bebi 21(3): e5

Research Article

Semi-supervised Learning for COVID-19 Image Classification via ResNet

Download106 downloads
  • @ARTICLE{10.4108/eai.25-8-2021.170754,
        author={Lucy Nwosu and Xiangfang Li and Lijun Qian and Seungchan Kim and Xishuang Dong},
        title={Semi-supervised Learning for COVID-19 Image Classification via ResNet},
        journal={EAI Endorsed Transactions on Bioengineering and Bioinformatics},
        volume={1},
        number={3},
        publisher={EAI},
        journal_a={BEBI},
        year={2021},
        month={8},
        keywords={COVID-19 Image Classification, Semi-supervised Learning, Residual Neural Network, Joint Optimization, Data Imbalance},
        doi={10.4108/eai.25-8-2021.170754}
    }
    
  • Lucy Nwosu
    Xiangfang Li
    Lijun Qian
    Seungchan Kim
    Xishuang Dong
    Year: 2021
    Semi-supervised Learning for COVID-19 Image Classification via ResNet
    BEBI
    EAI
    DOI: 10.4108/eai.25-8-2021.170754
Lucy Nwosu1, Xiangfang Li1,2, Lijun Qian1,2, Seungchan Kim1, Xishuang Dong1,2,*
  • 1: Center of Computational Systems Biology, Department of Electrical and Computer Engineering, Prairie View A&M University, Texas A&M University System, Prairie View, TX 77446, USA
  • 2: Center of Excellence in Research and Education for Big Military Data Intelligence (CREDIT Center), Department of Electrical and Computer Engineering, Prairie View A&M University, Texas A&M University System, Prairie View, TX 77446, USA
*Contact email: xidong@pvamu.edu

Abstract

INTRODUCTION: Coronavirus disease 2019 (COVID-19) is an ongoing global pandemic in over 200 countries and territories. Analysis of X-ray imaging data can play a critical role in timely and accurate screening and fighting against COVID-19.

OBJECTIVES: Supervised deep learning dominates COVID-19 pathology data analytics. However, it requires a substantial amount of annotated X-ray images to train models, which is often not applicable to data analysis for emerging events.

METHODS: The proposed model with two paths is built based on Residual Neural Network for COVID-19 image classification to reduce labeling efforts, where the two paths refer to a supervised path and an unsupervised path, respectively.

RESULTS: Experimental results demonstrate that the proposed model can achieve promising performance even when trained on very few labeled training image.

CONCLUSION: The proposed model can reduces the efforts of building deep learning models significantly for COVID-19 image classification.