Multi-Classification of Respiratory Diseases using Deep Learning

Abstract

Lung diseases can have serious health consequences and cause distressing respiratory symptoms. Chest X-rays are often used to diagnose lung disease because it provides important visual data about the lungs. This study presents a Custom ResNet50 model for analyzing patterns and predicting the presence of three diseases, namely pneumonia, tuberculosis, and COVID-19. The model is trained with a dataset of 5,700 chest X-rays from Kaggle. An accuracy of 98.45% is obtained, showing that the finetuned model outperforms traditional machine learning algorithms and accurately classifies different pulmonary diseases with a high level of confidence. This research has the potential to greatly improve the diagnostic process for pulmonary diseases and provide more accurate and efficient treatment options for patients. As a result, these diseases can be identified and treated early, reducing their severity and likelihood of transmission.