The goal of this project is to develop a significantly improved software for chromosome anomaly detection, by characterizing a chromosome banding pattern, displaying its essential characteristics, establishing idealized representations for specific types, and quantitatively comparing chromosomes to prototypes and to each other. This will improve the specificity of chromosome band pattern characterization, particularly with high resolution banding, in order to detect and quantify sub-visible band pattern alterations. Specifically, we will develop chromosome image analysis software that will (1) extract the essence of the band pattern similarities of a group of chromosomes, (2) display an idealized image showing the characteristics they have in common, (3) classify chromosomes by comparing each to the prototype for each homolog type, (4) compare homologs to each other to detect subtle abnormalities, (5) compare segments of chromosomes to detect anomalies, (6) develop prototypical chromosome images for groups of people manifesting genetically linked diseases and compare these with normals. We will use the Eigenvector method of decomposing chromosome images into components, selecting those components which reflect the relevant characteristics of the banding pattern, while eliminating components that reflect preparatory artefact, inter-chromosome variation, and other irrelevant differences. This new capability will significantly increase the ability of automated chromosome analysis instruments to evaluate chromosome alterations in neoplastic and in normal mammalian cells.
