Journals Proceedings

Abstract

Karyotyping is a technique used to display and study the human chromosomes for detecting abnormalities, genetic disorders or defects. M-FISH (Multiplex Fluorescent In-Situ Hybridization) provides color karyotyping. In this paper, naïve Bayes classification of M-FISH chromosome images based on watershed based chromosome segmentation is presented. It is observed that the classification of the watershed regions by using the naive Bayes classifier works better than pixel by pixel classification. By adding the feature, standard deviation along with mean of each region, improved classification accuracy was observed. The approach was tested on a database and found to provide an accuracy of 73%.