LCCC Seminar with Joel Saltz, MD (Stony Brook University)

About the Speaker:   Dr. Joel Saltz is a Distinguished Professor of Biomedical Informatics at Stony Brook University and a pioneer in computational pathology and AI. A Fellow of the American College of Medical Informatics, he has founded three Biomedical Informatics departments. He holds an MD and a PhD in Computer Science, completed his residency at Johns Hopkins, and is board-certified in Clinical Pathology.

Abstract: Routine Hematoxylin and Eosin (H&E) stained slides harbor a wealth of diagnostic and prognostic information that extends far beyond what conventional visual inspection can extract. This talk traces the evolution of computational pathology, from foundational digital infrastructure like the Virtual Microscope to advanced deep learning methodologies. I will discuss our recent pathology AI research, including latent conditional diffusion models for predicting immunohistochemistry (IHC) staining directly from H&E, the generation of privacy-preserving synthetic tissues, explainable pathology AI, and techniques for extracting pathology synoptic elements using biologically interpretable models. I will also describe our collaborative work with UNC in breast cancer, focusing on the predictive value of tumor-infiltrating lymphocytes (TILs) and the use of quantitative tissue morphology to accurately forecast clinical outcomes.