Chromosome classification via deep learning and its application to patients with structural abnormalities of chromosomes

Abstract

Background and objectiveKaryotyping is an important technique in cytogenetic practice for the early diagnosis of genetic diseases. Clinical karyotyping is tedious, time-consuming, and error-prone. The objective of our study was to develop a single-stage deep convolutional neural networks (DCNN)-based model to automatically classify normal and abnormal chromosomes in an end-to-end manner. MethodsWe analyzed 2,424 normal chromosomes and 544 abnormal chromosomes. A preliminary support vector machine (SVM) model was developed to evaluate the basic recognition performance on the dataset. A DCNN-based model was then proposed to process the same dataset. ResultsBy utilizing the SVM model, the classification accuracy of 24 normal chromosomes was 86.01 %. The 32 types of normal and abnormal chromosomes got an accuracy of 85.37 %. The accuracy of the DCNN-based model performing the 24 normal chromosomal classification was 91.75 %. The accuracy of the 32 type classification was 87.76 %. To differentiate eight common structural abnormalities, we obtained accuracies that ranged from 90.84 % to 100 %, and the values of the AUC ranged from 91.81 % to 100 %. ConclusionsOur proposed DCNN-based model effectively performed the karyotype classification in an end-to-end manner. It had the competence to be used as a prediction tool for abnormal karyotype detection and screening in genetic diagnosis without initial feature extraction. We believe our work is meaningful for genetic triage management to lower the cost in clinical practice.