Indirect supervision applied to COVID-19 and pneumonia classification

• Published in: Informatics in medicine unlocked Vol. 28; p. 100835
• Main Authors: Danilov, Viacheslav V., Proutski, Alex, Karpovsky, Alex, Kirpich, Alexander, Litmanovich, Diana, Nefaridze, Dato, Talalov, Oleg, Semyonov, Semyon, Koniukhovskii, Vladimir, Shvartc, Vladimir, Gankin, Yuriy
• Format: Journal Article
• Language: English
• Published: England Elsevier Ltd 01.01.2022; The Authors. Published by Elsevier Ltd; Elsevier
• Subjects: Classification; COVID-19; Deep learning; Indirect supervision; Pneumonia; Transfer learning; COVID-19; Deep learning; Pneumonia; Transfer learning; Classification; Indirect supervision
• ISSN: 2352-9148; 2352-9148
• DOI: 10.1016/j.imu.2021.100835

The novel coronavirus 19 (COVID-19) continues to have a devastating effect around the globe, leading many scientists and clinicians to actively seek to develop new techniques to assist with the tackling of this disease. Modern machine learning methods have shown promise in their adoption to assist the healthcare industry through their data and analytics-driven decision making, inspiring researchers to develop new angles to fight the virus. In this paper, we aim to develop a CNN-based method for the detection of COVID-19 by utilizing patients' chest X-ray images. Developing upon the inclusion of convolutional units, the proposed method makes use of indirect supervision based on Grad-CAM. This technique is used in the training process where Grad-CAM's attention heatmaps support the network's predictions. Despite recent progress, scarcity of data has thus far limited the development of a robust solution. We extend upon existing work by combining publicly available data across 5 different sources and carefully annotate the comprising images across three categories: normal, pneumonia, and COVID-19. To achieve a high classification accuracy, we propose a training pipeline based on indirect supervision of traditional classification networks, where the guidance is directed by an external algorithm. With this method, we observed that the widely used, standard networks can achieve an accuracy comparable to tailor-made models, specifically for COVID-19, with one network in particular, VGG-16, outperforming the best of the tailor-made models. [Display omitted] •Network training based on indirect supervision results in accuracy comparable to tailor-made networks made for distinguishing COVID-19 and pneumonia.•VGG-16 trained using guided attention has demonstrated the most accurate classification at a level of 88% and 84% on the validation and testing subsets respectively.•Standard deep learning approaches do not activate around patterns that point to COVID-19 or pneumonia.