為了加速PACS 傳輸影像的速度以及節省PACS的儲存空間,影像壓縮是必須的。影像壓縮可分為兩大類:無失真壓縮和失真壓縮。無失真壓縮其解壓縮影像可以完整的復原,所以常被應用於醫學影像上,可是其壓縮率有限,大概為1.5到3左右。另一方面失真壓縮雖然可以達到很高的壓縮率,但因為無法復原影像,所以壓縮完後可能會影響影像的品質。對失真壓縮而言,高壓縮比很容易達成但其代價就是影像變差。因此影像壓縮效果的探討在醫學影像上的應用變得相當的重要。目前JPEG2000是最受歡迎的影像壓縮方法而且它也是DICOM標準所支援的壓縮方法之一,本篇論文主要在評估醫學影像使用JPEG2000在不同壓縮比下的效果。 本研究主要以2部顯示器作為評估的工具,分別用以顯示原始影像及解壓縮影像。在評估影像品質之前,先根據DICOM PART 14和AAPM TG18的規範進行螢幕校正。使用的測試醫學影像其成像方式分別為CR、DR、CT、DF、DSA、MRI、DM、RF以及US,共計199張。而評估影像壓縮的效能,則依無失真與失真壓縮方式分開測試。無失真壓縮方面,利用差值影像來觀察其是否精確的復原。在失真壓縮方面評估方式,設定壓縮比的範圍由5到400。除了計算PSNR外,另外加入視覺測試,最後將結果用WINKS統計分析軟體進行實驗數據的分析。 實驗結果顯示本研究所使用的螢幕符合DICOM的規範,在無失真壓縮方面,影像皆可完整復原,但其壓縮比不高平均在3左右。在失真壓縮方面,各類不同的醫學影像中,可接受的壓縮率在10,而最高可達60。本結果將可提供醫療界在使用JPEG2000壓縮各類醫學影像的參考,此外也提供一個可行的螢幕校正機制。
In order to speed up the image transmission rate and to save disk space for PACS, applied image compression concept into system is needed. Image compression fall into two broad categories: lossless and lossy. Lossless compression method allows exact recovery of the original image, and is certainly the obvious choice for medical imaging applications. However, it achieves very modest compression ratios, typically from 1.5 to 3. On the other hand, the lossy compression method allows much higher compression ratios, but it does not allow exact recovery after compression which may affect the quality of images. Generally, higher compression ratios can be obtained at the expense of more image degradation. The chosen of compression method become an important factor for this application. Currently, JPEG2000 is one of most popular method for still image compression and one of suggested compression method in DICOM 3.0 standard. In this study, it evaluated the performance of JPEG2000 compression method for several different medical images which acquired from different modalities with different compressed ratio. In this study, two monitors are calibrated according to the standard of DICOM PART 14 and AAPM TG18, and then used to display the original image and decompressed image for the evaluation of image quality. Totally amounts to 199 medical images which from CR, DR, CT, DF, DSA, MRI, DM, RF, and US are used as testing images. The evaluation of compression methods in our study were classified into lossless and lossy compression two categories. In lossless compression evaluation, difference image method are used to verify the image if it could be recovered after decompressing. In lossy compression evaluation, first of all original images are compressed in different rate (from 5 to 400), and then calculated the PSNR value and verified with visual examination between original and decompressed images. Finally, the statistical analysis software, WINKS was used to analysis experimental data. The results show that the screens used in our study meet the requirement of DICOM standard. In lossless compression, we found all image in our study could be recovered completely with the compression rate of 3 in average. In lossy compression, Medical images acquired from different modalities could be in compression rate of 10 to 60. This result can be used as the reference for clinic people using JPEG2000 to compress medical images, in addition offer a feasible mechanism for screen calibration.