臺灣博碩士論文加值系統

心電圖(Electrocardiogram；ECG)使用一種非侵入式的檢測方式，來記錄心臟收縮和擴張運動時，所產生的電位變化，醫生可依照波型的變化判斷出可能的病症。長期的心電圖記錄需要花費大量的儲存空間來記錄12導程心電圖資料。為滿足如此龐大的資料儲存與傳輸需求，具保持臨床診斷資訊能力之心電圖資料壓縮技術乃因應而生。
本論文是第一個針對基於1-D RRO-NRDPWT且具重建品質保證機制心電圖壓縮系統，設計其VLSI的架構，藉此驗證此心電圖壓縮系統設計的可行性，並修改部分設計以方便硬體實現。本篇論文在1-D RRO-NRDPWT實現中，使用了區段累加演算法(SAA)的觀念實作，並特別對於邊界資料的累加設計一控制單元，達到降低硬體成本的付出，需要7785900um2的Total area和17mW Dynamic Power，與未精簡的架構相比，只需要其2.54%的硬體面積和6.46%Dynamic Power。並對1-D RRO-NRDPWT和非線性量化演算法進行整合，若考慮量化每段1024筆心電圖訊號在QOD機制遞迴運算下，最多3次內必定達到使用者設定的心電圖重建品質。而心電圖單一導程處理時間每段最多約為52n秒。在考量壓縮12導程之需求下，其處理最多約0.624u秒。如果以心電圖之取樣頻率與霍特系統一樣為1KHz的話，系統可有效地達到心電圖的即時運算功能。

Electrocardiogram (ECG) is a non-invasive modality recording of the electrical activity of the heart, and is widely used for clinical diagnosis of heart disease. However, storing the 12-lead ECG signal needs huge capacity. In order to keep the ECG signal clinical features, a lot of ECG compression methods were proposed.
Besides, maintaining reconstructed signals at a desired level of quality is crucial for lossy ECG data compression. Based on the 1-D RRO-NRDPWT, this study proposes an ECG compression system which has the ability of ECG compression quality control. For design the VLSI architecture of the ECG compression system, this paper modifies the original non-linear quantization algorithm and uses the concepts of the segment accumulation algorithm (SAA). Besides, the architecture proposed in this improves the control unit of Boundary data processing, which can reduce the hardware cost, only needs 7785900um2 area and 17mW Dynamic Power, Comparing with the direct implementation of 1-D RRO-NRDPWT, the propose architecture only needs 2.54% area and 6.46% Dynamic Power. This paper further considers the integration of 1-D RRO-NRDPWT and nonlinear algorithms. To achieve the desired ECG compression quality, this architecture only needs 3 iterations under the novel QOD method for each segment ECG signals. The maximum execution time for compressing single channel ECG signal is 52 ns. For considering 12-lead ECG signals, the maximum execution time is about 0.624 μs, which can meet the requirement of real-time ECG recording system.

目錄
中文摘要-------------------------------------------------i
英文摘要------------------------------------------------ii
致謝----------------------------------------------------iii
目錄----------------------------------------------------iv
表目錄--------------------------------------------------vi
圖目錄--------------------------------------------------vii
第一章 緒論-------------------------------------------1
1-1 介紹背景---------------------------------------1
1-2 研究動機與目的---------------------------------------4
1-3 論文架構---------------------------------------------5
第二章 小波轉換理論--------------------------------------6
2-1遞迴式離散週期性小波轉換(DPWT) --------------6
2-2非遞迴週期性離散週期性小波轉換(NRDPWT)-------9
2-3可逆式四捨五入線性轉換(RROLT)------------------------12
第三章 心電圖壓縮系統-----------------------------------14
3-1基於RRO-NRDPWT的心電圖壓縮系統--------------14
3-2 非線性量化演算法之基本原理--------------------------16
3-3壓縮系統失真分析------------------------------------21
3-4壓縮品質控制機制(QOD)應用於ECG壓縮-------------------24
3-5 QOD機制之失真度預測---------------------------------26
第四章 1-D RRO-NRDPWT與量化演算法硬體設計---------------28
4-1 1-D RRO-NRDPWT硬體規劃 --------------------29
4-1-1 運算架構規劃--------------------31
4-1-2 小波轉換有限位元分析------------33
4-2 量化演算法硬體規劃-------------------------34
4-2-1 量化尺度使用方式----------------34
4-2-2量化尺度有限位元分析-------------------------------35
4-2-3 硬體架構及實現方式--------------------------------37
4-2-4 讀取小波分解係數----------------------------------37
4-2-5 量化尺度讀取--------------------------------------37
4-2-6 量化處理------------------------------------------38
4-2-7 失真度計算----------------------------------------39
4-2-8 Step預測計算--------------------------------------42
第五章 驗證結果與探討-----------------------------------43
5-1 1-D RRO-NRDPWT模擬結果---------------------43
5-2 量化演算法模擬結果-------------------------52
5-2-1 量化處理------------------------56
5-2-2 失真度計算----------------------------------------56
5-2-3 量化階級預測計算----------------------------------59
5-2-4 量化尺度獲得--------------------------------------60
5-3 模擬結果分析----------------------------------------64
第六章 結論---------------------------------------------65
參考文獻------------------------------------------------66

參考文獻
[1]L. U. Z., D. Y. Kim, and W. A. Pearlman, “Wavelet compression of ECG signals by the set partitioning in hierarchical trees algorithm,” IEEE Transaction on Biomedical Engineering, vol. 47, pp. 849-856 July 2000.
[2] C. T. Ku, H. S. Wang, K. C. Hung, “High efficient ECG compression base on reversible round-off non-recursive 1-D discrete periodized wavelet transform,” Medical Engineering and Physics, pp. 1-18, Dec. 2006.
[3] C. T. Ku, H. S. Wang, K. C. Hung, and Y. S. Hung, “A Novel ECG Data Compression Based On Reversible Round-Off 1-D NRDPWT.” IEEE International Symposium on Intelligent Signal Processing and Communication System, pp. 602-605, Dec. 2006.
[4] C. T. Ku, H. S. Wang, K. C. Hung, and Y. S. Hung, “A novel ECG data compression method based on non-recursive discrete periodized wavelet transform,” IEEE Transaction on Biomedical Engineering, vol. 53, pp.2577-2583, Dec. 2006.
[5] I. Daubechies, “Ten lectures on wavelet,” Ser. No.61 in CBMS-NSF series in Applied Mathematics. Philadelphia, PA: SIAM, 1992.
[6] 黃啟仁, “一維可逆式四捨五入非遞迴離散週期性小波轉換之FPGA硬體實現,” 國立高雄第一科技大學電腦與通訊工程系研究所碩士論文, 2009.

[7] 陳廷翰, “以FPGA實現一基於可逆四捨五入非遞迴離散週期小波轉換心電圖壓縮系統之量化演算法,” 國立高雄第一科技大學電腦與通訊工程系研究所碩士論文, 2009.

[8]MIT心律不整心電圖資料庫 http://www.physionet.org/physiobank/database/mitdb/