Pneumonia Classification using CNN-GAN

Pneumonia is a lung condition that causes inflammation of the lungs because of the collection of fluid in the air sacs present in the alveoli. It can damage one or both lung this disease. According to the survey report conducted by the World Health organization, this disease affects children, infants, and people above the age of sixty-five. As the physical examinations done by radiologists cannot always provide an accurate result in the disease detection, and are also a bit time consuming so such deep learning algorithms come in use which decreases the load on the medical personals by automating the required processes. In this article three categories are taken in consideration listed as normal, viral, and bacterial pneumonia Chest X-rays the most common diagnostic tool for these diseases are used in the form of scanned images. The accumulated images from online repositories were taken and with the help of the GAN algorithm present dataset size was increased which helped in more effective training and testing of the model later with help of CNN the classification of the disease was done thereby using CNN and GAN approach together achieved an accuracy of 98.76% which when compared with other techniques was greater. Therefore, with help of this proposed system, ability to detect pneumonia is done more quickly and even with a small dataset classification of the disease can be done. It can play a good role for providing an aid for radiologists and doctors working in this medical field.