The automatic classification of normal and abnormal chromosomes using image analysis

Abstract

The mutagenic effect of chemicals is assessed using a procedure which examines their effect on the shapes of chromosomes in cells. Even slight changes in chromosome configurations can adversely affect an organism as a whole. The identification of damage in cells is time consuming. Automating the identification would considerably reduce the procedure time. This paper gives a new way of characterizing damaged chromosomes and distinguishing them from normal chromosomes. The methodology builds on previous work which is relevant, but not directly related to the problem.

The approach is based on the fact that all normal chromosomes have a characteristic X shape, which can be described by their medial axis. The most common chemical mutation structures, exchange figures, do not exhibit this characteristic. Using the medial axis in conjunction with other features, it is possible to classify a chromosome as normal or abnormal.