新型多導程胎兒心電監控系統

A NEW MULTI-LEAD ECG SYSTEM FOR FETUS MONITORING

尤仲瑋(Chung-Wei Yu)

指導教授 ： 徐良育

中原大學/工學院/生物醫學工程研究所/碩士(2014年)

https://doi.org/10.6840/cycu201400821

摘要

利用量測胎兒心電圖預測胎兒的健康狀況，在懷孕期間是非常重要的。而非侵入式的方法所的到的訊號相較於侵入式的方式所得到的訊號較小也包含較多的雜訊，在得到乾淨的胎兒心電訊號前必須先做去除雜訊。
本研究設計一新型多導程胎兒心電擷取系統，利用五個市售心率帶上的橡膠電極置於托腹帶上，作為前端的電極，並利用德州儀器出產的MSP430F149微控制器透過SPI(Serial Peripheral Interface)串列傳輸的方式控制ADS1298晶片取代以往多導程胎兒心電擷取系統的前端線路，將線路簡化，減少採集母親的腹部訊號時的雜訊以及空間。最後再將訊號擷取進入微控制器內進行演算並偵測胎兒心率。
在演算法的部分， 將本研究前端電極採集到的訊號進行分組，選出一個通道作為主要腹部訊號另外三個通道的訊號平均作為參考訊號。為了要進行自動分組，本研究計算每通道的訊號中胎兒的RR平均間期標準差與RR平均間期的比值，並發現當比值越小時，該組訊號的正確率會越高。因此可以依據此比值來進行自動分組。 另一方面，一開始先將主要腹部訊號上的母親QRS波進行刪除，再將其以及參考訊號輸入LMS自適應濾波內進行濾波，將除了胎兒心跳訊號以外的雜訊都濾除。此外將偵測到的前十跳的母親QRS波做平均得到一個母親QRS波樣板。若胎兒QRS與母親QRS波重疊，則利用母親QRS樣板與主要腹部訊號上的母親QRS波進行相減，將相減後的結果加回處理過後的訊號上。最後再進行胎兒的QRS偵測，即可得到胎的心率。
本研究也將系統實際應用在孕婦上，結果顯示，電極系統可以成功採集母親腹部訊號。而演算法則是利用PhysioNet Chllange 2013 Set A資料庫中的訊號進行驗證。結果證明本研究所提議的演算法可以成功提取出胎兒心電訊號。而胎兒QRS偵測的平均正確率達66%。但因為每組訊號品質不一，當訊號品質較差時就會影響演算法的正確率。
結論，本研究完成多導程胎兒心電監控系統，並利用演算法成功分離母親以及胎兒心電訊號，成功偵測胎兒的QRS，並計算胎兒心率。

關鍵字

胎兒心電訊號 ； 自適應濾波 ； 多導程心電系統

並列摘要

It is very important, during pregnancy, to measure fetal electrocardiogram(FECG) for predicting fetal health condition. However, when compared with invasive method, the signal obtained by non-invasive methods is small and is contaminated by different noises. We need to remove these noises before a clean fetal electrocardiogram can be obtained for processing.
This study designs a new multi-electrode FECG monitoring system by placing five rubber electrodes, similar to the rubber electrode on commercial heart rate belt, on a maternity belt to detect abdomen signals. The MSP430f149 micro-controller from the Texas Instrumentation was used to control the ADS1298, using serial peripheral interface (SPI), in order to simplify the circuit and to reduce noise. Finally, the signal is fetched into micro-controller to detect and compute the fetal heart rate.
For algorithm valuation, four signals from the electrodes have to be grouped, one of them was chose as main abdomen signals, and the other three signals were averaged as reference signals. For the purpose of automatic grouping, the ratios between average RR interval and standard division of RR interval were computed for each channel. It is found that when this ratio is smaller the accuracy of that channel is higher. Thus, this ratio can be used for automatic grouping of the input signal. On the other hand, we first remove the maternal QRS on the main abdomen signal and put it, along with the reference signal, into the LMS adaptive filter to remove the noises other than the fetal ECG. Additionally, the first ten maternal QRSs were averaged to generate a template. When the fetal QRS is overlapped with maternal QRS, this template was subtracted with the corresponding maternal QRS in the main abdomen signal. After subtraction, the residue was added back to the main abdomen signal. Finally, after the detection of fetal QRS, the fetal heart rate can be obtained.
This study also uses the proposed system on a pregnant woman. The results indicate that this system can collect abdomen signals successfully. By using the PhysioNet Challenge 2013 Set A database, the proposed signal processing algorithm was validated. The results demonstrate that the proposed method can extract fetal ECG and the averaged accuracy of fetal QRS detection is 66%. However, the quality of acquired signal will affect the averaged accuracy.
In conclusion, the study constructs a multi-lead ECG system for fetal monitoring. The proposed signal processing algorithm can successfully separates the maternal and fetal ECG, detects the fetal QRS, and computes the fetal heart rate.

並列關鍵字

Fetal Electrocardiography ； Adaptive Filter ； Multi-lead ECG System

參考文獻

[1] 陳麒合, "以適應性消噪濾波器由孕婦心電圖分離並分析胎兒心電圖," 亞洲大學生物資訊研究所碩論文 100.

[3] J. G. Webster, "Medical Instrumentation Application and Design," Elsevier B.V, 1992.

[4] T. Wheeler, A. Murrilis, and T. Shelley, "Measurement of The Fetal Heart Rate During Pregnacy by A New electrocardiographic Technique " An International Journal of Obstetrics & Gynaecology, vol. 85, pp. 12-17, 1978.

[5] S. Abboud and A. Beker, "An improved detection algorithm in fetal electrocardiography," Journal of Electrocardiology, vol. 22, pp. 238-242 1989.

[7] D. Graupe, "Time Series Analysis: Identification and Adaptive Filtering," R.E. Krieger Pub ,1984 .