A Deep Learning System for Detecting Severe Lung Diseases including Tuberculosis, Pneumonia, and COVID-19 through Digital Chest X-rays | Chapter 1 | New Horizons of Science, Technology and Culture Vol. 4

 

Lung diseases are the most common and dangerous health issues worldwide. Illnesses such as Tuberculosis (TB), Pneumonia, and COVID-19 are vital to worldwide health and need accurate detection for effective treatment. In this study, a deep learning-based model was proposed for multiclass classification to automatically diagnose these illnesses using chest X-ray (CXR) images. The model employs a Convolutional Neural Network (CNN) framework that has been trained on publicly available datasets to categorise CXRs as COVID-19, Pneumonia, Tuberculosis, and No-Findings. Data pre-processing techniques were employed, including image resizing and normalisation, along with stratified data splitting. The proposed model was evaluated with an accuracy rate of 98.5%, demonstrating strong performance throughout all classes, with precision, recall, and F1-score exceeding 96%. The Pneumonia category achieved the highest recall (99.8%), while the No-Findings category showed balanced performance with 99.4% recall and a 99.2% F1-score. The findings illustrate the model's reliability for practical application in clinical decision support systems. The project's future development will focus on enhancing the model's capability to handle various types of input data, including X-rays, CT scans, and other radiological imaging formats, to boost its versatility and effectiveness in multiple diagnostic contexts.

 

Author(s) Details

B. Sarada
Department of CSE(AI&ML), Ramachandra College of Engineering, AP, India.

 

Please see the book here:- https://doi.org/10.9734/bpi/nhstc/v4/5805