Issues associated with deploying CNN transfer learning to detect COVID-19 from chest X-rays
- PDF / 1,587,278 Bytes
- 15 Pages / 595.276 x 790.866 pts Page_size
- 21 Downloads / 182 Views
SCIENTIFIC PAPER
Issues associated with deploying CNN transfer learning to detect COVID‑19 from chest X‑rays Taban Majeed1 · Rasber Rashid2 · Dashti Ali3 · Aras Asaad4 Received: 17 May 2020 / Accepted: 25 September 2020 © Australasian College of Physical Scientists and Engineers in Medicine 2020
Abstract Covid-19 first occurred in Wuhan, China in December 2019. Subsequently, the virus spread throughout the world and as of June 2020 the total number of confirmed cases are above 4.7 million with over 315,000 deaths. Machine learning algorithms built on radiography images can be used as a decision support mechanism to aid radiologists to speed up the diagnostic process. The aim of this work is to conduct a critical analysis to investigate the applicability of convolutional neural networks (CNNs) for the purpose of COVID-19 detection in chest X-ray images and highlight the issues of using CNN directly on the whole image. To accomplish this task, we use 12-off-the-shelf CNN architectures in transfer learning mode on 3 publicly available chest X-ray databases together with proposing a shallow CNN architecture in which we train it from scratch. Chest X-ray images are fed into CNN models without any preprocessing to replicate researches used chest X-rays in this manner. Then a qualitative investigation performed to inspect the decisions made by CNNs using a technique known as class activation maps (CAM). Using CAMs, one can map the activations contributed to the decision of CNNs back to the original image to visualize the most discriminating region(s) on the input image. We conclude that CNN decisions should not be taken into consideration, despite their high classification accuracy, until clinicians can visually inspect and approve the region(s) of the input image used by CNNs that lead to its prediction. Keywords Coronavirus · Convolutional neural network · Deep learning · Class activation maps · COVID-19
Introduction
* Aras Asaad [email protected] Taban Majeed [email protected] Rasber Rashid [email protected] Dashti Ali [email protected] 1
Department of Computer Science and Information Technology, College of Science, Salahaddin University, Erbil, Kurdistan Region, Iraq
2
Department of Software Engineering, Faculty of Engineering, Koya University, Koya KOY45, Kurdistan Region, Iraq
3
Independent Researcher, Toronto, ON, Canada
4
Oxford Drug Design, Oxford Centre for Innovation, New Road, Oxford OX1 1BY, UK
The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus causing COVID-19, has become a pandemic since its emergence in Wuhan, China in Dec 2019 [1]. The death toll from the infection escalated and many health systems around the world have struggled to cope. A critical step in the control of COVID-19 is effective and accurate screening of patients so that positive cases receive timely treatment and get appropriately isolated from the public; a measure deemed crucial in curbing the spread of the infection. Reverse-transcription polymerase
Data Loading...
