The Application Study Of Medical Image Enhancement Based On The Fuzzy Mathematics Theory

Digital image processing(DIP) is an important technique for Computer medical diagnosis and treatment. However, due to the influence of X optical signal, the working environment of medical imaging device and imaging equipment condition, the medical image is contaminated with various noises. Therefore, it is necessary to obtain a higher quality image by using the DIP to enhance images, i.e., to filter noise, highlight the region of interest or image edge, which provides the basis for further analysis and calculations.In medical image analysis, the main challenge is that there exists uncertain similarity for various targets, which will lead to the difficulty to separate targets. Moreover, even in the imaging procedure, the edge of target is largely overlapped with the background so that the target boundary is blurred. Beside this, medical image processing lacks the certain information for the further analysis, which will usually need the prior knowledge. However, it is difficult to depict these image characteristics with the classical mathematical expression. Therefore, it is necessary to describe the medical image based on the fuzzy theory. Based on this, this dissertation aims to utilize fuzzy theory to model the medical image information, and then develops the corresponding methods to denoise and enhance. The main contribution of this dissertation is as below :1. As for the enhancement in the transform domain, an adaptive enhancement based on fuzzy theory is developed. This approach can transform the membership function in the fuzzy space, which can reasonably stretch the image gray to enhance the image detail information while compress noise. The comparison with the single level fuzzy enhancement demonstrates the effectiveness of the improved approach.2. Utilizing the more sensitivity of human eyes to the image detail than the noise, we improved the image enhancement algorithm based on physiological characteristics of human eyes. The application to the actual MRI confirms that the proposed method can extract the edge information more reliably compared with the image enhancement based on Lee or LIP models.3. By introducing fuzzy entropy, we realized a denoising approach based on fuzzy entropy. In this approach, a physical model corresponding to the degraded multi pixel values image. Then the fuzzy entropy is constructed to reflect theimage information. Finally, the minimum entropy criteria is utilized to denoise MRI images. The comparison with the mean and medium denoising methods shows that the developed approach can effectively remove the noises in medical images.4. Based on the developed fuzzy processing methods, a image analysis system with flexible extension is developed, which has the fundamental image processing functions such as display, clip, and enhancement.