Real-Time Decoding of Leg Motor Function and Dysfunction from the Subthalamic Nucleus in People with Parkinson’s Disease
摘要
We developed a real-time decoding framework that can accurately predict leg motor functions, as well as key aspects of walking, from local field potentials (LFP) recorded from the subthalamic nucleus of patients with Parkinson’s disease. Concretely, we designed decoders that can predict locomotor states, gait events, modulations in force during obstacle avoidance, and freezing of gait episodes while participants walked freely in unconstrained conditions. Our algorithms employed the full spectrum of LFP recorded bilaterally, either through externalized deep brain stimulation (DBS) leads connected to an external, high-resolution amplifier (six bipolar channels, Fs = 8 kHz), or wirelessly using a last-generation implantable stimulator with sensing capabilities (Percept PC, Medtronic, two bipolar channels, Fs = 250 Hz). These results represent the first neural decoding of leg motor function operating in real-time from therapeutically implanted DBS electrodes. Considering the large number of patients treated worldwide with DBS implants, as well as the capabilities of newest commercial stimulators, our results pave the way for the design and widespread deployment of closed-loop neuromodulation therapies that address gait deficits with new closed-loop approaches.