Varifocal-Net: A Chromosome Classification Approach Using Deep Convolutional Networks

Abstract

Chromosome classification is critical for karyotyping in abnormality diagnosis. To expedite the diagnosis, we present a novel method named Varifocal-Net for simultaneous classification of chromosomes type and polarity using deep convolutional networks. The approach consists of one global-scale network (G-Net) and one localscale network (L-Net). It follows three stages. The first stage is to learn both global and local features. We extract global features and detect finer local regions via the G-Net. By proposing a varifocal mechanism, we zoom into local parts and extract local features via the L-Net. Residual learning and multi-task learning strategies are utilized to promote high-level feature extraction. The detection of discriminative local parts is fulfilled by a localization subnet of the G-Net, whose training process involves both supervised and weakly-supervised learning. The second stage is to build two multi-layer perceptron classifiers that exploit features of both two scales to boost classification performance. The third stage is to introduce a dispatch strategy of assigning each chromosome to a type within each patient case, by utilizing the domain knowledge of karyotyping. Evaluation results from 1909 karyotyping cases showed that the proposed Varifocal-Net achieved the highest accuracy per patient case (%) of 99.2 for both type and polarity tasks. It outperformed state-of-the-art methods, demonstrating the effectiveness of our varifocal mechanism, multi-scale feature ensemble, and dispatch strategy. The proposed method has been applied to assist practical karyotype diagnosis.

CIR-Net: Automatic Classification of Human Chromosome Based on Inception-ResNet Architecture

“…Recently, several deep learning-based approaches have been employed in chromosome classification, such as Siamese Networks [1], Attention Based Sequence Learning [8], vanilla Convolutional Neural Network (Vanilla-CNN) [9] and Varifocal-Net [10].…”

“…Yulei et al [10] proposed Varifocal-Net for chromosome classification using deep learning. This approach consists of one global-scale network (G-Net) and one local-scale network (L-Net).…”

“…This paper focuses on chromosome classification because it is a very significant but a labor-intensive stage in auto-karyotyping. Although many researches [1], [8], [9], [10] have made some contribution for chromosome classification tasks, classifying chromosomes accurately and robustly in the clinical application on the condition of the insufficient labeled dataset is still a challenging task for the following reasons:…”

Deep Metric Learning Network using Proxies for Chromosome Classification and Retrieval in Karyotyping Test

“…Researchers proposed and used various machine learning and deep learning techniques for automating the karyotyping process and obtained encouraging results. Most studies used deeply stacked convolutional neural networks for chromosome classification (Hu et al, 2019;Zhang et al, 2018;Sharma et al, 2018b and some studies employed feature based deep learning methods (Qin et al, 2019;Jindal et al, 2017).…”

DeepACE: Automated Chromosome Enumeration in Metaphase Cell Images Using Deep Convolutional Neural Networks

“…However, despite that there are some methods have been developed to solve classification [7,10] and segmentation [2,6] problems of chromosomes, very few of the researches have tried to develop automated chromosomes enumeration method. Gajendran et.al [4] presented a study by combining a variety of preprocessing methods and counting algorithm based on topological analysis, but the error rate is high.…”