Abnormal higher-order network interactions in Parkinson's disease visual hallucinations.

This article focuses on altered brain network organization underlying visual hallucinations in Parkinson's disease, investigated through advanced dimensionality reduction techniques applied to resting-state functional MRI data. The study analyzed 77 Parkinson's disease patients (31 with visual hallucinations, 46 without) and 19 healthy controls, revealing that patients with hallucinations exhibit compression of the unimodal-heteromodal gradient, indicating increased functional integration between sensory and higher-order networks, particularly between the visual and default mode networks. Complementary analysis using t-distributed stochastic neighbour embedding (t-SNE) identified distinct alterations in prefrontal and other regions, suggesting additional complex network reconfigurations not captured by traditional functional connectivity methods. These findings support a systems-level model where disrupted hierarchical brain network communication contributes to hallucination susceptibility and highlight dimensionality reduction as a valuable tool for elucidating neural signatures of perceptual disturbances across neuropsychiatric disorders.