考慮PRD與最大誤差之心電圖壓縮演算法設計

ECG Compression Algorithm Design with PRD and Maximal Error Considerations

張詒茹(Yi-Ju Chang)

指導教授 ： 曾傳蘆

國立臺北科技大學/機電學院/電機工程系所/碩士(2006年)

摘要

在醫療應用上，全天候的追蹤與記錄心電圖訊號使得資料量非常龐大，而且在傳輸心電圖資料時，常受到有限頻寬的限制，因此必須在考慮重建品質的前提下對心電圖進行壓縮，以免造成醫師誤判。本論文的研究目的為發展出一個可同時考慮百分比均方根誤差(Percent Root-Mean-Square Difference, PRD)與最大時域誤差等壓縮品質的小波轉換心電圖壓縮演算法，並利用向量量化與長度變動編碼法(Run Length Encoding, RLE)等壓縮編碼技術，來產生良好的壓縮比與品質。
在本論文中，利用小波轉換前、後之能量關係式，並配合臨界值取捨方式在所規定之PRD範圍中搜尋最佳壓縮比，而不須反小波轉換即可預估壓縮後之PRD。除此之外，就醫生的觀點，誤差太大亦會造成病症上之誤判，因此本論文考慮心電圖壓縮後的最大誤差限制。結合PRD與最大誤差之品質控制機制將使得生醫訊號在有失真壓縮下仍能維持一定醫學診斷的品質要求。進行前述壓縮後，本論文利用向量量化與RLE以達到較佳的編碼壓縮，以期在品質要求下，有良好的壓縮比。在向量量化中，接受端接收到資料串後，僅以簡單的查表動作即可重建原始生醫訊號，大大減少傳輸的位元數來達到最有效的壓縮。且因為在心電圖資料中，常有大量的0及1連續出現，針對此現象使用RLE進行編碼，以提高壓縮效率。
最後，本論文利用MATLAB軟體來撰寫小波轉換及相關編碼演算法，並使用MIT-BIH資料庫中的多筆心電圖紀錄來進行模擬。模擬結果顯示本論文所提出理論的確具有可行性與有效性。

關鍵字

心電圖(ECG) ； 小波轉換 ； 向量量化 ； PRD ； 最大誤差 ； 資料壓縮

並列摘要

In medical applications, the amount of electrocardiogram (ECG) data increases dramatically for long-term recording time. When sending the recorded ECG data, transmission rate is limited by the bandwidth of continuous channel and hence compression of ECG signals is often used. Also, in order to avoid miss diagnosis of symptoms, the reconstruction quality of ECG compression algorithm is critical. To solve the mentioned difficulties, this thesis is devoted to developing a quality-guaranteed wavelet-based algorithm that has PRD and maximal error control. Moreover, the algorithm produces good compression ratio provided that this thesis utilizes vector quantization and the Run Length Encoding (RLE) method.
By means of the energy relationship between time domain and wavelet domain representations, for those compression methods based on wavelet transform and thresholding, one can easily search an optimal threshold that gains best compression ratio and meets the specified PRD without taking inverse wavelet transform. Furthermore, since excessively error magnitude will lead the doctor to make a wrong diagnosis, this thesis investigates the maximal error of reconstructed signals. Combining with PRD control and the maximal error control, the proposed algorithm provides a better quality of the reconstructed medical signals. After compressing the raw ECG signals, this thesis utilizes the vector quantization technique and RLE scheme to achieve higher compression ratio by encoding. When the receiver receives the encoded data, the decoder only uses a simple table-look-up to reconstruct the medical signals. In addition, because there is a large number of 0’s and 1’s in the transformed ECG signals, this thesis can use the RLE to increase the compression ratio by reducing the redundant bits.
Finally, the MATLAB software is used to implement the compression algorithm. The ECG signals in the MIT-BIH database are adopted to test the program. The results of Simulation show that the feasibility and effectiveness of the algorithm proposed in the thesis.

並列關鍵字

ECG ； wavelet transform ； vector quantization ； PRD ； maximal error ； data compression

參考文獻

[17] 趙樹年，以小波轉換以及統一的向量量化架構進行失真到無失真的心電圖壓縮，碩士論文，中原大學電子工程學系碩士班，桃園，2004。

[3] J. L. Cardenas-Barrera and J. V. Lorenzo-Ginori, "Mean-shape vector quantizer for ECG signal compression," IEEE Transactions on Biomedical Engineering, vol. 46, no. 1, 1999, pp. 62-70.

[4] J. Chen and S. Itoh, "A wavelet transform-based ECG compression method guaranteeing desired signal quality," IEEE Transactions on Biomedical Engineering, vol. 45, no. 12, 1998, pp. 1414-1419.

[5] H. Lee and K. M. Buckley, "ECG data compression using cut and align beats approach and 2-D transforms," IEEE Transactions on Biomedical Engineering, vol. 46, no. 5, 1999, pp. 556-564.

[6] S. G. Miaou and H. L. Yen, "Quality driven gold washing adaptive vector quantization and its application to ECG data compression," IEEE Transactions on Biomedical Engineering, vol. 47, no. 2, 2000, pp. 209-218.