以個人電腦叢集技術支援高效能與安全之醫學影像傳送系統之建置與研究

The Establishment and Research on a System for Efficient and Secure Distribution of Medical Images by Using Personal Computer Clustering Technique

楊仁華(Yang Ren Hua)

指導教授 ： 陳士農

亞洲大學/資訊學院/資訊工程學系碩士班/碩士(2003年)

摘要

隨著醫學影像逐漸利用數位化的方式來處理，以及各式各樣有關醫學影像應用的發展，使得數位醫學影像的研究越來越顯得重要。不管是在遠距醫療或是遠距醫學教學中的醫學影像都會需要運用到大量的運算，其中包含了影像壓縮與影像傳送，影像壓縮的目的在於希望能夠在降低位元率的同時又可以盡可能的保持影像品質。除此之外還希望能夠藉著使用個人電腦叢集來達到以往利用傳統平行電腦處理大量資料時所擁有的效能。
本論文主要的貢獻是建置一個以個人電腦叢集為基礎的醫學影像傳送系統，首先利用小波轉換對影像進行頻帶的轉換與切割，接著利用個人電腦叢集對醫學影像進行壓縮與加密，最後再將這些檔案儲存在伺服器端的資料庫中。一旦位於客戶端的使用者對影像提出需求，系統會利用漸進式傳輸的方式傳送影像，也就是使用者首先會看到的是影像的低頻部份，如此可以讓使用者對此張醫學影像進行初步的瀏覽，如果使用者需要更進一步瀏覽此張影像，系統會將資料庫中其他用來加強影像的區塊逐步送至客戶端，將影像還原成原始的醫學影像。我們希望這個系統能達到高效能與安全這兩個特點，同時此系統還可以對影像進行選
擇性區塊選取的動作，也就是在進行小波轉換前系統會將影像中重要的部份選取出來，在之後的影像壓縮步驟進行不失真壓縮，使影像中重要的資訊不會遺失，進而達到完整保存影像品質的目的。

關鍵字

醫學影像(medical image) ； 小波轉換(wavelet transform) ； 影像壓縮(image compression) ； 個人電腦叢集(PC cluster)、 ； PVM(Parallel Virtual Machine)

並列摘要

Accompanied with processing medical image by digitalization procedures and various medical image applications, research in digital medical image is getting more and more attention nowadays. Both distant medical treatment and distant medical education require massive computerization, which includes technologies of image compression and image transition. The purpose of image compression is to balance image size and image quality. Additionally, image compression makes it possible to achieve a great performance while processing considerable data through only PC cluster rather than traditionally parallel computer techniques.
This dissertation is to try to contribute a medical image transition system which is established based on PC cluster. First, in terms of image itself, wavelet transform is used to transform and divide subband. After that, we digitalize and encrypt medical images via PC cluster. At last, these digitalized and encrypted files are
stored in the database server. Once end users request for certain images, system will progressively transit images to the other end, and this means that the low band part of image will show up in the client side. This technique enables users to browse this medical image in the first place. If detailed image is further requested, system will restore original medical image by the means of transiting missed image data from the database. This system is expected to achieve high efficiency and high security. This is due to the medical image transition system can select selective blocks within the image; system itself would select the most significant parts of the image before processing wavelet transform. In the other words, in order to keep completed image information, the most crucial information will not be lost after lossless compression technology.

參考文獻

[陳衍良 2000]

[Averbuch 1996]

A. Averbubh, D. Lazar and M. Israeli, “Image compression using wavelet transform and multiresolution decomposition,” IEEE Trans. On Image Process, 5(1)：4 -15, 1996

[Bruckmann 1998]

[Clunie 2000]