A Survey of Multimodal Foundation Models in Computational Pathology

PhD Qualifying Examination


Title: "A Survey of Multimodal Foundation Models in Computational Pathology"

by

Mr. Jin CHENG


Abstract:

Recent advancements in large-scale multimodal foundation models have
significantly accelerated computational pathology, shifting the focus from
algorithmic development toward clinical deployment where high benchmark accuracy
must be accompanied by interpretability, robustness, and systemic
trustworthiness. To navigate this transition, this survey systematically
reviews the evolution of foundational architectures and the methodological
pathways designed to bridge the gap between raw predictive power and clinical
utility. Specifically, we explore how the field tackles persistent technical
bottlenecks by examining the design of multi-instance learning architectures
for gigapixel context modeling, alongside strategies utilizing privileged
information learning and test-time adaptation to mitigate severe label scarcity
and distribution shifts. Beyond these algorithmic solutions, the survey projects
a fundamental evolution toward systemic clinical intelligence through holistic
multi-omics integration. Ultimately, we advocate for a shift from static
predictive models to multimodal agentic systems that leverage dynamic tool use
and retrieval-augmented generation, delineating a pathway toward reasoning-
capable and trustworthy artificial intelligence for clinical pathology.


Date:                   Friday, 6 March 2026

Time:                   2:00pm - 4:00pm

Venue:                  Room 2132C
                        Lift 22

Committee Members:      Dr. Hao Chen (Supervisor)
                        Dr. Dan Xu (Chairperson)
                        Dr. Terence Tsz Wai Wong (CBE)