COVID-19 diagnosis prediction using classical-to-quantum ensemble model with transfer learning for CT scan images

ABSTRACT

At present, there are problems of low detection efficiency and accuracy in chest CT images of COVID-19 as well as limited computational power of deep learning model training. Developing a classical-to-quantum (CQ) ensemble model with transfer learning to efficiently detect patients with COVID-19 using chest CT images.: Attributes were extracted from chest CT scans using pre-trained networks ResNet50, VGG16 and AlexNet, while dressed quantum circuits were used as classifiers. The overall accuracy of the CQ method based on three aforementioned networks on the chest CT dataset is 83.2%, 86.2% and 85.0%, respectively. The proposed ensemble model has a precision of 89.0% for pneumonia samples, an overall accuracy of 88.6% and a pneumonia class recall rate of 83.0%. In addition, to further verify the robustness of the ensemble model, breast ultrasound and brain tumour images were used in it. The suggested ensemble approach is effective for classifying and detecting medical pictures with complicated features, particularly for detecting COVID-19 patients using chest CT images.

KEYWORDS:

• COVID-19
• deep learning
• transfer learning
• dressed quantum circuits
• ensemble model

Additional information

Funding

This work was supported by National Natural Science Foundation of China [grant number 61902158].

Notes on contributors

Wenqian Li

Wenqian Li, Master of Engineering, main research directions: computer vision and medical image processing.

Xing Deng

Xing Deng, Ph.D., Associate Professor, Research Interests: Computer Vision and Medical Image Processing.

Haorong Zhao

Haorong Zhao, Master of Engineering, main research directions: computer vision and medical image processing.

Haijian Shao

Haijian Shao, Ph.D., Associate Professor, main research directions: computer vision and medical image processing.

Yingtao Jiang

Yingtao Jiang, Ph.D., Professor, Research Interests: Computer Vision and Medical Image Processing.