI resting state networks to cluster subjects and find neurofunctional subtypes

•C-ICA is a new method for discovering neurofunctional subtypes from rs-fMRI data.•Patients are clustered based on differences in ICA-derived resting state networks.•Neurofunctional subtypes may increase the understanding of disease heterogeneity.•We successfully validated C-ICA in simulation studies and an empirical study.•C-ICA shows a better clustering performance compared to competing clustering methods.