應用於生醫訊號特徵萃取及偵測之數位訊號處理系統

Design and Implementation of Digital Signal Processing Systems for Biomedical Features Extraction and Detection

謝其軒(Hsieh, Chi Hsuan)

指導教授 ： 黃元豪

國立清華大學/電機資訊學院/通訊工程研究所/博士(2016年)

摘要

近來，由於穿戴式裝置的崛起以及老年人口比例的增加，整合式的生醫信號處理系統變得愈趨重要。在這篇論文當中，我們提出並實做了兩套先進的生醫數位信號處理系統。
生醫的信號處理基本上都可被歸類為特徵萃取以及偵測的問題。藉由合適的特徵萃取，我們在這篇論文當中發展了兩套高效率而低複雜度的生醫數位信號處理系統。第一套系統是一個利用眼動產生的腦波信號來控制電腦的腦機系統。我們利用了通訊系統常被使用的脈衝編碼解調演算法，實現了一個不需乘法器的眼動偵測系統。除此之外，我們還提出了針對這套演算法的訓練模型來調整演算法中會用到的參數。我們用FPGA實做了這套系統並且達到了百分之89.7的偵測率。
藉由這套系統，使用者可戴著腦波偵測儀器對下六種眼動的指令來控制電腦。
第二套系統則是一個基於超寬頻雷達的呼吸信號特徵萃取系統。在這套系統當中，我們提出了新的呼吸模型以及對應的特徵萃取演算法來偵測出除了呼吸頻率以外其他更多的呼吸特徵，例如吸氣速度，呼氣速度，呼吸深度，以及呼吸停等的時間比例等。這些特徵不只可用做醫學上評估診斷的潛在指標，它們同時也可以被視為壓縮後的呼吸模型信號，當我們只在意這些特徵時，每一次呼吸的記錄都可以只留下這些特徵，而省下長時間呼吸信號監測原本需要的大量儲存容量或是遠端照護中原本需要的傳輸頻寬。我們實做了一套實時的雷達呼吸信號特徵萃取平台，雷達呼吸信號由超寬頻雷達晶片送入我們的FPGA信號處理平台，信號處理平台再針對每一次的呼吸週期做信號處理並送出對應的特徵資料送至電腦上顯示。

關鍵字

超寬頻雷達 ； 呼吸信號 ； 腦電波 ； 人機介面

並列摘要

Nowadays, the implementation of biomedical integrated systems are attracting more attention than before due to the emerging industry of wearable devices and the rapid growth
of elderly population. In particular, efficient biomedical signal processing systems are in demand for various applications. This dissertation aims to develop advanced digital signal processing (DSP) systems for wireless biomedical applications.
Many problems in the field of biomedical signal processing can be reduced to a task of feature extraction and event detection. This kind of problem generally treats a set of
measurements and asks for the recognition of some patterns. Through appropriate feature extraction from the targeted signal, we can develop efficient signal processing algorithms to perform different tasks. This dissertation proposes two digital signal processing biomedical systems. The first one is an electroencephalography (EEG) based brain-computer interface (BCI) utilizing eye commands. This system first uses a low-complexity edge detector to extract the sharp edges of the eye movement events. Then, we use pulse width demodulation (PWDM) to further classify the eye commands with only addition operations. Also, a training mechanism is proposed to facilitate the detection of eye commands. Users wearing an EEG headset can give six eye commands including glancing toward four directions and winking of the left or right eye. This proposed system is implemented with FPGA and
achieves a detection rate of 89.7% in the experiments.
The second proposed digital signal processing biomedical system is an ultra-wideband(UWB) radar signal processing platform for human respiratory feature extraction. In this
system, we propose a new respiration model and an iterative correlation search algorithm with early termination to acquire additional respiratory features such as the inspiration and expiration speeds, respiration intensity, and respiration holding ratio. These features, the parameterized and compressed respiratory signals, can provide physical information to facilitate clinical diagnosis and help to manage a more efficient database for the respiration monitoring system. The proposed respiratory feature extraction algorithm is designed and
implemented using the proposed UWB radar signal processing platform including a radar front-end chip and an FPGA chip. The proposed radar system can detect human respiration
rates at 0.1 to 1 Hz and facilitates the real-time analysis of the respiratory features of each respiration period. Moreover, the parameterized waveforms are used to construct a new diagnosis method for respiratory diseases in clinical trials.

並列關鍵字

UWB ； Respiration ； EEG ； BCI

參考文獻

[1] M. A. Hanson et al., “Body area sensor networks: challenges and opportunities,” Computer, vol. 42, no. 1, pp. 58–65, Jan. 2009.

[2] T. Calabria, “Software PACE detect with ADS1292,” in Texas Instruments, 2015.

[3] J. Malmivuo and R. Plonsey, Bioelectromagnetism: principles and applications of bioelectric and biomagnetic fields. Oxford University Press, 1995.

[5] A. Funase, T. Hashimoto, T. Yagi, A. K. Barros, A. Cichocki, and I. Takumi, “Research for estimating direction of saccadic eye movements by single trial processing,” in 29th Int. Conf. IEEE Eng. Med. Biol. Soc.,, 2007, pp. 4723–4726.

[7] S. S. Gupta, S. Soman, P. G. Raj, R. Prakash, S. Sailaja, and R. Borgohain, “Detecting eye movements in EEG for controlling devices,” in IEEE Conference on Computational Intelligence and Cybernetics, 2012, pp. 69–73.