Articles

Swallowing Pattern Classification Method Using Multichannel Surface EMG Signals of Suprahyoid and Infrahyoid Muscles

Masahiro Suzuki, Makoto Sasaki, Katsuhiro Kamata, Atsushi Nakayama, Isamu Shibamoto, Yasushi Tamada
Vol. 9 (2020) p.10-20

The ability to fine-tune the movement of swallowing-related organs and change the swallowing pattern to fit the volume of a bolus, texture and the physical properties of the food to be swallowed is referred to as the swallowing reserve. In other words, it is the response capability of food swallowing to avoid choking and aspiration. Herein, we focus on the coordination of the suprahyoid and infrahyoid muscles activities, which are closely related to swallowing movement, as a first step to develop a method to evaluate swallowing reserve, which declines due to neuromuscular disease, muscle weakness caused by aging, to mention a few. First, using two 22-channel electrodes, we measured the surface electromyography (sEMG) signals of suprahyoid and infrahyoid muscles during the following four swallowing conditions: combining two bolus volumes (3 and 15 mL water) and two techniques (normal and effortful swallow). Then, we verified whether the difference in swallowing patterns based on swallowing conditions can be classified from sEMG signals using three machine learning methods; namely, the real-time classification, comprehensive classification, and image recognition method. In the real-time classification method, the mean classification accuracy (MCA) for the four swallowing conditions was as low as 81.5%, indicating that the difference between swallowing conditions performed in a period of approximately 1 s cannot be classified sufficiently by this method. In the comprehensive classification method that applies a majority decision to all the classification results from the start to the end of swallowing, which can be obtained every 16 ms, MCA was 95.1%. Furthermore, in the image recognition method, the change of a series of sEMG signals in the swallowing movement was converted into swallowing pattern image, and the images were classified using a combination of deep convolutional neural networks and support vector machine (SVM). Compared with the comprehensive classification method, the number of training samples for the image recognition method was only 1/26, but the MCA reached 95.7%. This method, which can noninvasively evaluate swallowing patterns that change slightly based on swallowing conditions, could be applied to early detection of reduced swallowing function or a state of frailty (dysphagia potential) in aged individuals.

READ FULL ARTICLE ON J-STAGE