Segmentation Of CDNA Microarray Image Using Fuzzy C-means Algorithm Optimized By Particle Swarm

For gene contains a large number of genetic information, researches on it have profound significance. However, early methods can't fix the problem of low efficiency in dealing with high-throughput information. A gene chip technology with high efficiency and accuracy has been focused from it appears in 1980s to now. Gene chips have been applied in many fields, and image processing is a critical integral step which can obtain information in the gene chip.The thesis focuses on the image processing of gene chip. The key steps of image processing include image preprocessing, grid location, image segmentation, evaluation of segmentation algorithm and signal extraction, and they are introduced respectively in the thesis. Image preprocessing and gridding are the preparations which facilitate the segmentation. Segmentation is one of the most steps, which can affect the final signal extraction result. The thesis mainly focuses on the segmentation of gene chip image. The entire process of segmentation is introduced, including algorithm and evaluation of algorithm. After summarizing the previous algorithm, an adaptive segmentation algorithm based on Fuzzy c-means clustering is proposed. Besides this, an improved Fuzzy c-means clustering algorithm optimized by particle swarm is also proposed. The improved algorithm has better performance in noisy situation and can't easily fall into local optimum.A variety of evaluation criteria for segmentation have been described, and the thesis proposes a novel ultimate measurement accuracy criteria based on gene expression ratio. Using these criteria, segmentation algorithm are evaluated and compared.