A Novel and Robust Fatty Liver Recognition Method Based on Hybrid Generative Adversarial Networks on Ultrasound Images

The liver is the biggest gland in the body and is responsible for a wide variety of processes. The digestive system breaks down the food and liquids a person consumes so that the body can use them. It also aids in the fight against diseases by filtering out dangerous elements from the blood. Liver damage can be caused by exposure to viruses or harmful substances. The focus of this research will concentrate on developing methods to forecast the onset of the illness using machine learning (ML) algorithms. In this study, we introduce an information enhancement strategy for generating synthetic medical pictures with Generative Adversarial Networks (GANs). In order to increase the size and variety of the training set, we offer a training scheme that employs both traditional data augmentation and GAN approaches for synthetic data augmentation. In this article, we used many different types of deep neural network architecture to forecast the likelihood of developing liver illness. The most precise deep neural network framework was used to forecast the ultimate result. Our GAN model is so exact that it can accurately forecast the onset of liver disease. The best model for making the prediction, to determine if a person has liver disease, is then used by the system. Accuracy, precision, recall, F-measure, as well as area under the curve (AUC) were used to compare and contrast the suggested model with the existing model SegNet, both of which were developed to foretell the incidence of liver disease. The results of the study demonstrated that the suggested model evaluated better than the SegNet models in all metrics, including accuracy (96%), recall (93%), precision (95%), and specificity (96%). We further demonstrate that the combined use of synthetic pictures and real-world augmented data may considerably boost the effectiveness of GAN networks for the categorization of Nonalcoholic Fatty Liver Disease (NAFLD).