State entropy reveals the reconfiguration of brain complexity in Parkinson’s disease

The brain is a complex dynamic system comprising multiple information-processing units. The characterization of brain systems and understanding of their organization and reconfiguration in different states remains an ongoing challenge. This study aims to explore the organization of brain complexity in Parkinson’s disease (PD). In this study, we developed a framework to measure the complexity of each unit and explore aberrant patterns in PD. The resting-state blood oxygen level-dependent signals were converted into a state series with spatiotemporal coding of the whole brain system, and subsequently calculated the Shannon entropy of the state series (referred to as state entropy). In a total of 152 participants (PD patients n ​= ​73, normal controls n ​= ​79), we revealed a descent entropy gradient from the neocortex, particularly the default mode network (DMN), to subcortical systems. Subsequent application of a two-sample t -test and multiple comparison correction revealed that the state entropy of the visual, sensorimotor, and limbic networks was significantly increased in the PD group. Additionally, we obtained the person-based similarity index (PBSI) of state entropy within each network to measure individual variability, finding that patients with PD have decreased PBSI scores on the visual, sensorimotor, and limbic networks, and increased PBSI of entropy on the DMN. Notably, when testing the classification performance of state entropy and PBSI metrics, we found that individual variability was more effective in recognizing patients with PD. Our study reveals the organization of brain complexity across healthy and PD states, and suggests that personalized measures of brain reconfiguration have the potential to help develop clinical assistant diagnostic systems.