SIAMESE NEURAL NETWORKS FOR PANDEMIC DETECTION USING CHEST RADIOGRAPHS
International Journal on Technical and Physical Problems of Engineering
; 14(2):104-110, 2022.
Article
in English
| Scopus | ID: covidwho-1940041
ABSTRACT
The recent developments in the field of deep learning have enabled the efficient diagnosis of medical imaging for determining a broad set of diseases. To reduce the spread and impact of the pandemic (COVID virus), machine learning techniques can be used to diagnose and predict the disease using chest X-ray images. In this research, we present an approach using Siamese Convolutional Neural Network (SCNN) to classify chest x-ray images into four classes, namely pandemic, Severe-COVID, Pneumonia and Normal. We present a comparative study between the performance of our Siamese network and other pre-trained CNN architectures i.e. VGG-16 and ResNet50 in this research. The model performance is tested by merging two publicly available datasets COVID-Chest-Xray dataset and Chest X-Ray Images (Pneumonia). We achieved an accuracy of 98% on Siamese ResNet50 which gives the best performance in contrast to 95% on VGG-16, 93% on ResNet50 and 96% on Siamese VGG-16. © 2022, International Organization on 'Technical and Physical Problems of Engineering'. All rights reserved.
Search on Google
Collection:
Databases of international organizations
Database:
Scopus
Language:
English
Journal:
International Journal on Technical and Physical Problems of Engineering
Year:
2022
Document Type:
Article
Similar
MEDLINE
...
LILACS
LIS