EAG(electroarthrography)신호를 사용하는 딥러닝 기반 무릎 운동 패턴 분류

Abstract

Electroarthrography(EAG) is a technology that uses electrodes attached to the knee surface to measure and detect the flow potential generated in the articular cartilage of the knee joint under load and to observe joint load through this. In this paper, an experiment was conducted to confirm and establish whether there are differences in characteristics depending on the operation of the EAG signal. In this paper, we collected and analyzed EAG signals based on previously unexplored movements and electrode positions. The characteristics of EAG were comprehensively analyzed and established using three movements: sitting posture, straight legs, limited parallel squat, and passive angle change according to knee angle change. In addition, we proposed a method to classify differences in EAG signals according to movement using CNN(Convolutional Neural Networks), a deep learning technology. This paper achieved 93.59% accuracy in distinguishing EAG signals corresponding to various movements through deep learning, which provides evidence for the distinct characteristics of EAG signals depending on movement. This is expected to serve as a basis for verifying the feasibility of using EAG in actual clinical practice in the future, provide a deeper understanding of knee joint health and exercise effects, and become the basis for suggesting new directions.