Classification of X-ray images of COVID-19 based on CNN and improved swin transformer model

Fubao Song; Jiaqing Mo; Jiangwei Zhang

doi:10.1117/12.3005845

10 October 2023 Classification of X-ray images of COVID-19 based on CNN and improved swin transformer model

Fubao Song, Jiaqing Mo, Jiangwei Zhang

Author Affiliations +

Proceedings Volume 12799, Third International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2023); 1279954 (2023) https://doi.org/10.1117/12.3005845
Event: 3rd International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2023), 2023, Kuala Lumpur, Malaysia

Abstract

Faced with the rapid spread of COVID-19, nucleic acid testing methods can detect positive cases relatively quickly. Still, the time-consuming detection and frequent false-negative issues have led to a sharp increase in the demand for alternative diagnostic tools for COVID-19. In this paper, using medical imaging technology and deep learning technology, a model combining a convolutional neural network and an improved Swin Transformer network is designed to detect chest X-ray images of COVID-19. The image is input into the convolutional layer to extract the local details of the image. Then, to solve the problem that some heads do not play a role in calculating multi-head self-attention due to too small a weight, a learnable bias parameter is added to each individual computing head to enhance the specific weight of each head. Experiments show that this method has a recognition rate of 98.25% for chest X-ray images of COVID-19. Indicators such as recall rate and F1 score have been improved compared with some current methods.

(2023) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Fubao Song, Jiaqing Mo, and Jiangwei Zhang "Classification of X-ray images of COVID-19 based on CNN and improved swin transformer model", Proc. SPIE 12799, Third International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2023), 1279954 (10 October 2023); https://doi.org/10.1117/12.3005845

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

;

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE