Klasifikasi Citra X-Ray Dada Deteksi Pneumonia Menggunakan Metode Convolutional Neural Network

This research focuses on classifying pneumonia images with Convolutional Neural Networks and testing the performance of CNN architecture model ResNet-50V2 with MobileNetV2 on pneumonia detection using lung X-ray images. In addition, the effect of enhancing image contrast with CLAHE is also compared. The dataset used in this study amounted to 5,840 X-ray images of lung pneumonia taken from the Kaggle.com website. Three detection performance parameters were used, namely accuracy; loss; and F1-Score. The performance of ResNet-50V2 is better than that of MobileNetV2. Both architectural models also implement CLAHE to increase the contrast of the image, CLAHE provides good results from both model performance. The accuracy value obtained reached 93% with an F1-Score value of 91%. In addition, the performance of pneumonia detection using lung X-rays with ResNet50-V2 can be improved by increasing the image contrast using CLAHE.