Research on Local Field Potential Signal Classification Algorithm Based on Transfer Learning

The local field potential signals (LFPs), as a vital signal for studying the mechanisms of deep brain stimulation (DBS) and constructing adaptive DBS contain information related to the motor symptoms of Parkinson's disease (PD). This paper proposed a Parkinson's disease state recognition algorithm based on the idea of transfer learning.The algorithm uses continuous wavelet transform (CWT) to convert one-dimensional LFPs into two-dimensional gray-scalogram images and color images respectively, and adds a Bayesian optimized random forest (RF) classifier to replace the three fully connected layers used in the classification task in the VGG16 model, to realize the pathologic status identification of PD and normal state of parkinsonian patients. It was found that consistently superior performance of gray-scalogram images over color images. The proposed algorithm achieved an impressive accuracy of 97.76%, outperforming feature extractors such as VGG19, InceptionV3, ResNet50, and the lightweight network MobileNet. This algorithm has high accuracy and can monitor the status of patients in real time without manual feature extraction, and only apply DBS stimulation when in PD state, effectively improving the closed-loop adaptive DBS treatment effect.