Deep learning techniques have begun applied to the practice of medicine. Deep learning may, for example, assist physicians by showing pathogenic structures within diseases, including cysts, tumors, or infected tissue, or by recognizing objects such as cellular nuclei, or tumorous liver growths. However, the development of diagnostic deep learning techniques has been hampered by a scarcity of clinical image data. This is particularly true in the dental field.
Accordingly, we propose a method of generating dental images using Generative Adversarial Networks (GANs). In this paper, we compare the performance of a Mask R-CNN model for image segmentation trained on a dataset that included GAN-generated data against a Mask R-CNN model trained with a dataset that did not include GAN-generated data.
Ultimately, we observed improved performance by the Mask R-CNN that employed our GAN-generated images.