以吞嚥及呼吸感測器偵測吞嚥時之咽部上提與吞嚥行為演算法開發

Abstract

目前使用於喉部的穿戴裝置，已顯示可監測到吞嚥的訊號，但多數研究僅是示範性質，並未深入研究訊號與實際吞嚥過程之間的相關性。本研究目標是使用具高度可撓性、柔軟性的吞嚥感應貼布，作為吞嚥監測的非侵入式穿戴感測器，並結合呼吸感測器量測的呼吸訊號，開發可觀察吞嚥生理訊號的穿戴型體外監測裝置。本研究亦開發一套資料擷取分析系統，用以研究呼吸、舌骨及甲狀軟骨之特徵波形，並建立呼吸結合吞嚥之分析演算法，為吞嚥評估之標準。本研究首先對吞嚥感應貼布進行性能測試，以抗拉伸實驗驗證吞嚥感應貼布具高柔性的特性，平均楊氏模數達1.7324 MPa，這代表其貼在皮膚表面時，不會限制吞嚥時所造成的皮膚自然運動。重複性實驗結果顯示，吞嚥感應貼布在長時間使用中能維持穩定的壓電輸出特性。本研究以擬喉部裝置模擬吞嚥行為驗證吞嚥感應貼布與實際人體吞嚥行為的相關性。本研究亦將此裝置進行人體實驗驗證，並以即時喉部吞嚥影像來驗證吞嚥感應貼布量測訊號的準確性。實驗結果顯示，吞嚥感應貼布能準確捕捉個人吞嚥動作的不同波形，並識別特徵時間點，成功比出較年輕與年長健康受試者的訊號差異，分析出舌骨及甲狀軟骨移動至最大位移的時間點在呼吸暫停區間內的比例分別為 年輕男性93.33%及80%、年輕女性80%及93.33%、年長男性為80.56%及97.22%、年長女性為95.56%及97.78%。本研究亦針對重複唾液吞嚥實驗開發演算法，其準確率在年輕與年長健康男女性受試者中監測吞嚥次數的誤差在+/-1次以內皆可達到100%。同時，本研究的實驗結果亦顯示每位受試者皆都有其特徵的吞嚥表現波型，這將可對臨床上的分析提供了重要的依據。總結，本研究開發一種可監測吞嚥行為的穿戴裝置，並開發演算法可將吞嚥行為進行量化分析，並以實驗驗證此裝置具有作為即時監測吞嚥的個人化穿戴式裝置。

Current studies on wearable devices have demonstrated that swallow activities can be detected by attaching physical sensors to the larynx area. However, most studies only do demonstrations and have not conducted in-depth studies on the correlation between the signals and the actual swallowing process. The aim of this study is to use a highly flexible and compliant swallow patch sensor (SPS) for non-invasive swallowing sensing. A respiratory sensor also is integrated to serve as a bio-feedback sensor. Using this swallowing sensing system, a data acquisition and analysis system also are developed to study the characteristic waveforms of respiration, and movements of hyoid bone and thyroid cartilage during the swallowing process. The combination of this swallowing monitoring system can become a standard method to assess the swallowing process. The compliance of the SPS is verified by measuring its average Young's modulus, and it was 1.7324 MPa. The experimental results also demonstrated that the SPS has good repeatability and can be used for an extended period of time. These results suggest that the SPS does not restrict the swallow-induced skin surface, and it can monitor swallow activities for long-term usage. Furthermore, a 3D-printed thyroid cartilage is used to study the signal correlation with the movement of the thyroid cartilage, and it is applied to human studies. Our studies demonstrate that the SPS can accurately capture the different waveforms of each individual’s swallowing movements. The signal differences between young and senior healthy subjects can be distinguished. The time for the hyoid bone and thyroid cartilage moved to the farthest point can be compared with the apnea period, which is 93.33% and 80% for young men, 80% and 93.33% for young women, 80.56% and 97.22% for senior men, and 95.56% and 97.78% for senior women. This study also developed an algorithm for repeated saliva-swallowing experiments. Its accuracy can reach 100% +/- 1 error. Finally, the experimental results of this study also show that each subject has his own characteristic swallowing performance wave pattern, which will provide an important basis for clinical analysis. In summary, this study developed a wearable device and an algorithm to quantitatively analyze swallowing behavior for the application of a personalized wearable device.