Seminars 25/26

The TIA seminar is a regular event, organised by the Tissue Image Analytics Centre, that usually takes place on the 1st and 3rd Mondays of each month between 2pm and 3pm. We invite researchers and leaders in the area of computational pathology and associated areas to discuss their work and stimulate thought-provoking discussions.

Take a look at the details of TIA seminars during 2024-25 academic year, most of which are also available on our YouTube channel.

Please contact the TIA Seminar Organiser Dr Adam Shephard via email (adam.shephard@warwick.ac.uk) if you are interested in attending a particular seminar below, or even giving a seminar. Alternatively, please fill-out the following form if you are interested in registering for the seminar series mailing list TIA Seminar Series Registration.

Note. We will continue to update this page with more details about the upcoming seminars.

Autumn 2025

Dr Nadieh Khalili

Radboud UMC, Nijmegen, The Netherlands (click here for a short bio)

13 October 2025

Title: Towards Multimodal AI for Personalized Cancer Treatment

Abstract: The integration of diverse data types—such as histopathology, radiology, and omics—offers a powerful opportunity to develop AI models that improve cancer diagnosis and treatment selection. In this talk, I will present recent efforts from our group in developing and validating multimodal deep learning frameworks that predict treatment-relevant biomarkers directly from routine diagnostic images. I will highlight case studies in bladder and prostate cancer and discuss challenges related to data harmonization, model generalization, and clinical translation.

Watch the seminar by clicking here.

---

Dr Tapabrata Chakraborty

University College London, London, UK (click here for a short bio)

27 October 2025

Title: Conformal Quantification of Predictive Uncertainty in Health AI

Abstract: For AI to be trusted for use in high risk applications like healthcare, it is important to ensure (AI Assurance) that the predictive uncertainty is within guaranteed bounds which can be achieved through conformal prediction based calibration at user defined level of significance.

Watch the seminar by clicking here.

---

Malte Kuehl

Aarhus University, Aarhus, Denmark (click here for a short bio)

10 November 2025

Title: Scaling multiplexed protein imaging analysis for subcellular-resolution pathology

Abstract: Proteins and their spatial distribution constitute the functional foundation of biology, offering critical insights into health and disease. However, probing pathological tissues at subcellular resolution remains challenging, from experimental design through image acquisition to computational analysis. I will present our recent work on Pathology-oriented multiPlexing (PathoPlex) published in Nature, which introduces key innovations spanning quality control, experimental design, and image acquisition, alongside spatiomic, a novel software library that accelerates spatial proteomics analysis at the pixel level and beyond. Finally, I will discuss how these developments, combined with advances in spatial statistics, agentic AI, and collaborative community efforts in open-source software, offer potential pathways for addressing remaining challenges in spatial proteomics.

Watch the seminar by clicking here.

---

Prof Jana Lipkova

University of California, Irvine, Irvine, USA (click here for a short bio)

17 November 2025

Title: Too Many Models, Too Few Benchmarks: Addressing the benchmarking crisis in AI for pathology

Abstract: The field of AI in pathology is experiencing a benchmarking crisis. New models emerge weekly, each claiming superiority over predecessors, yet the community lacks rigorous, reproducible tools to validate these claims. In this talk, I will present two open benchmarks from our group to address this gap. The first benchmark systematically evaluates state-of-the-art pathology foundation models (FMs) for survival prediction in oncology. We assess model generalization across multiple external cohorts, compare architectural paradigms (patch- vs. slide-level, uni- vs. multimodal), probe model interpretability and investigate weather FM allow to build pan-cancer generalists model that replace disease-specific models and generalize to rare cancers and conditions not seen during the model training? The second benchmark introduces a framework for safe AI deployment in pathology, focusing on out-of-distribution (OOD) detection and model “guardrails.” We propose a new taxonomy for OOD scenarios in pathology and a public benchmark that stress-tests existing OOD models across diverse data shifts and real-world variability. Both benchmarks, including all datasets, standardized splits, and evaluation protocols, are publicly released to serve as standards for model evaluation and comparison, thereby accelerate further advances and deployment of AI in pathology.

Watch the seminar by clicking here.

---

Prof Jin Tae Kwak

Korea University, Seoul, South Korea (click here for a short bio)

21 November 2025

Title: Pathology knowledge-guided AI models in computational pathology

Abstract: Computational pathology is undergoing a profound transformation driven large-scale artificial intelligence (AI), particularly the seamless integration of multiple data modalities through multimodal large language models (MLLMs). In this talk, I will present our group’s recent efforts in developing knowledge-guided computational pathology tools that utilize vision-language models for various diagnostic tasks. I will discuss how vision-language models, combined with expert knowledge, can aid in analyzing complex tissue patterns, enhancing interpretability, and achieving accurate and reliable diagnostic performance.

Watch the seminar by clicking here.

---

Dr Linda Studer

Radboud UMC, Nijmegen, The Netherlands

8 December 2025

Title: TBC

Abstract: TBC