無線多導極心電圖系統動作假影雜訊抑制 之使用混合式適應性濾波器

Abstract

本論文研究重點是針對在運動情況下量測心電圖(Electrocardiography, ECG)所產生的動作假影(Motion Artifacts)雜訊，分析在不同運動情況下，動作假影雜訊對不同導極的影響，觀察加速度計在不同導極之間的敏感度，參考加速度計的資訊來使用混合式適應性濾波器(Hybrid Adaptive filter)架構做動作假影雜訊的抑制，並且實作無線多導極心電圖系統(Wireless Multi-Lead ECG System)。 一般標準十二導極心電圖儀在實際使用上有諸多的限制，且因為多導極且有線帶來許多不便，另外，心電圖資料無法提供長時間的監控，也沒有有效的抑制機制來處理運動產生的雜訊。我們團隊實作的無線多導極心電圖系統，透過演算法將三導極心電圖資訊轉換出標準十二導極的心電圖資訊，在使用上大大增加了便利性，無線多導極心電圖系統搭載加速度計能應用在動作雜訊抑制、跌倒偵測警示、偵測導極貼位，此系統還能夠長時間監控心電圖並在雲端提供即時的病症預警。 傳統上使用加速度計適應性濾波器(G-sensor Adaptive Filter)抑制動作假影雜訊會因為加速度值和雜訊的關聯性很低造成訊號的失真，而使用相關平均適應性濾波器(Adaptive Coherent Averaging)抑制有時會藉由平均效應將原本沒有雜訊的R-R區間(R-R interval)引入雜訊。混合式適應性濾波器結合了兩者的優點，利用加速度的變化來模擬動作情況，消除相關平均適應性濾波器帶來的平均效應。我們在實作混合式適應性濾波器時，藉由分析不同的運動在不同導極上加速度值與雜訊的關聯性，針對不同運動來選擇是否使用加速度計適應性濾波器，透過這種方式，我們可以在不同運動情況下確保ECG的訊號品質

This thesis studies on the motion artifacts generated by measuring Electrocardiography (ECG) in exercise, and analyze the effects of motion artifacts on different lead under different motion conditions. Observe the accelerometer sensitivity between the different ECG lead, refer to the accelerometer information to use Hybrid Adaptive Filter architecture to do motion artifacts noise suppression, and implement wireless multi-lead ECG system. The general standard 12-lead ECG have a lot of restrictions in practical use. Because multi-lead and wired ECG brings a lot of inconvenience to user. Also, 12-lead ECG can’t provide long-term ECG monitoring, and there is no effective inhibition mechanism to deal with the noise generated by the motion. Our team implement the wireless multi-lead ECG system, through the algorithm transform 3-lead ECG into standard 12-lead ECG, greatly increased the convenience in use. Wireless multi-lead ECG system equipped with accelerometer can be applied in the motion noise suppression, fall detection alerts and detection of the placement of the electrode. This system can monitor ECG for a long time and provide immediate symptom alerts in the cloud. Traditionally, G-sensor Adaptive Filter used to suppress motion artifacts. Distortion caused by the low correlation between acceleration and noise. Adaptive Coherent Averaging sometimes will introduces noise into noiseless R-R interval by average effect. Hybrid Adaptive Filter combines the advantages of both. Use changes in acceleration to simulate the motion and eliminate the average effect caused by Adaptive Coherent Averaging. In the implementation of Hybrid Adaptive Filter, we analyze the correlation between acceleration and noise in different exercise and different lead and choose whether to use g-sensor adaptive filter. In this way, we can ensure ECG signal quality under different exercise conditions.