Advances in Information Systems (General Track)

Paper Type

Complete

Paper Number

1742

Description

Covid-19 Diagnosis needs new Information Systems technologies as Deep learning methods, especially in medical image screening. We aim to review the applications of deep learning augmented systems in Covid-19 predictions with the help of a large literature collection from four major databases IEEE explore, ACM, Web of Science, and PubMed. We have identified three major research themes from the current literature, Image Classification, Image segmentation, and evaluation methods for DL models. Among the DL techniques, Transfer Learning is identified as the most popular method for different tasks on Chest X-rays and CT scans. Pre-trained models such as ResNet, VGG, DenseNet, and Unet are widely used in the covid-19 diagnosis. While these models are pre-trained on natural images, a Chest X-ray image pre-trained model CheXnet is gaining popularity in Covid-19 image tasks helping in improving accuracies of classifications.

Share

COinS
 
Aug 9th, 12:00 AM

Applications of Deep Learning Augmented Systems for Covid-19 Predictions- A Literature Review

Covid-19 Diagnosis needs new Information Systems technologies as Deep learning methods, especially in medical image screening. We aim to review the applications of deep learning augmented systems in Covid-19 predictions with the help of a large literature collection from four major databases IEEE explore, ACM, Web of Science, and PubMed. We have identified three major research themes from the current literature, Image Classification, Image segmentation, and evaluation methods for DL models. Among the DL techniques, Transfer Learning is identified as the most popular method for different tasks on Chest X-rays and CT scans. Pre-trained models such as ResNet, VGG, DenseNet, and Unet are widely used in the covid-19 diagnosis. While these models are pre-trained on natural images, a Chest X-ray image pre-trained model CheXnet is gaining popularity in Covid-19 image tasks helping in improving accuracies of classifications.