Cycle Generative Adversarial Networks for Cell Diagnostics in Liver Fibrosis

Abstract

In pathological diagnosis, special stains are applied to identify disease areas. However, these methods are complex and time-consuming. In this study, to address these limitations, we developed a novel disease-specific virtual staining method using a cycle generative adversarial network, a type of image-generation artificial intelligence. Using rat livers, the model transformed simple hematoxylin and eosin stains to staining images for disease determination more quickly than traditional experimental methods. The evaluation using mathematical indices reinforces the method’s validity and clinical applicability, suggesting a significant advancement in pathology that could streamline diagnostic workflows and improve patient outcomes.