Curve Structure Detection In Biological Image In The Zebrafish

At post-genome times, tremendous amount of images and relevant data generated from high-throughput screening in the biological assays in order to study the function of genome, the relationship of the genotype and phenotype, drug discovery and Systems Biology. Accurate analysis and quantitative interpretation of these data are in great need to answer the given biological question. So automatic or semiautomatic statistical analysis and system modeling are required.Zebrafish has recently emerged as an invaluable vertebrate system for disease modeling and drug discovery, and been considered as one of the most important biology for the study of biomedicine. Developing a computerized zebrafish image processing and analysis pipeline would be a significant step towards reproducible quantitation of Zebrafish phenotypes in large-scale or high-throughput imaging studies. As an integral part of the ZFIQ (Zebrafish image quantitator) project of Harvard Medical School and Biomedical Imaging and Analysis Joint Lab in Northwestern Polytechnic University, a semiautomatic and an automatic curvilinear detection methods were proposed in the thesis in order to better understand the relationship between genotype and phenotype. The main work of the thesis was summarized as follows:1. Classified the curvilinear detection methods into six categories, and mainly introduced the pattern recognition techniques, model-based approaches and tracking-based approaches. Finally the applications of the curvilinear detection methods in biomedicine were presented, especially the applications in the detection of Zebrafish axon, neurites in fluorescence microscopy images and vessels in CTA images.2. Proposed a semiautomatic curvilinear detection method based on the Dijkstra's shortest-path algorithm. Firstly, the modified Hessian matrix was constructed. Then the eigenvalues and eigenvectors were computed, which were used to construct the cost function. Once the start point and the endpoint were set, the direction map corresponding to the start point was formed. Then the curve was tracked in the reverse direction from the endpoint to the start point according to the Dijkstra's shortest-path algorithm. Thus the curve was detected. The experiment results on Zebrafish axon images of different modalities indicated that the method can detect the main branches of the curvilinear structure and determine the length of the curve.