AIVIS, a Seoul AI Hub-affiliated company under the Seoul Metropolitan Government, announced on the 12th that it received the Certificate of Merit (Outstanding Poster Award) at the 2025 Annual Conference of the Korean Society of Artificial Intelligence in Medicine (KoSAIM). This recognition officially acknowledges the technical achievement of AIVIS’s joint research project, conducted by AIVIS’s AI Research Team (Sungyong Jang, Daehong Lee) and Professor Joseph Jung of the Department of Pathology, The Catholic University of Korea College of Medicine. The study, titled “Enhancing Accuracy in Immunohistochemistry (IHC) Quantification Using a Transformer-Based Next-Generation AI Model,” was recognized for its innovation and performance. IHC (Immunohistochemistry) staining is an essential diagnostic procedure in pathology. However, accurately identifying and quantifying only the diagnostically relevant “tumor cells” among the vast number of cells on a slide remains a major challenge. Conventional deep learning AI models, such as CNNs, have shown limitations in distinguishing tumor cells from non-tumor cells (e.g., immune or normal cells), leading to reduced accuracy and reliability in quantitative analyses. To address this issue, the AIVIS research team developed a new AI model based on the latest Vision Transformer (ViT) architecture. This model enables AI to clearly distinguish individual nuclei as either “tumor” or “non-tumor” cells, even in complex tissue environments, thereby improving the precision of quantitative IHC analysis. The research team validated the model’s performance using 239 gastric carcinoma (stomach cancer) tissue samples for the representative IHC quantification biomarker Ki-67. Results showed that the existing CNN-based AI model (YOLOv9) exhibited significant discrepancies—13 to 21 percentage points lower accuracy compared to manual measurements by pathologists. In contrast, the AIVIS model reduced this gap to just 1.2 percentage points under a “free-hand” setting where pathologists manually defined regions of interest. This near-perfect alignment with expert analysis demonstrates that the ViT-based model is highly effective for complex IHC quantification tasks. Through this KoSAIM Outstanding Poster Award, AIVIS aims to further showcase the technological competitiveness of Seoul AI Hub’s innovation ecosystem on a global stage and expand its partnerships in the field of precision medicine worldwide. Daehong Lee, CEO of AIVIS, stated, “This award is especially meaningful because it recognizes AIVIS’s next-generation AI technology at Korea’s most prestigious medical AI society. We will continue to advance our R&D efforts to overcome the limitations of existing methods and deliver reliable, AI-powered quantitative pathology solutions that can be trusted in real-world clinical settings.”