心電圖(Electrocardiography)是對於心臟電生理活動的紀錄，也是現今用來診斷心臟疾病的主要方法。隨著長時連續測量多導程心電訊號，伴隨而來的龐大資料將造成傳輸與儲存空間的負擔。有著能夠即時且高壓縮率的多導程心電訊號演算法能夠有效的減少龐大資料所造成的負擔。然而壓縮可分為有損壓縮與無損壓縮。即使現今有損壓縮技術能將失真率控制在可接受的範圍，在醫療診斷上仍可能因失真而造成誤診。對於醫療人員若能提供有效的無損心電訊號壓縮，不但能提升在心臟疾病診斷上的準確度，也便於接下來的應用。 本論文利用多導程線性預測模組來降低導程之間的相關性、變動式線性預測模組來降低導程內的相關性。最後送入自調整的哥倫布編碼來提升整體系統的壓縮率。我們利用MIT-BIH Arrhythmia(MIT-BIH db)[1]、Physikalisch-Technische Bundesanstalt(PTBdb)[2]資料庫來做壓縮效率的評估。MIT-BIH db資料庫包含48組兩導程的心電訊號， PTB db資料庫包含來自290位測試病人的549組12導程心電訊號。結果顯示本論文所提供之壓縮演算法應用在MIT-BIH db資料庫平均壓縮率能達2.809x，應用在 PTB db資料庫平均壓縮率能達到4.073x。到最後我們將此演算法實現在嵌入式系統中。評估壓縮演算法展示遠距醫療的效果。 ;Electrocardiography is the recording of the heart electrical activity and used to diagnose heart disease nowadays. The diagnosis requires a large amount of time for acquiring enough multi-channel data normally. The storage and transmission of 12 lead ECG data will result in massive cost. With Multi-channel ECG lossless compression which has high compression ratio and capability of real time processing, we can effectively reduce the loading of huge data. However, compression algorithm can divide into lossless an lossy compression. Although data distortion has been controlled in acceptable range, it may cause some misdiagnosis while the diagnosis may occur. In this thesis, we propose a multi-channel ECG compression technique which has multi-channel linear prediction and adaptive linear prediction technique for removing redundancy in intra-channel and inter channel correlation respectively. Finally, entropy will feed into self-adjust Golomb-Rice codec for increase compression ratio. We evaluate the performance by calculating the compression ratio (CR) with MIT-BIH Arrhythmia(MIT-BIHdb) database[1] and Physikalisch-Technische Bundesanstalt database (PTBdb)[2]. The result shows the average compression ratio of proposed method is 2.809x in MIT-BIH database and 4.073x in PTB database. Finally, we implement proposed algorithm on the embedded development board for demonstrate the result of telemedicine application.
