Restoration of Medical Images Using Genetic Algorithms

Abstract

Image restoration is still one of the most important areas of medical image processing. Image restoration concerns about the removal or reduction of degradations in an image that could happen during the acquisition process. Being able to restore a medical image helps providing a better diagnosis and treatment. One of the most common blurring is the motion blur. Many restoration algorithms were proposed to solve the image restoration problem such as Wiener Filter, Lucy-Richardson and Blind Deconvolution Algorithms. These algorithms have varied performance, computational complexity, and abilities to deal with noisy images. They also require the knowledge of the Point Spread function (PSF) such that image deconvolution can be implemented. Restoration of an image is extremely reliant on the quality of the estimation technique used to find an accurate PSF parameters (i.e. motion length and motion angle). In this paper, we adopt Genetic Algorithms (GAs) to find the optimal PSF parameters such that a Wiener filter can be used for image restoration. We adopted number of statistical evaluation criteria to asses the quality of our proposed method. We applied our method on a number of medical images with various additive Gaussian noise. The developed results show that our proposed algorithm, PSF generated by GAs, is showing better results compared to other known methods in the literature in the absence of the real PSF.