透過您的圖書館登入
IP:34.227.191.136
  • 學位論文

適用於長期光體積描述訊號監控之訊號處理演算法選擇機制

Signal Processing Selection Mechanism for Long-Term PPG Monitoring

指導教授 : 吳安宇
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


穿戴式裝置,顧名思義就是容易攜帶,並利用科技技術開發出可以隨時隨地追蹤和分析穿戴者資訊的裝置。近年來穿戴式裝置的市場價值逐漸提升,也在新冠肺炎疫情的催化下,穿戴式裝置得以在醫療領域嶄露頭角。由於穿戴式裝置的可攜帶性和長期量測生理訊號的特性,使它可以廣泛使用在生理監測的應用上,以光體積描述訊號為例,能用於心律監測,血壓預測,心臟疾病的偵測等等相關應用。即便穿戴式裝置有以上好處,然而,從消費者的觀點出發,穿戴式裝置會有兩點和工程問題相關的挑戰,1)不準確資料的量測,由於光體積描述訊號的量測方式使訊號容易受到運動偽影影響,進而產生錯誤的量測資訊,2)電池續航力不足的問題,穿戴式裝置本身電池容量有限,難以負荷長期監測下的訊號處理運算。因此,本文針對以上提到的兩項挑戰,基於訊號處理演算法上提出相對應的解決技術。 針對光體積描述訊號的不穩定性問題,本文基於心律監測的應用,提出心電訊號輔助下的光體積描述訊號品質評估系統(ECG-aided PPG Signal Quality Assessment System),透過評估訊號品質來篩選訊號,並利用心電訊號和光體積描述訊號之間的生理連結,來提供更為客觀的訊號品質定義,進而達到準確除去高預測誤差訊號的效果。而針對電池能耗的問題,我們從訊號處理演算法來著手,改善傳統訊號處理作法中單一前處理下運算資源的浪費,本文首先提出條件性前處理系統(Conditional Preprocessing System),引出並非所有訊號都需要做前處理的概念,原本就具有資訊量的訊號不需要額外花費前處理的資源去計算,以減少前處理的資源浪費,接著,更進一步提出輕量訊號前處理系統(Lightweight DSP Preprocessing System),利用多個輕量前處理演算法的組合來取代一個高複雜度的前處理演算法,並藉由評估訊號特性來選擇訊號適合做哪種的前處理方式,達到維持高準確度,但能大幅下降訊號處理時間的好處。

並列摘要


Wearables are portable devices that can track and analyze wearer data anytime and anywhere. In recent years, the market value and the popularity of wearable devices have gradually increased. Furthermore, under the impact of the epidemic outbreak, the role of wearable devices has been expanded in the healthcare sector. Due to the portability of wearable devices and the ability of long-term monitoring, they can be widely used in different physiological applications. For example, photoplethysmography (PPG signals) can be used for heart rate monitoring, blood pressure prediction, cardiovascular disease detection, and so on. However, from the consumer's point of view, there are two challenges related to engineering issues with wearable devices: 1) inaccurate data measurement, the optical measurement of the PPG signal makes it sensitive to motion artifacts, which will lead to the incorrect measurement result. And 2) the problem of battery draining fast, the limited battery of the wearable device cannot support long-term computation of signal processing. Based on the two challenges mentioned above, this thesis aims to propose corresponding solutions from the signal processing perspective of the wearable. To deal with the stability issues of PPG signals, we propose an ECG-aided PPG signal quality assessment system for the application of heart rate estimation. We utilize the physiological connection between the ECG signal and the PPG signal to provide a more objective definition of signal quality. Hence, it can achieve a higher rejection rate of low-quality signals. As for the battery issue of the wearable, we propose a conditional preprocessing system and introduce the concept that not all signals require to be preprocessed. Signals informative enough do not require additional preprocessing effort to enhance signal quality. Hence, this framework can reduce the computation waste of preprocessing. Next, a lightweight DSP preprocessing system is proposed to replace a power-hungry preprocessing algorithm with multiple lightweight preprocessing algorithms. By evaluating the signal characteristics and selecting which preprocessing method is suitable for each signal, this framework can maintain high accuracy and significantly reduce processing time for signal processing.

參考文獻


[1] K. Guk, G. Han, J. Lim, K. Jeong, T. Kang, E.­K. Lim, and J. Jung, “Evolution of wearable devices with real­time disease monitoring for personalized healthcare,” Nanomaterials, vol. 9, no. 6, p. 813, 2019.
[2] Grand View Research, “Wearable Technology Market Size: Industry Report, 2020­2027.” https://www.grandviewresearch.com/industry-analysis/ wearable-technology-market, June 2020. Accessed: 2021­04­01.
[3] Ericsson ConsumerLab, “Wearable technology and the IoT: Consumer views on wearables beyond health and wellness.”https:// www.ericsson.com/en/reports-and-papers/consumerlab/reports/ wearable-technology-and-the-internet-of-things?fbclid= IwAR0IHFxiJnr0K0wiCF7EQLEaT-c22wJe3qUyBcVTa_P4BJEMNh0d1FDHpGk, Mar. 2016. Accessed: 2021­04­01.
[4] C. Orphanidou, T. Bonnici, P. Charlton, D. Clifton, D. Vallance, and L. Tarassenko, “Signal­quality indices for the electrocardiogram and photoplethysmogram: Derivation and applications to wireless monitoring,” IEEE Journal of Biomedical and Health Informatics, vol. 19, no. 3, pp. 832–838, 2015.
[5] Builtin, “What is Wearable Technology? Examples of Wearables.” https:// builtin.com/wearables. Accessed: 2021­04­01.

延伸閱讀