Research
Multi-Class Tissue Segmentation of Histopathology Images
Pathologists assess tissue samples on glass slides via microscope, but this process suffers from intra-/inter-observer variability. We develop deep learning-based tissue segmentation models to objectively and reproducibly quantify histopathology images from various cancer types.
Efficient Annotation for Whole Slide Images
Annotating giga-pixel-sized whole slide images is time-consuming and labor-intensive. Furthermore, pathologists have limited time to annotate them. We introduce an efficient approach to annotate whole slide images to train segmentation models.
Survival Outcome and Treatment Response Prediction
It is our ultimate goal to predict survival outcome and treatment response for personalized medicine. We correlate prognostic factors quantified from histopathology images with patient outcome using interpretable AI models.
