Abstract
Olfactory dysfunction is an early sign of such neurodegenerative diseases as Parkinson's (PD) and Alzheimer's (AD), and is often present in Mild Cognitive Impairment (MCI), a precursor of AD. Understanding neuro-temporal relationships, i.e., functional connectivity, between olfactory eloquent structures in such disorders, could shed light on their basic pathophysiology. To this end, we employed region-based analyses using resting-state functional magnetic resonance imaging (rs-fMRI) obtained from cognitively normal (CN), MCI, and PD patients with cognitive impairment (PD-CogImp). Using machine learning (linear and ensemble learning), we determined whether the identified functional patterns could classify abnormal function from normal function. Olfaction, as measured by objective testing, was found to be most strongly associated with diagnostic status, emphasizing the fundamental association of this primary sensory system with these conditions. Consistently lower functional connectivity was observed in the PD-CogImp cohort compared to the CN cohort among all identified brain regions. Differences were also found between PD-CogImp and MCI at the level of the orbitofrontal and cingulate cortices. MCI and CN subjects had different functional connectivity between the posterior orbitofrontal cortex and thalamus. Regardless of study group, males showed significantly higher connectivity than females in connections involving the orbitofrontal cortex. The logistic regression model trained using the top discriminatory features revealed that caudate was the most involved olfaction-related brain structure (accuracy = 0.88, Area under the Receiver operator characteristic curve of 0.90). In aggregate, our study demonstrates that resting functional connectivity among olfactory eloquent structures has potential value in better understanding the pathophysiology of several neurodegenerative diseases.