Deepnet for Detecting Analyzable Metaphases

Abstract

Automated Karyotyping System (AKS) is an essential computer aided system for chromsome image analysis, that in turn, helps the cytogenetic experts for the diagnosis, prognosis and treatment evaluation of genetic disorders and cancers. Many challenges have been faced by researchers for designing a fully automated system. One among them is the detection of analyzable metaphases, which are the input to the system. Conventional machine learning as well as deep learning techniques were adopted by researchers to classify the analyzable and unanalyzable metaphases. Here as well, a Convolutional Neural Network (CNN) is proposed to efficiently detect analyzable metaphases. It is found that the testing accuracy of the classifier is 85% eventhough the dataset is scarce.