Abnormal Chromosome Identification Using Hybrid Image Segmentation Method

Abstract

Image processing is a technique used for collecting and analyzing the visual information by a mechanized streamlined PC using models and methods. Chromosomes are highly organized and discrete subunits of nuclear genomes of humans, animals, and plants. The number of chromosome is species specific and any alteration in chromosome, either numerical or structural, is known as chromosomal aberration. The manual segmentation of overlapping chromosome is the tedious and experiences intra and intra-changeability impacts. This paper proposed an Enhanced Probabilistic Neural Network for chromosome classification is proposed. After that we proposed the Fuzzy Classifier strategy gave the most elevated classification accuracy among the current classifiers. It reduces the computation and time complexity as well as achieves better classification accuracy. There is around 4 % to 5% increase in generally speaking classification accuracy with fuzzy and statistical classifiers respectively compared to different approaches.