| Abstract [eng] |
Radiological images allow for quick and accurate diagnosis of diseases. However, radiologists are often unable to devote much time to the analysis of X-rays, which can lead to the omission of important details. To alleviate this problem, a number of computer-aided detection and diagnostic methods have been proposed, of which the deep-learning method is the latest to be applied. The aim of this study is to identify new possibilities for more efficient X-ray image processing, find the most efficient model architectures, and explore the possibilities of synthetic image generation capabilities of generative adversarial network models to improve the training and recognition accuracy of deep learning models. |
