편집
3,168
번
바뀜
편집 요약 없음
[[image:Research_simulation.png|1200px|center]]
== Digital Staining Using Generative Adversarial Networks ==
Digital staining is an innovative staining technique that uses artificial intelligence to overcome the limitations of traditional histological staining (HS) performed by experts. This technology was developed based on generative adversarial networks (GANs), using supervised and unsupervised learning approaches. In addition, to address the differences in staining quality and color between laboratories, single and multi-domain transformation staining methods were developed. The developed models prevented overstaining, distinguished backgrounds more accurately than traditional unsupervised methods, and achieved staining results more similar to those of HS. Furthermore, a Virtual Staining GUI system was developed to enhance usability. This study proved the feasibility of applying supervised learning in environments without paired datasets and presented the potential for advancing multi-staining research using unstained images. These advancements are expected to improve the accuracy of pathological analyses and enable automated staining methods for various pathological images.
[[image:Research_digitalstaining.png|2000px|center]]