心律變異分析演算法之研究

A Study of Heart Rate Variability Analysis Algorithm

謝易珉(Yi-Ming Hsieh)

指導教授 ： 張朝陽

國立虎尾科技大學/電機資訊學院/資訊工程系碩士班/碩士(2016年)

摘要

近年來隨著醫療科技逐漸的進步，使得人類壽命的延長與老年人口的增加，然而，由於出生率的下降使得整體人口出現高齡化的現象，因為環境、飲食、壓力等因素的變化，疾病隨著年齡的增長而增加，因此長期的看護治療逐漸的被重視，在這些疾病中心血管疾病占了一大部分，為了降低因病所發生的不幸，數種感測儀器被普遍的使用來量測患者的心電圖訊號 (Electrocardiogram, ECG)，經由分析心電圖訊號我們可以得到許多疾病的徵兆，透過檢測結果我們可以將資訊通知患者，或上傳資料至後端伺服器提供醫護人員與第三者資訊，以達到醫護人員及早診斷症狀與第三者的即時呼救，為了分析疾病症狀我們將提出一個簡單而即時性的檢測方法，此方法主要包含了三個部分，第一部分為訊號的前置處理，我們使用過濾器去移除雜訊干擾並透過波形轉換獲得特徵波形，此操作可增強QRS複合波的振幅並將其餘波形振幅給壓制下來，第二部分為R波峰的檢測，R波峰檢測是分析心電圖最為關鍵的部分，而在檢測時主要受到不規則的型態變化與雜訊所影響，導致可能的波峰檢測錯誤，因此將對輸入訊號作波處理，並且我們在波峰檢測階段加入雜訊判斷的步驟，以降低因雜訊而造成的檢測錯誤，第三部分為心律變異分析，此部分將以R波峰為基準對波形做分析，其中包含了波峰間隔與計算分析波形型態來取得特徵參數，透過波峰間隔與參數來做心律異常判斷，並針對資料庫中的幾種症狀做檢測，檢測訊號來源為MIT-BIH心律異常資料庫，資料庫包含了48筆半小時的雙通道訊號，本文所提出的R波檢測方法達到99.69百分點的平均準確度、99.78百分點的靈敏度以及99.91百分點的正確波峰預測度，而本文所提出的症狀分析達到64.89百分點的整體準確度，實驗結果顯示波形型態分析容易受到雜訊以及突然的變化所影響。

關鍵字

心電圖 ； 波形轉換 ； R波檢測 ； 心律不整分析 ； 雜訊濾除

並列摘要

Along with the advancement of medical technology in recent years, it prolongs the human life and increase the elderly population. However, a lower birth rate causes the aging of the population. Because of the changes in human diet and the pressures of life. Some diseases when the human age increases. Therefore, the long-term treatment becomes an important issue. Among the diseases, the cardiovascular disease is a major of them. To reduce the disease onset, several instruments are commonly used to obtain the electrocardiogram (ECG) signal of the patient. Additionally, it can distinguish many indications of the illness by analysis the electrocardiogram signal. The analyzed information can be uploaded to the cloud systems to provide the inquiry for the users. Furthermore, medical professionals can employ the information to diagnose the symptoms or to call for help immediately. In order to analyze the symptom of disease, we propose a simple and instantaneous detection algorithm. The proposed method includes three stages. The first stage is the signal preprocessing. We use the filter to remove the noise interference and obtain the feature waveform by waveform transformation. It can enhance the QRS complexes and depress the unnecessary waveform. The second stage is to detect the R-peak location. It is a crucial stage in the proposed method. The detection method is influenced by irregular morphology and noise. It may cause the false detection. Therefore, we add the noise judgment into the peak detection stage to reduce the false detection. The third stage is the arrhythmia analysis. This stage operation is based on the R-peak. It contains the interval of each R-peak and the values of the waveform feature. We judge the arrhythmia by using the interval and eigenvalues and detect some symptoms in the database. The detection electrocardiogram data are from MIT-BIH arrhythmia database. It includes the 48 half-hour of two channel records. The proposed method of R-peak detection achieves average detection accuracy of 99.69 percent, sensitivity of 99.78 percent, and positive predictivity of 99.91 percent. The proposed method of symptom detection achieves overall detection accuracy of 64.89 percent. Experimental result show that the waveform morphological analysis is easily affected by noise and sudden change.

並列關鍵字

electrocardiogram (ECG) ； waveform transformation ； R-peak detection ； arrhythmia analysis ； noise filtering

參考文獻

[8] M. S. Manikandan and K. P. Soman, “A novel method for detecting R-peak in electrocardiogram (ECG) signal,” Biomedical Signal Processing and Control, vol. 7, pp. 118-128, 2012.

[9] P. Phukpattaranont, “QRS detection algorithm based on the quadratic filter,” Expert Systems with Applications, vol. 42, pp. 4867-4877, 2015.

[10] A. G. Ramakrishnan, Senior Menber, IEEE, A. P. Prathosh, and T. V. Ananthapadmanabha, “Threshold-independent QRS detection using the dynamic plosion index,” IEEE Siganl Processing Letters, vol. 21, pp. 554-558, 2014.

[11] Yun-Chi Yen and Wen-June Wang, “QRS complexes detection for ECG signal: the difference operation method,” Computer Methods and Programs in Biomedicine, vol. 91, pp. 245-254, 2008.

[12] L. Sathyapriya, L. Murali, and T. Manigandan, “Analysis and detection R-peak detection using modified pan-tompkins algorithm,” IEEE International Conference on Advanced Communication Control and Computing Technologies, pp. 483-487, 2014.