Automatic chromosome classification using medial axis approximation and band profile similarity

Abstract

Automated chromosome classification is an essential task in cytogenetics and has been an important pattern recognition problem. Numerous attempts were made in the past to characterize chromosome to perform clinical and cancer cytogenetics research. It is important to determine optimum features and develop feature extraction schemes for chromosome classification. In this paper we propose novel approaches for medial axis determination and profile matching of human chromosome. The medial axis determination plays a critical role for precise and complete extraction of band patterns of chromosomes. The features of the band profile obtained along the axis are then used in the classification process. In particular, the medial axis is obtained by simple cross-section analysis and the classification is accomplished through iterative pairing of band profiles. According to the experimental results, the developed automatic system can efficiently extract band profiles of the chromosomes along their medial axes, and satisfactory chromosome classification results can be obtained.