Chromosome Segmentation using Skeleton Generated from Rule-based Contour Vector Matching

S. Pornsuksiri and R. Lipikorn (Thailand)

Keywords

Karyotype, Overlapping chromosome segmentation, Rulebased contour vector, skeleton

Abstract

Karyotype is the way chromosomes in each living cell are organized in order to study and evaluate their character istics, such as their position, size, and shape. Karyotype can be used to determine abnormality, the cause of dis ability or birth defects, and gender. Thus, karyotyping is a common task in cytogenetic tests which consists of separating chromosomes then arranging and pairing chromosomes. However, one of the main problems in chromosome karyotyping is how to segment touching and overlapping chromosomes which makes segmentation difficult to perform. This paper presents a robust chromosome segmentation using skeleton generated from block decomposition and our proposed rule-based contour vector matching which allows us to detect separation points of touching and/or overlapping chromosomes and then segment them correctly. For the experiments, each chromosome image is converted to a binary image and a skeleton of each chromosome in an image is generated using block decomposition and rule based contour vector matching. By using these proposed method, we obtain not only skeletons of chromosomes but also other properties, such as relative length, area, and den sity, that are very useful for chromosome pairing which is the next step in karyotyping. The experimental results were compared with those of other existing methods and they confirm the novelty of our proposed method.

Important Links:

Go Back