A Human Tissue Complex Impedance Measurement System for Swallowing Action Recognition

Abstract

Electrical impedance myography (EIM) is a noninvasive, painless, rapid, and low-cost method for assessing muscle health proposed in the past two decades. It operates by the nonintrusive injection of a weak, high-frequency current to acquire the electrical characteristics of muscle tissue, thereby determining its physiological properties. In this article, a portable human tissue complex impedance measurement system based on EIM is proposed. Through in-phase and quadrature (I/Q) demodulation, this system is capable of real-time display of the amplitude and phase of human tissue complex impedance on a PC. The reliability of the system was validated by measuring the complex impedance of the Fricke-Morse impedance models under excitations at different frequencies and of the relaxed human neck muscles under a single-frequency excitation across different environmental parameters. Furthermore, to reveal the correlation between muscle tissue complex impedance and muscle states, the swallowing action recognition model based on convolutional k-nearest neighbors (CKNNs) is constructed and deployed in the proposed system. With high accuracy and low floating point operations (FLOPs), CKNN performs effectively in swallowing action recognition. Compared to other swallowing action recognition systems, the proposed system exhibits reliable classification capabilities, achieving a recognition accuracy of 95.09%.