Publication

Abstract : Respiratory diseases are one of the major health concerns worldwide; especially in places with limited medical facilities. This project helps doctors to detect these diseases quickly and easily by using chest X-Ray images and artificial intelligence (AI). This proposed paper addresses key issues such as high computational cost and the challenge of real-time deployment. A customized 9-layer Convolutional Neural Network (CNN) is designed to classify six types of lung disease from X-rays. The model achieved a test accuracy of 93.68% and performed better than ResNet-18 in all key metrics. The confusion matrices showed consistent results in the training, validation, and test sets. GradCAM++ heat maps also matched actual disease-affected regions, making the model easier to trust. The system can be used in rural or resource-limited hospitals to help doctors to make quick and reliable decisions. It can support early diagnosis and better treatment using just X-rays and a lightweight AI model.