Application of Deep Learning Algorithms for Discerning the Presence of Pneumonia

Abstract

Pneumonia is a fatal disease that involves inflammation of the air sacs in the lungs, resulting in breathing difficulties. As a result, early discovery of the condition is critical, as it can be deadly in its later stages, leading to respiratory issues. Chest X-rays have long been used to reliably diagnose Pneumonia. Human-assisted diagnosis, on the other hand, has constraints such as the availability of an expert and involves significant expenditures for the necessary equipment. As a result, there has been a spike in demand for alternative methods for identifying pneumonia from chest x-rays using Deep Learning. To aid this, the usage of Convolutional Neural Networks (CNN) and other classification algorithms has increased and is now being used in important decision-making processes in the medical profession. This paper compares several convolutional neural network architectural models such as VGG-16, InceptionV3, DenseNet169, and CNN for identifying the presence of pneumonia using chest x-ray images, and finally, the models have been evaluated using performance metrics for better analysis.