Swallowing Events Recognition Method Based on Complex Impedance Pharyngography and Nyquist Plots

Abstract

Early screening is of great significance to reduce the incidence of dysphagia, and accurate recognition of swallowing events (SEs) becomes an inevitable step during the screening and treatment of dysphagia. Impedance pharyngography (IPG) has been developed as a promising noninvasive SE detection method in the last 20 years, but the conventional IPG technique can only measure impedance amplitude and misses the equally important phase information. To further improve the analysis capability of the IPG technique, this article proposes a new complex IPG (CIPG) detection method, which can dynamically extract the complex impedance (impedance amplitude and phase) information during the swallowing process. First, the CIPG measurement principle based on integer-period digital lock-in amplifier (IP-DLIA) is proposed, and an FPGA-based CIPG system is built. Then, an intelligent algorithm for SE recognition is designed in which the Nyquist plots (impedance real part versus imaginary part) are adopted as 2-D-image inputs of the residual nets convolutional neural network (CNN) algorithm. Finally, an SE recognition experiment is performed, in which different SEs, including dry swallowing, drinking water, drinking yogurt, eating bread, and coughing, are recorded using the CIPG system. The results show that the overall SE recognition accuracy reaches 97.8%, which is much higher than those based on the traditional IPG method or other sensors. This study demonstrates the effectiveness and superiority of the proposed CIPG technology and the intelligent SE recognition algorithm and lays theoretical and technical foundations on further developing an early screening method for dysphagia based on CIPG.