Feature Extraction for the C Lassification of Human Chromosomes from G-Band Images Using Wavelets

- Chromosomes contain genes that provide the coded information for human beings to grow, develop and function. Any change in the number, size or structure of the chromosome leads to chromosomal abnormalities which will cause birth defects. However many of these defects are preventable, if detected earlier. The main objective of this work is to determine good features to classify human chromosomes and to detect chromosomal abnormalities from G-Band chromosome images. As a first step, chromosome images are analyzed using Discrete Wavelet Transform (DWT) to get coefficients which contain information about the banding pattern. The banding patterns allow a chromosome to be reliably differentiated from other chromosomes of same size and centromere position. From the coefficients, statistical features are calculated. A neural network may be used for further classification using these features. abnormalities