Diagnosis and multi-classification of lung diseases in CXR images using optimized deep convolutional neural network

A deep learning (DL) architecture is proposed in this study for the multi-class classification of COVID-19, lung opacity, lung cancer, tuberculosis (TB), and pneumonia. There are two distinct models, namely Classification_1 and Classification_2 in this research. Classification_1 detects the lung dis...

Full description

Saved in:
Bibliographic Details
Published in:Soft computing (Berlin, Germany) Germany), 2024-04, Vol.28 (7-8), p.6219-6233
Main Authors: Ashwini, S., Arunkumar, J. R., Prabu, R. Thandaiah, Singh, Ngangbam Herojit, Singh, Ngangbam Phalguni
Format: Article
Language:English
Subjects:
Accuracy
Algorithms
Application of Soft Computing
Artificial Intelligence
Artificial neural networks
Automation
Classification
Computational Intelligence
Control
COVID-19
Datasets
Deep learning
Engineering
Image retrieval
Lung cancer
Lung diseases
Machine learning
Mathematical Logic and Foundations
Mechatronics
Medical imaging
Neural networks
Opacity
Pneumonia
Robotics
X-rays
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A deep learning (DL) architecture is proposed in this study for the multi-class classification of COVID-19, lung opacity, lung cancer, tuberculosis (TB), and pneumonia. There are two distinct models, namely Classification_1 and Classification_2 in this research. Classification_1 detects the lung diseases and Classification_2 classifies the different lung diseases. The hyperparameters and architecture of the DL models are tuned by grid search optimization (GSO) to get accurate results. To meet the DL criteria, the enormous number of CXR images of COVID-19, lung opacity, pneumonia, lung cancer, TB, and normal images of 3615, 6012, 5856, 20,000, 1400, and 100,192, respectively, were reduced, normalized, and randomly divided. According to the experimental findings, our proposed model beat previous research with 99.82% accuracy in Classifcation_1 and 98.75% accuracy in Classification_2. The suggested paradigm offered greater performance, enabling medical professionals to identify and treat patients more rapidly and effectively.
ISSN:1432-7643
1433-7479
DOI:10.1007/s00500-023-09480-3