[Paper Review] Medical Imaging with Deep Learning for COVID- 19 Diagnosis: A Comprehensive Review

Subrato Bharati, Prajoy Podder+2 more|arXiv (Cornell University)|Jul 13, 2021

COVID-19 diagnosis using AIMedicine145 references33 citations

TL;DR

A comprehensive review of deep learning applications in COVID-19 medical imaging (X-ray and CT) and drug discovery, highlighting top DL methods and their performance, plus future directions.

ABSTRACT

The outbreak of novel coronavirus disease (COVID- 19) has claimed millions of lives and has affected all aspects of human life. This paper focuses on the application of deep learning (DL) models to medical imaging and drug discovery for managing COVID-19 disease. In this article, we detail various medical imaging-based studies such as X-rays and computed tomography (CT) images along with DL methods for classifying COVID-19 affected versus pneumonia. The applications of DL techniques to medical images are further described in terms of image localization, segmentation, registration, and classification leading to COVID-19 detection. The reviews of recent papers indicate that the highest classification accuracy of 99.80% is obtained when InstaCovNet-19 DL method is applied to an X-ray dataset of 361 COVID-19 patients, 362 pneumonia patients and 365 normal people. Furthermore, it can be seen that the best classification accuracy of 99.054% can be achieved when EDL_COVID DL method is applied to a CT image dataset of 7500 samples where COVID-19 patients, lung tumor patients and normal people are equal in number. Moreover, we illustrate the potential DL techniques in drug or vaccine discovery in combating the coronavirus. Finally, we address a number of problems, concerns and future research directions relevant to DL applications for COVID-19.

Motivation & Objective

• Summarize how DL models are applied to medical imaging for COVID-19 detection (X-ray, CT) and compare performance.
• Explain DL-based tasks beyond classification (localization, segmentation, registration) in COVID-19 imaging.
• Discuss DL applications in drug/vaccine discovery and identify current challenges and future research directions.

Proposed method

• Review of DL approaches for image classification, localization, segmentation, and registration in COVID-19 imaging.
• Compilation and comparison of reported accuracy metrics on X-ray and CT datasets.
• Synthesis of DL techniques applied to drug and vaccine discovery for COVID-19.

Experimental results

Research questions

• RQ1 What DL architectures and preprocessing strategies yield the highest accuracy for COVID-19 detection in X-ray and CT images?
• RQ2 What are the best-performing DL methods for COVID-19 image classification and related tasks (localization, segmentation, registration)?
• RQ3 What are the potential DL approaches and challenges in supporting drug/vaccine discovery for COVID-19?

Key findings

• Highest X-ray classification accuracy reported is 99.80% using InstaCovNet-19 on a dataset of 361 COVID-19, 362 pneumonia, and 365 normal cases.
• Highest CT classification accuracy reported is 99.054% using EDL_COVID on a 7500-sample dataset with equal numbers of COVID-19, lung tumor, and normal cases.
• DL techniques extend beyond classification to localization, segmentation, and registration in COVID-19 imaging.
• DL applications in drug or vaccine discovery for COVID-19 are discussed as a potential area of impact.
• The paper also addresses problems, concerns, and future research directions in DL for COVID-19.