iSparse kmeans: a two-step clustering approach for big dynamic functional network connectivity data

Abstract BackgroundDynamic functional network connectivity (dFNC) estimated from resting-state functional magnetic imaging (rs-fMRI) studies the temporally varying of functional integration between brain networks. In a typical dFNC pipeline, a clustering stage to summarize the connectivity patterns that are transiently but reliably realized over the course of a scanning session. However, identifying the right number of clusters through a conventional clustering criterion computed by running the algorithm repeatedly, over a large range of cluster numbers is time-consuming and requires substantial computational power even for typical dFNC datasets, and the computational demands become prohibitive as datasets become larger and scans longer. Here we developed a new dFNC pipeline, called iterative sparse kmeans or iSparse kmeans, to analyze large dFNC data without having access to huge computational power. MethodIn iSparse kmeans, we implement two-step clustering. In the first step, we randomly use a sub-sample dFNC data and identify several sets of states at different model orders. In the second step, we aggregate all dFNC states estimated from all iterations in the first step and use this to identify the optimum number of clusters using the elbow criteria. Additionally, we use this new reduced dataset and estimate a final set of states by performing a second kmeans clustering on the aggregated dFNC states from the first k-means clustering. To validate the reproducibility of iSparse kmeans, we analyzed four dFNC datasets from the human connectome project (HCP). ResultsWe found that both conventional kmeans and iSparse kmeans generate similar brain dFNC states while iSparse kmeans is 27 times faster than the traditional method in finding the optimum number of clusters. We show that the results are replicated across four different datasets from HCP. ConclusionWe developed a new analytic pipeline which facilitates analysis of large dFNC datasets without having access to a huge computational power source. We validated the reproducibility of the result across multiple datasets.