Lightweight Classification Network for Pulmonary Tuberculosis Based on CT Images

Authors

  • Junlin Tian School of Mathematics and Statistics, Lanzhou University, China https://orcid.org/0000-0002-1409-9342
  • Yi Zhang The First Hospital of Lanzhou University, China https://orcid.org/0000-0002-7632-1922
  • Junqiang Lei Radiological Clinical Medicine Research Center of Gansu Province, China & Intelligent Imaging Medical Engineering Research Center of Gansu Province, China https://orcid.org/0000-0002-7632-1922
  • Chunyou Sun School of Mathematics and Statistics, Lanzhou University, China
  • Gang Hu Computer Information Systems Department, State University of New York at Buffalo State, USA

DOI:

https://doi.org/10.37965/jait.2023.0153

Keywords:

tuberculosis case classification, CNN, Transformer, lightweight network

Abstract

With the continuous development of medical informatics and digital diagnosis, the classifification of tuberculosis (TB) cases from computed tomography (CT) images of the lung based on deep learning is an important guiding aid in clinical diagnosis and treatment. Due to its potential application in medical image classifification, this task has received extensive research attention. Existing related neural network techniques are still challenging in terms of feature extraction of global contextual information of images and network complexity in achieving image classifification. To address these issues, this paper proposes a lightweight medical image classifification network based on a combination of Transformer and convolutional neural network (CNN) for the classifification of TB cases from lung CT. The method mainly consists of a fusion of the CNN module and the Transformer module, exploiting the advantages of both in order to accomplish a more accurate classifification task. On the one hand, the CNN branch supplements the Transformer branch with basic local feature information in the low level; on the other hand, in the middle and high levels of the model, the CNN branch can also provide the Transformer architecture with different local and global feature information to the Transformer architecture to enhance the ability of the model to obtain feature information and improve the accuracy of image classifification. A shortcut is used in each module of the network to solve the problem of poor model results due to gradient divergence and to optimize the effectiveness of TB classifification. The proposed lightweight model can well solve the problem of long training time in the process of TB classifification of lung CT and improve the speed of classifification. The proposed method was validated on a CT image data set provided by the First Hospital of Lanzhou University. The experimental results show that the proposed lightweight classifification network for TB based on CT medical images of lungs can fully extract the feature information of the input images and obtain high-accuracy classifification results.

Metrics

Metrics Loading ...

Downloads

Published

2023-01-11

How to Cite

Tian, J., Zhang, Y., Lei, J., Sun, C., & Hu, G. (2023). Lightweight Classification Network for Pulmonary Tuberculosis Based on CT Images. Journal of Artificial Intelligence and Technology, 3(1), 25–31. https://doi.org/10.37965/jait.2023.0153

Issue

Section

Research Articles