互動可程式之智慧衣心電圖監測系統

近年來大多數穿戴式設備都伴隨著生理訊號量測的功能，然而這些穿戴式設備都不包含心臟疾病即時分析功能。本研究以織物電極配合ADS1298類比前端擷取12導程心電訊號，並以TMS320c5515數位訊號處理器實現小波轉換對心電訊號進行分析，最後透過cc2541低功耗藍芽將分析結果傳輸至行動裝置進行顯示與儲存。為了降低硬體成本，本研究使用定點數(fixed-point)數位訊號處器做為運算核心，並以互動式的方式讓使用者選擇想要分析的心臟疾病，系統就會根據使用者選擇的選項從12導程中挑選特定的幾個導程進行心臟疾病分析。藉由上述的方式不僅可以減少處理器的負擔，同時也可以提高分析的準確率與專一性。目前本研究實現了陣發性心房顫動與心室期外收縮這兩種常見的心律不整的分析。本研究透過MIT-BIH arrhythmia database、MIT-BIH atrial fibrillation database 進行QRS波偵測與心臟疾病分類性能測試。由測試結果得知，MIT-BIH arrhythmia database與MIT-BIH atrial fibrillation database平均QRS波偵測準確率分別為97.66±4.92%與96.76±1.7%，而平均分類準確率分別為92.22±8.98%與83.65±4.47%。透過MIT-BIH資料庫測試結果，證明本研究提出的演算法確實可成功偵測這兩種心臟疾病。本研究亦使用科技接受模式問卷調查使用者對於本研究裝置使用情況，結果顯示整體平均分數為3.93±0.77。其中在心電訊號實際量測上對於受測者的使用情況與演算法性能影響最大因素為智慧衣的設計。當織布電極移動或電極接觸不良時，心電訊號品質會下降許多。最後，科技接受模式結果顯示使用者對於智慧衣的外觀設計接受度得分最低。這表示智慧衣的外觀與量測穩定性仍需加強，但整體而言，使用者對於本研究裝置接受度高。

關鍵字

心電圖；低功耗藍芽；織物電極；類比前端；小波轉換

並列摘要

In recent years, most of the wearable devices have the capability of physiological signal measurement function. However ,these wearable devices do not include real time heart diseases analysis capability. This study used fabric electrodes and ADS1298 analog front end to capture 12-lead electrocardiogram (ECG) signals. The TMS320c5515 digital signal processor (DSP) was used to realize wavelet transform and to analyze ECG signals. Finally, the analyzed result was transmitted through cc2541 bluetooth low power energy (BLE) to the mobile device for display and storage. In order to lower the hardware cost, a fixed-point processor was used as the computational core, and a interactive method was introduced that allow the user to target a particular cardiac disease of interest. The system then choose specific signal configuration from the 12 lead ECG for this selected cardiac disease to start analysis. By doing this, it can not only reduce the burden on the processor but also increase the accuracy and specificity of the disease analysis. In this, the analysis of the two most common heart diseases, atrial fibrillation (AF) and premature ventricular contractions (PVC) were realized. QRS complex detection and disease classification were evaluated using MIT-BIH arrhythmia database and MIT-BIH atrial fibrillation database, in this study. The averaged QRS complex detection accuracies are 97.66±4.92% and 96.76±1.7% for MIT-BIH arrhythmia database and MIT-BIH atrial fibrillation database, respectively. On the other hand, the averaged classification accuracies are 92.22±8.98% and 83.65±4.47% for MIT-BIH arrhythmia database and MIT-BIH atrial fibrillation database, respectively. From the above demonstrate that the proposed algorithm can successfully identify these two cardiac diseases. The technology acceptance model questionnaire also used to investigate the usage of this device. The result shows that the averaged overall score are 3.93±0.77. In addition, the design of smart clothes deeply affects the algorithm performance and subject's using situation. When fabric electrodes are moving or have poor contact, the quality of ECG signals are lower. In conclusion, result of technology acceptance questionnaire shows that the smart clothes design score the lowest in all the questions. This means that the smart clothes design and measurement stability should be enhanced in the future. However, the overall acceptability of the proposed system was high.

並列關鍵字

Electrocardiogram ； Bluetooth Low Power ； Fabric electrodes ； Analog front end ； Wavelet transform

參考文獻

[1] 行政院衛生署.102年國人主要死因分析.

[26] 張庭瑄, "以行動裝置實現智慧衣心電圖分析系統,"生物醫學工程系, 2014.

[2] Cuiwei Li, Chongxun Zheng, and Changfeng Tai, "Detection of ECG Characteristic Points Using Wavelet Transforms," IEEE Transaction on Biomedical Engineering, vol. 42, no. 1, pp. 21-28, January 1995.

[3] Mohammed Bahoura , "DSP implementation of wavelet transform for real time ECG wave forms detection and heart rate analysis," Computer Methods and Programs in Biomedicine 52, 1997, pp. 35-44.

[7] 鄭郁儐, "建構Android平台之DICOM ECG 擷取及管理系統," 生物醫學工程系, 2011.

國際替代計量

互動可程式之智慧衣心電圖監測系統

全文下載

主題瀏覽