Explainability Of Artificial Intelligence For Diagnosing COVID-19 From Chest X-Rays
2021 International Conference on Computational Performance Evaluation, ComPE 2021
; : 598-603, 2021.
Article
in English
| Scopus | ID: covidwho-1831736
ABSTRACT
This COVID-19 pandemic has overburdened the government and the healthcare system of many countries around the world. It has brought up the need for a fast and accurate diagnosing method. Artificial intelligence (AI) is having a notable role in different aspects of the pandemic- contact tracing, epidemiology, medical diagnosis and prognosis, and drug development. Deep learning has found its application in the diagnosis of COVID-19 chest X-rays (CXR) using convolution neural nets. Many architectures have been used and transfer learning is the most preferred approach. These models have proven to be fast and accurate in COVID-19 diagnosis. However, one key element that has prevented the use of AI in clinical practice is its lack of transparency and explainability. In this paper, we use the ResNet-50 pre-trained model for classifying the CXR of COVID-19 patients from pneumonia and normal patients. We then use explainability algorithms to visualize the model features and verify the explainability of the model. © 2021 IEEE.
Full text:
Available
Collection:
Databases of international organizations
Database:
Scopus
Language:
English
Journal:
2021 International Conference on Computational Performance Evaluation, ComPE 2021
Year:
2021
Document Type:
Article
Similar
MEDLINE
...
LILACS
LIS