Evaluación del rendimiento de una Red Neuronal Generativa Antagónica en la Generación de un banco de imágenes de radiografías panorámicas dentales humanas

Abstract

From engineering, different technological tools have been proposed to support the exercise of health professionals, in such developments large volumes of medical information are analyzed and processed with which the populations that are intended to be beneficiaries are characterized. However, to access a person's medical record data, it is necessary to have their permission, as it is private and confidential information, so obtaining several miles of data can become tedious, delay the investigative process or limit the results. Such is the case of several projects that have been developed in the area of Dentistry, particularly work related to the use of digital image processing algorithms and artificial intelligence techniques, in which the authors raise the need to access banks more images to improve the performance of your solutions. An alternative is the creation of realistic synthetic images through digital editing and recently through the implementation of antagonistic generative neural networks. For this reason, in this document, the performance of an Antagonic Generative Neural Network is evaluated in the generation of an image bank of panoramic human dental radiographs from 300 real images, which verified the quality of the images generated by the algorithm of machine learning through a survey that was carried out to dental health professionals and people who do not know whatsoever on the subject,