Intrinsic dimension estimation of the fMRI space via sparsity-promoting matrix factorization - Counting the `brain cores' of the human brain

Abstract

Functional Magnetic Resonance Imaging (fMRI) is a powerful non-invasive tool for localizing and analyzing brain activity. This study focuses on one very important aspect of the functional properties of human brain, specifically the estimation of the level of parallelism when performing complex cognitive tasks. Using fMRI as the main modality, the human brain activity is investigated through a purely data-driven signal processing and dimensionality analysis approach. Specifically, the fMRI signal is treated as a multi-dimensional data space and its intrinsic `complexity' is studied via sparsity-promoting matrix factorization in the sense of blind-source separation (BSS). One simulated and two real fMRI datasets are used in combination with Independent Component Analysis (ICA) for estimating the intrinsic (true) dimensionality via detection of statistically independent concurrent signal sources. This analysis provides reliable data-driven experimental evidence on the number of independent active brain processes that run concurrently when visual or visuo-motor tasks are performed. The results prove that, although this number cannot be defined as a hard threshold but rather as a continuous range, however when a specific activation level is defined, an estimated number of concurrent processes or the loose equivalent of `brain cores' can be detected in human brain activity.