Twice Optimization System For Image Segmentation Based On Active Contour Model

The imaging technology has been essential means for medical diagnosis in present. Image segmentation is not only the base of comprehension for medical image, but also the bottle neck restricting the development and application of other related image processing technology. It is important for Bio-medical research, clinic diagnosis and pathology analysis.Magnetic resonance imaging (MRI) is un-interventional and no damnification, which has extensive use in clinic diagnosis. The image can provide data with high resolution, and its merit of high-contrast and high SNR (signal-to-noise ratio) of parenchyma especially for diagnosis of brain etc. The article describes the segmentation of lateral ventricles in MRI. Compared to traditional segmentation, Active Contour Model (ACM) is more suitable for medical image processing. According the superincumbent thought, it can get fine result for the medical image with high noise and weak boundary through various constraint conditions.It has been 30 years since the ACM was published. In this period, a lot of improved algorithms or extensional algorithms based on this model appeared, and were applied in medical image processing for different problems. In this article, the segmentation of Lateral Ventricles is not satisfied in sharp angular point through the traditional ACM. In order to improve segmentation quality with manageability, it adds genetic algorithm (GA). It is a searching algorithm based on population not individual, so the probability of getting into local optimum point is low. It differentiates the quality of individuals through fitness function and there is no more other operation. The article describes a twice optimization algorithm for the segmentation of Lateral Ventricles and establishes a segmentation system. Firstly, it applies the traditional ACM optimizing the original contour and the result just draws the smooth part of the object image, and failure is in sharp angular point; secondly, it applies genetic algorithm for second optimization and get the satisfied result, and the fitness function bases the energy function of defined ACM in former step.The Twice Optimization algorithm provides a feasible solution for segmentation of sharp angular point in medical image.