Seminars

All talks are on Wednesdays at 4pm in the John Morgan Building, Class of '62 Auditorium & are recorded unless otherwise noted

 

 

November 11, 2026

Spyridon Bakas, PhD, Associate Professor of Pathology & Laboratory Medicine, University of Indiana

 

Hosted by AI2D - invited by Christos Davatzikos

BakasAbstract: The modern landscape of data-driven cancer research and care is similarly complex across health systems and universities. AI is reshaping the landscape of medicine, by helping us see what we cannot see through subvisual cues hidden in images and tissue, as well as by operational AI already embedded in clinical workflows across the diagnostic continuum of cancer patients. Value-guided evidence generation and generalizable AI models of clinical utility require access to large datasets through either public releases, or multi-institutional collaborations (such as the ReSPOND consortium). World leading example studies of federated learning are shown as a paradigm shift for multi-institutional collaboration, enabling privacy-aware model development and benchmarking across sites without moving data. A comprehensively validated computational pathology study is shown that delivers conclusive glioma diagnosis using histology alone, when clinical practice mandates the use of molecular profiling. Looking ahead, preliminary results are shown for integrated, multi-modal diagnostic, prognostic, & predictive models, combining histology, radiology, and molecular views of disease.

Bio: Spyridon (Spyros) Bakas is an internationally recognized leader in AI-driven oncology, federated learning, & computational pathology, with Elsevier ranking him among the world’s top 0.5% most-cited scientists across disciplines. He is an Endowed Chair Associate Professor and the Computational Pathology Division Director in the Dept. of Pathology & Laboratory Medicine, with secondary appointments in Neurosurgery; Radiology; Biostatistics & Health Data Science; & Computer Science, at Indiana University in Indianapolis (IN, USA). Outside IU, he is in the MICCAI Society Board of Directors & the Chair of the clinical AI-RANO cooperative group. Dr Bakas leads a federally funded research group focused on bridging biomedical imaging & AI to address clinical requirements. His work extends to the creation of multi-institutional multi-modal datasets & federated learning that protect patient privacy while enabling global collaboration. He has co-authored >150 peer-reviewed manuscripts & >100 abstracts, with collaborators across academic ranks & disciplines, as well as with industry. His publications have shaped current best practices in computational oncology, influencing both clinical workflows & translational AI research.

Dec 9, 2026

Ragini Verma, PhD, Professor of Radiology, University of Pennsylvania

 

Hosted by AI2D - invited by Christos Davatzikos

Abstract: TBA

Bio: TBA

Decmber 16, 2026

Zhijian Yang, PhD, Research Scientist at Google

 

Hosted by AI2D - invited by Christos Davatzikos

Abstract: Foundation models are transforming Health AI by learning generalizable representations from large-scale unlabeled data, enabling efficient development of new health features, models, and products with lightweight adaptation and limited labels. In this talk, Zhijian will trace the development of health foundation models from clinical imaging to everyday physiological monitoring. The talk will begin with a 3D MRI foundation model that learns from large-scale scans and radiology reports, as well as label-efficient adaptation of segmentation foundation models via prompting and preference optimization. Moving from the hospital to everyday life, Zhijian will introduce a hierarchical foundation model pretrained on millions of hours of smartwatch sensor data (photoplethysmography and accelerometry) and evaluated across chronic, acute, and incident diseases. He will also present its application to blood pressure monitoring, spanning from clinical validation to deployment on consumer devices. The talk will conclude with a discussion of opportunities to connect everyday health data with clinical data

Bio: jian Yang, PhD, is a Research Scientist at Google working on Health AI. Dr. Yang's current research develops foundation models and representation learning methods for longitudinal wearable sensor data, such as photoplethysmography and accelerometry from consumer smartwatches and fitness trackers, to enable passive, continuous monitoring of health and early detection of disease in everyday life. Before joining Google, Dr. Yang was an AI Scientist at GE HealthCare, leading the development of a vision‑language foundation model for 3D MRI. Dr. Yang received a PhD in Applied Mathematics and Computational Science from the University of Pennsylvania, advised by Dr. Christos Davatzikos. Dr. Yang's doctoral research developed generative and weakly supervised deep learning methods that integrate MRI and genetic data to disentangle the heterogeneity of brain aging and neurodegenerative disease, with publications in Nature Medicine, Nature Communications, and ICLR.