Electrophysiological biomarkers reflecting the pathological activities in the basal ganglia are essential to gain an etiological understanding of Parkinson's disease (PD) and develop a method of diagnosing and treating the disease. Previous studies that explored electrophysiological biomarkers in PD have focused mainly on oscillatory or periodic activities such as beta and gamma oscillations. Emerging evidence has suggested that the nonoscillatory, aperiodic component reflects the firing rate and synaptic current changes corresponding to cognitive and pathological states. Nevertheless, it has never been thoroughly examined whether the aperiodic component can be used as a biomarker that reflects pathological activities in the basal ganglia in PD. In this study, we examined the parameters of the aperiodic component in hemiparkinsonian rats and tested its practicality as an electrophysiological biomarker of pathological activity. We found that a set of aperiodic parameters, aperiodic offset and exponent, were significantly decreased by the nigrostriatal lesion. To further prove the usefulness of the parameters as biomarkers, acute levodopa treatment reverted the aperiodic offset. We then compared the aperiodic parameters with a previously established periodic biomarker of PD, beta frequency oscillation. We found a significantly low negative correlation with beta power. We showed that the performance of the machine learning-based prediction of pathological activities in the basal ganglia can be improved by using both beta power and the aperiodic component, which showed a low correlation with each other. We suggest that the aperiodic component will provide a more sensitive measurement to early diagnosis PD and have the potential to use as the feedback parameter for the adaptive deep brain stimulation.
Keywords: Aperiodic component; Machine learning; Parkinson's disease; Prediction of dopamine depletion.
Copyright © 2022 The Authors. Published by Elsevier Inc. All rights reserved.