Deep learning approach for analysis and characterization of COVID-19

Indrajeet Kumar, Sultan S. Alshamrani, Abhishek Kumar, Jyoti Rawat, Kamred Udham Singh, Mamoon Rashid, Ahmed Saeed AlGhamdi

研究成果: Article同行評審

1 引文 斯高帕斯(Scopus)

摘要

Early diagnosis of a pandemic disease like COVID-19 can help deal with a dire situation and help radiologists and other experts manage human resources more effectively. In a recent pandemic, laboratories perform diagnostics manually, which requires a lot of time and expertise of the laboratorial technicians to yield accurate results. Moreover, the cost of kits is high, and well-equipped labs are needed to perform this test. Therefore, other means of diagnosis is highly desirable. Radiography is one of the existing methods that finds its use in the diagnosis of COVID-19. The radiography observes change in Computed Tomography (CT) chest images of patients, developing a deep learning-based method to extract graphical features which are used for automated diagnosis of the disease ahead of laboratory-based testing. The proposed work suggests an Artificial Intelligence (AI) based technique for rapid diagnosis of COVID-19 from given volumetric chest CT images of patients by extracting its visual features and then using these features in the deep learning module. The proposed convolutional neural network aims to classify the infectious and non-infectious SARS-COV2 subjects. The proposed network utilizes 746 chests scanned CT images of 349 images belonging to COVID-19 positive cases, while 397 belong to negative cases of COVID-19. Our experiment resulted in an accuracy of 98.4%, sensitivity of 98.5%, specificity of 98.3%, precision of 97.1%, and F1-score of 97.8%. The additional parameters of classification error, mean absolute error (MAE), root-mean-square error (RMSE), and Matthew's correlation coefficient (MCC) are used to evaluate our proposed work. The obtained result shows the outstanding performance for the classification of infectious and non-infectious for COVID-19 cases.

原文English
頁(從 - 到)451-468
頁數18
期刊Computers, Materials and Continua
70
發行號1
DOIs
出版狀態Published - 2021

All Science Journal Classification (ASJC) codes

  • 生物材料
  • 建模與模擬
  • 材料力學
  • 電腦科學應用
  • 電氣與電子工程

指紋

深入研究「Deep learning approach for analysis and characterization of COVID-19」主題。共同形成了獨特的指紋。

引用此