應用混合式編碼於動態醫學影像之壓縮

Using Hybrid Coders for Dynamic Medical Image Compression

葉進儀(Jinn-Yi Yeh) ； 蕭嘉贊(Jia-Zan Shiao) ； 張浚銘(Jun-Ming Chang) ； 傅家啟(Ja-Chih Fu)

《大葉學報》 13卷1期 (2004/06) Pp. 77-87

https://doi.org/10.7119/JDYU.200406.0077

摘要

隨著科技的進步及發展，醫療影像數位化已經受到許多專家學者的重視，藉由資訊的量化、無失真影像壓縮與傳輸的技術、數位化資料儲存、再配合電腦輔助診斷系統，如此能提供專業醫師在診斷上能有一快速及正確的診斷參考，藉以避免醫療延遲及資源的浪費，提高遠距醫療的診斷價值。本研究結合離散小波轉換、三角形區塊比對於動態影像壓縮技術、及算術編碼來消除或縮減在磁振造影影像中，任一或多種的重複性，以得到資料壓縮的效果，而達到符合網路化的需求，以期對影像的傳輸與儲存有更大助益。在研究中以數位化的左心室磁振動態影像資訊和腦部功能性磁振動態影像資訊作為壓縮的實例，並以高峰訊號雜訊比（peak signal-to-noise ratio，PNSR）值和壓縮比率（compression ratio，CR）來作績效的評估，實驗結果發現透過此研究架構來對醫學動態影像壓縮，可得到一極佳之PSNR值和CR值。

關鍵字

動態影像壓縮 ； 磁振造影影像 ； 高峰訊號雜訊比 ； 壓縮比率

並列摘要

During the current decade, many researchers have focused on the application and development of digital medical images. Because of the deployment of data quantification, lossless image compression, digital data retrieval, and computer-aided diagnosis systems, doctors can diagnose a patient quickly, easily, and correctly, even from a long-distance clinic. Therefore, image-compression techniques become increasingly important in digital image processing. In this study, a novel hybrid coder is developed for dynamic medical image compression. The kernel techniques include discrete wavelet transformation (DWT), triangle-block matching algorithms, and arithmetical coding to reduce temporal redundancy and to achieve a favorable lossless compression rate. The experimental design has two data sets, one of which is dynamic magnetic resonance images (MRI) for the human left ventricle. The other set is functional magnetic resonance images (fMRI) for the human brain. The peak signal-to-noise ratio (PNSR) and the compression ratio (CR) are used to evaluate the performance of this approach. The experimental results show that the PNSRs and CR for both cases are primed by applying the proposed method.

並列關鍵字

dynamic medical image compression ； magnetic resonance image ； peak signal-to-noise ratio ； compression ratio