Pneumonia is a lung disease that could be caused by bacteria, viruses, fungi, or parasites. The pulmonary cysts are filled with fluid, causing croup and mucus cough. Usually, observation of the patient's lung condition is performed through X-rays. However, the quality of X-ray images tends to be less than optimal. Therefore, a CAD-based automation system was developed. In this paper, a new chest X-rays dataset for pneumonia cases is classified by using Convolutional Neural Network (CNN). This study examines the CNN performance in handling the new dataset. The data were obtained from the Kaggle platform. In total, there were5,840 images occupied in this study, consisting of 1,575 normal lung images and 4,265 pneumonia lung images. The data were divided into training and testing data, with the amount of data 5,216 and 624 images on each, respectively. The CNN activation function applied the Rectifier Linear Unit (ReLU) function, Adam optimization function, and epoch as many as 200times. Based on the test results, the average accuracy and loss values are sequentially at 89.58% and 47.43%. The results of this test indicate that the CNN method is quite capable of classifying the pneumonia cases.
Copyrights © 2020