Data generation from medical image using generative adversarial networks

The use of deep learning methodologies is greatly beneficial for applications that seek to automatically generate captions or descriptions from images. Deep learning networks offer significant support, especially in the medical field. In the diagnosis of diseases, doctors can use applications developed to generate captions and descriptions from medical images automatically. Generating textual descriptions from images is one of the most challenging scientific endeavors in image processing. In this study, we focus on the use of Generative Adversarial Networks (GAN) in medical image processing to support artificial intelligence in medical diagnosis. The goal is to create a natural language processing system to accurately describe X-ray images to help doctors diagnose and treat patients. The model in the study is designed to understand chest X-rays and generate descriptions from these X-ray images. To generate predictive annotation text, the "Open-I" dataset consisting of X-ray images and their corresponding diagnostic texts was selected. A Global Attention Convolutional Neural Network model was used to perform these operations. Analyses on the dataset showed that the designed system works effectively with accuracy values of 0.37 for the BLEU-1 metric and 0.39 for the Rouge-L metric.