EVOKED PHASE COHERENCE AS A BIOMARKER FOR ADAPTIVE NEUROMODULATION IN RAT MODEL OF PARKINSON'S DISEASE

Sammanfattning: Neuromodulation, such as spinal cord stimulation (SCS) and deep brain stimulation (DBS), has been shown to modulate pathophysiological brain activity and provide symptomatic therapy for several neurological disorders, including Parkinson’s Disease. The effectiveness of this therapy could likely be further improved by neuromodulation that is adaptive, delivering stimulation more selectively, by monitoring a biomarker in recorded brain signals, which indicates the presence of a pathological state. In the treatment of Parkinson’s Disease, the most commonly proposed solutions for adaptive neuromodulation are relying on excessive beta-band oscillatory activity as a biomarker, which is however often highly variable between patients during movement and in conjunction with neuromodulatory treatment, such as levodopa. These limitations hinder broader use of this biomarker and prompts further research for alternative solutions. In this work, we instead present the use of a novel feature of evoked electrophysiological activity, which utilizes the inter-trial phase coherence between stimulation pulses, to classify parkinsonian brain states in 6-OHDA lesioned rats. We developed a method, which relates to the rate of decay in inter-tral phase coherence, evoked by single SCS or DBS pulses, that is able to statistically separate experimental conditions recorded from a dopaminergic depleted hemisphere from conditions a non-depleted hemisphere, while also being able to separate conditions with levodopa treatment from conditions without treatment. For animals undergoing SCS we can classify phase decay measurements from pharmacologically treated or untreated parkinsonian states, using a Bayesian model, with a high accuracy and strong classifier performance for a single channel (AUC 0.85 – 0.99) in the motor cortex and striatum. In ongoing experiments, similar implementation of adaptive DBS is being evaluated. Our results support the implementation of our feature in a protocol aimed at performing closed-loop neuromodulation in the 6-OHDA rat model of Parkinon’s Disease, that can serve as the basis for further studies.