A Study of Genetic Algorithms on Polygonal Approximation and Medical Image Analysis

• Main Authors: Shu-Chien Huang, 黃樹乾
• Other Authors: Yung-Nien Sun
• Format: Others
• Language: en_US
• Published: 1999
• Online Access: http://ndltd.ncl.edu.tw/handle/99245402951516812333

博士 === 國立成功大學 === 資訊工程研究所 === 87 === Genetic algorithms mimic the process of natural evolution, the driving process for the emergence of complex and well-adapted organic structures. In the natural world, the fittest individuals survive, reproduce and then compete with each other. Genetic algorithms can simulate the survival of the fittest individuals and produce the near-optimal and stable solutions in complex search space for the optimization of problems in diverse fields. The first focus of this dissertation is on the applications of genetic algorithms to polygonal approximation. In general, there are two basic approaches associated with the polygonal approximation of n ordered points with a uniform error norm: (1) given the number of breakpoints, m, find a polygonal approximation with m breakpoints, such that its distance from the curve is minimal among all the approximations with m breakpoints, and (2) given an error , find the polygonal approximation with the minimal number of breakpoints such that the polygon is distant from the curve by no more than . Most of these existing approximation methods have the disadvantage of result dependency on the selection of starting points and the given arbitrary initial solutions. They usually yield unacceptable result if the starting point or the initial solution is inappropriate. In order to overcome the disadvantages described above, the proposed method tackles the polygonal approximation problems by employing the genetic algorithms. In the genetic process, three genetic operators, namely selection, crossover and mutation, and the fitness function have been introduced for polygonal approximation. Experiments have showed promising results and fast convergence of the proposed method. The second focus of this dissertation is on the applications of genetic algorithms to medical data analysis. We employ genetic algorithms to surface reconstruction. The 3-D surface can be constructed from a series of cross-sectional contours. First, the feature points are extracted from each CT contour by the polygonal approximation technique. Second, it then produces the triangular meshes by connecting corresponding feature points to form triangles between any two consecutive contours. Third, these triangular meshes are rendered. Our approach is able to obtain an efficient display while preserving the 3-D object shape. On the other hand, we proposed a new approach to obtain the initial endocardial border. This approach is capable of handling the inherent problems of image-dropout in echocardiogram. The initial contour estimation followed by GA-based deform model provides near-optimal endocardial contour refinement. Clinical validations of the methods proposed in this dissertation have been carried in the Veterans General Hospital-Kaohsiung and National Cheng-Kung University Hospital. Experimental results have demonstrated that the genetic algorithm is potentially a valuable tool for medical data analysis.