This is a past event
Computational pathology represents a transformative paradigm in precision medicine, leveraging artificial intelligence to enhance diagnostic accuracy and streamline clinical workflows. This presentation showcases a comprehensive research portfolio addressing critical challenges across multiple pathological domains through advanced AI methodologies. Our work encompasses progressive context encoders for anomaly detection (P-CEAD) on whole slide images to assist tumor segmentation in melanoma and colorectal cancer. Additionally, we developed conditional GAN-based solutions for pathologist pen marking removal and interpretable attention-based multi-instance learning frameworks that predict cell-of-origin in diffuse large B-cell lymphoma using quantitative pathology features. Our innovations extend to cutting-edge applications including vision-language model approaches for virtual pathologist systems, and virtual staining techniques. These methodologies demonstrate superior performance in identifying complex histological patterns while maintaining clinical interpretability and workflow integration. Collectively, this research illustrates how computational pathology is revolutionizing precision medicine through intelligent automation, enhanced diagnostic capabilities, and improved patient care outcomes across diverse medical specialties.
- Speaker
- Quincy Gu
- Venue
- Teams (online)