Human Visual Lossless Method of Diagnostic Medical Image Analysis in Wavelet Domain
The JPEG algorithm used to analyze frequency domain of medical diagnostic image is not proper for processing high resolution medical diagnostic image because of blocking effect, leading to loss of information. Thus, in the paper, JPEG2000 algorithm including Discrete Wavelet Transform (DWT) was employed and we used lossless (5, 3) tap filter and (3, 5) tap filter for decomposition and synthesis, respectively. Using DWT, input image were decomposed to 7 sub-bands and we analyze each sub-bands by abstracting statistical value of them. Also, with analyzing result, in the synthesis process, we performed selective band rejection according to priority of each sub-bands. The original input image and restored image applied the proposed algorithm are compared with 2 evaluation methods, Mean Squared Error (MSE) and Entropy. Consequently, in the human visually respect, we achieved lossless restored image with selective band rejection and it can be effective for processing high resolution and large size medical diagnostic image.
Keywords - Diagnostic Medical Image, Band rejection, Power, Entropy, MSE (mean squared error), DWT (Discrete Wavelet Transform)