Using Support Vector Machine and Generative Adversarial Network for Multi-Classification of Pneumonia Disease
Sanchit Vashisht, Bhanu Sharma, Shweta Lamba
Abstract
Due to the buildup of fluid in the air sacs found in the alveoli, pneumonia is a lung ailment that results in lung inflammation. This illness can harm one or both lungs. Children, newborns, and those over 65 are all affected by this disease, according to the World Health Organization's survey study. Deep learning algorithms are used to reduce the workload on medical personnel by automating the necessary procedures since physical exams performed by radiologists do not always yield an accurate result in the illness identification process and are also somewhat time-consuming. In this article, the classifications of normal, viral, and bacterial pneumonia are taken into account. The most commonly used method of diagnosing these disorders is chest X-ray scans. The current dataset was made larger with the help of the GAN algorithm, which made it possible to train and tests the model more effectively. Later on, the disease was categorized with the help of SVM; when compared to other approaches, the SVM and GAN approach's combined accuracy was 98%. As a result, even with a small amount of data, the suggested method enables quicker diagnosis and classification of pneumonia. It might be of assistance as a second opinion to radiologists and other medical professionals who work in this area.