The past decade has seen several attempts to employ the entropy of neuroimaging signals as a potential biomarker for cognitive decline or traumatic brain injury (C. Y. Liu et al. 2013; Adhikari et al. 2017; Li et al. 2018). Not all these studies properly account for the distributed nature of cognition, however, which raises the possibility of erroneous estimates of global entropy. This thesis proposes a novel means of estimating the complexity of fMRI signals and demonstrates its efficacy in detecting the effects of psychiatric disease on neuroimaging signals. The method determines the minimum number of orthogonal dimensions necessary to capture nonrandom signal dynamics, then projects the dynamic functional connectivity signal into the resultant low-dimensional space. In this space, the dynamic functional connectivity signal¿s entropy may be estimated along each dimension independently and summed to find the total entropy per subject, thus avoiding the need to estimate interregional effects. Tests on two independently collected datasets indicate that this pipeline can distinguish between healthy controls and psychiatric patients, and that a Hopf bifurcation-based effective connectivity model is able to recover meaningful differences between control and patient groups when trained in this space.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados