CheXNet for the Evidence of Covid-19 Using 2.3K Positive Chest X-rays

• Published in: Recent Trends in Image Processing and Pattern Recognition Vol. 1576; pp. 33 - 41
• Main Authors: Santosh, KC, Ghosh, Supriti
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2022; Springer International Publishing
• Series: Communications in Computer and Information Science
• Subjects: Chest X-ray; CheXNet; Covid-19; Deep learning
• ISBN: 3031070046; 9783031070044
• ISSN: 1865-0929; 1865-0937
• DOI: 10.1007/978-3-031-07005-1_4

CheXNet is not a surprise for Deep Learning (DL) community as it was primarily designed for radiologist-level pneumonia detection in Chest X-rays (CXRs). In this paper, we study CheXNet to analyze CXRs to detect the evidence of Covid-19. On a dataset of size 4, 600 CXRs (2, 300 Covid-19 positive cases and 2, 300 non-Covid cases (Healthy and Pneumonia cases)) and with k(=5) fold cross-validation technique, we achieve the following performance scores: accuracy of 0.98, AUC of 0.99, specificity of 0.98 and sensitivity of 0.99. On such a large dataset, our results can be compared with state-of-the-art results.