Impact of referencing scheme on decoding performance of LFP-based brain-machine interface

摘要

Objective. There has recently been an increasing interest in local field potential (LFP) forbrain-machine interface (BMI) applications due to its desirable properties (signal stability andlow bandwidth). LFP is typically recorded with respect to a single unipolar reference which issusceptible to common noise. Several referencing schemes have been proposed to eliminate thecommon noise, such as bipolar reference, current source density (CSD), and common averagereference (CAR). However, to date, there have not been any studies to investigate the impact ofthese referencing schemes on decoding performance of LFP-based BMIs. Approach. To address thisissue, we comprehensively examined the impact of different referencing schemes and LFP featureson the performance of hand kinematic decoding using a deep learning method. We used LFPschronically recorded from the motor cortex area of a monkey while performing reaching tasks.Main results. Experimental results revealed that local motor potential (LMP) emerged as the mostinformative feature regardless of the referencing schemes. Using LMP as the feature, CAR wasfound to yield consistently better decoding performance than other referencing schemes overlong-term recording sessions. Significance. Overall, our results suggest the potential use of LMPcoupled with CAR for enhancing the decoding performance of LFP-based BMIs.