TIA Centre Seminars
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 2023-24 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.
Spring 2025
| Date | Speaker | Title |
| 7th January 2025 (Tuesday 10AM) |
Azra Raza Columbia University Medical Center |
Redefining Hope: The Early Detection Revolution in Cancer Care |
| 13th January 2025 |
Tong Ding Harvard Medical School |
|
| 20th January 2025 |
Lee Cooper Northwestern Medicine |
|
| 3rd February 2025 |
Kun-Hsing Yu Harvard Medical School |
A pathology foundation model for cancer diagnosis and prognosis prediction |
| 17th February 2025 |
Lucy Godson National Pathology Imaging Co-operative (NPIC) |
|
| TBC 2025 |
Danielle Belgrave GSK |
|
| TBC 2025 |
Max Lu Harvard Medical School |
Autumn 2024
| Date | Speaker | Title |
| 21st October 2024 |
Peter Neidlinger Technical University Dresden |
Benchmarking foundation models as feature extractors for weakly-supervised computational pathology |
| 4th November 2024 |
Jack Breen University of Leeds |
Foundation Models and Multiple Instance Learning Methods for Ovarian Cancer Subtype Classification |
| 11th November 2024 |
Nicholas Trahearn Institute of Cancer Research |
|
| 25th November 2024 |
Spencer Watson University of Lausanne |
Investigating Glioblastoma Recurrence with Spatial Multi-Omics
|
| 11th December 2024 (Wednesday 2PM) |
Hanwen Xu University of Washington |
Towards learning patient level representations for better clinical outcome |
Autumn 2024
Peter Neidlinger
Technical University Dresden, Dresden, Germany (click here for a short bio)
21 Oct 2024
Title: Benchmarking Foundation Models as Feature Extractors for Weakly-Supervised Computational Pathology
Abstract: Advancements in artificial intelligence have driven the development of numerous pathology foundation models capable of extracting clinically relevant information. However, there is currently limited literature independently evaluating these foundation models on truly external cohorts and clinically relevant tasks to uncover adjustments for future improvements. In this study, we benchmarked ten histopathology foundation models on 13 patient cohorts with 6,791 patients and 9,493 slides from lung, colorectal, gastric, and breast cancers. The models were evaluated on weakly-supervised tasks related to biomarkers, morphological properties, and prognostic outcomes. We show that a vision-language foundation model, CONCH, yielded the highest performance in 42% of tasks when compared to vision-only foundation models. The experiments reveal that foundation models trained on distinct cohorts learn complementary features to predict the same label and can be fused to outperform the current state of the art. Creating an ensemble of complementary foundation models outperformed CONCH in 66% of tasks. Moreover, our findings suggest that data diversity outweighs data volume for foundation models. Our work highlights actionable adjustments to improve pathology foundation models.
Watch the seminar by clicking here.
Jack Breen
University of Leeds, Leeds, UK (click here for a short bio)
4 Nov 2024
Title: Foundation Models and Multiple Instance Learning Methods for Ovarian Cancer Subtype Classification
Abstract: Ovarian cancer histological subtyping is a vital diagnostic task as the subtypes represent vastly different diseases with varied treatment options and prognoses. It has previously been an underrepresented task in AI subtyping research. In this talk, I will discuss my thesis "Artificial Intelligence for Ovarian Cancer Diagnosis from Digital Pathology Slides", which focuses on the robust validation of multiple instance learning (MIL) subtyping models, and explores factors limiting the clinical utility of these models. Histopathology foundation models show great promise across many tasks, but analyses have typically been limited by arbitrary hyperparameters that were not tuned to the specific task - we recently reported the most rigorous single-task validation of histopathology foundation models to date (https://arxiv.org/abs/2405.09990). We have also explored the effects of replacing standard MIL methods with a multi-resolution graph network (https://arxiv.org/abs/2407.18105), alongside earlier studies focused on the computational efficiency of classification. Our other work has included investigating the effects of chemotherapy on morphological subtyping, predicting treatment response effectiveness, and the automated detection of metastases in the lymph nodes and omentum. The modern classifiers have drastically improved diagnostic performance, though some vital hurdles still need to be cleared for these models to achieve clinical utility.
Watch the seminar by clicking here.
Dr Nicholas Trahearn
Institute of Cancer Research, London, UK (click here for a short bio)
11 Nov 2024
Title: Investigating the Spatial Diversity of the Prostate Cancer Microenvironment using Artificial Intelligence
Abstract: Current prostate cancer risk predictors are not able to fully capture a patient’s risk of recurrence at the time of diagnosis, with cure rates varying substantially between patients in the same risk category, and is particularly challenging for the broad range of patients assessed as “high-risk”. In this talk I will discuss work from our recent Nature Cancer paper (https://doi.org/10.1038/s43018-024-00787-0), which assesses the impact of newly derived cancer evolution metrics as predictors of outcome within a clinical trial of high-risk prostate cancer patients. Patients enrolled on this trial received a combination of androgen deprivation therapy (ADT) and intensity-modulated radiotherapy (IMRT), and have received a median of 12 years of post-treatment follow-up. Using over 600 FFPE blocks from patients enrolled on the trial, we generated a large dataset of matched histology and whole genome sequencing to test our evolutionary metrics. Through Artificial Intelligence based Digital Pathology tools, we characterised the tumour microenvironment, at diagnosis, on a per-gland and cellular level. From this, we saw that our AI-derived Gleason grading was able to outperform that of the trial pathologist, raising interesting questions regarding current grading guidelines. Furthermore, our new measure of morphological evolvability was shown to be an independent predictor of prostate cancer recurrence and metastasis, events that would occur up to a decade later. Remarkably, these patterns were also reflected in our analysis of genomic evolvability and, when combined, produced a prediction of patient outcome that was stronger than the sum of its parts. By encapsulating different aspects of cancer evolution at different scales, this study demonstrates the added value of a multi-modal strategy to prostate cancer risk prediction.
Watch the seminar by clicking here.
Dr Spencer Watson
University of Lausanne, Lausanne, Switzerland (click here for a short bio)
25 Nov 2024
Title: Investigating Glioblastoma Recurrence with Spatial Multi-Omics
Abstract: Glioblastoma recurrence is currently inevitable despite advances in standard-of-care treatment. An alternative approach of targeting the tumor microenvironment, specifically tumor-associated macrophages via CSF-1R inhibition was found to dramatically regress established tumors in preclinical trials. While tumor regression was sustained in ~50% of individuals, recurrent tumors emerged over time in the remaining subset. These recurrences were ubiquitously associated with fibrotic scars that had formed during treatment. This same fibrotic response to treatment was also observed following surgical resection, radiotherapy, and in patient recurrence samples. Investigating the complex evolving post-treatment tumor landscape required integrating multiple -omics approaches with new spatially resolved transcriptomics. Spatial multi-omics analyses of the post-treatment tumor microenvironment identified specific fibrotic domains as pro-tumor survival niches that encapsulated surviving glioma cells, which inhibited immune surveillance while maintaining the tumor cells in a dormant state. Integrated analyses revealed specific pathways in the initiation of fibrosis that could be targeted in combination with anti-glioma therapy. These combinatorial therapies inhibited treatment-associated fibrosis, and significantly improved survival in preclinical trials of anti-CSF-1R therapy.
Watch the seminar by clicking here.
Hanwen Xu
University of Washington, Seattle, USA (click here for a short bio)
11 Dec 2024
Please note this seminar will be held on Wednesday at 2PM.
Title: Towards learning patient level representations for better clinical outcome
Abstract: Multi-modal medical data, including radiology and pathology imaging data, and genomics data are generated over the long cancer patient journey. These data are pivotal in guiding clinicians in clinical decision making, thus greatly impact how people optimize the clinical outcome. The complexity of interpreting massive multi-modal medical data for each individual patient highlights the need for a computational approach to learn the comprehensive representation. Developing such a multi-modal foundation model is faced with significant challenges, especially considering the long patient context. Here we develop a two-stage model architecture that can efficiently handle the long context problem in the medical domain. We primarily start by applying our model to whole slide pathology images, and trained the first slide-level foundation model GigaPath that shows superior performance on 25 subtyping and pathomics tasks. GigaPath is also aligned with real-world pathology reports, and shows capability in handling multi-modal tasks. Taken together, GigaPath is a good starting point towards developing multi-modal medical foundation models for better clinical outcome.
Watch the seminar by clicking here.
Spring 2025
Prof Azra Raza
Columbia University Medical Center, New York, USA (click here for a short bio)
6 Jan 2025
This is a special event, being held on Monday from 7-8:30PM, in person, at The Oculus Auditorium (OC1.05).
Title: Rhymes of Healing: Bridging Poetry & Medicine
Abstract: Few art forms can articulate the ineffable, foster resilience, and provide solace in the face of life’s trials more beautifully than poetry. Yet, its profound healing potential has not been mined for more practical applications to the daily lives of both the doctor and patient. Drawing from personal experiences in medicine and cancer research, as well as a lifelong devotion to poetry, I will examine how Urdu poetry, with its intricate metaphors and deeply emotional resonance, and English poetry, with its rich narrative and structural diversity, offer complementary pathways to healing. Together, they embody a universal language of empathy and connection that transcends cultural and linguistic barriers. The rhymes of healing serve as a therapeutic lens through which we can process grief, find strength, and celebrate hope. And then, there is the pure ecstasy of apprehending the deeper meaning of poetry’s ability to capture the soul in a moment of reflection.
To attend this event, please register here.
Prof Azra Raza
Columbia University Medical Center, New York, USA (click here for a short bio)
7 Jan 2025
This is a special edition of our TIA Centre Seminar Series, being held on Tuesday at 10AM.
Title: Redefining Hope: The Early Detection Revolution in Cancer Care
Abstract: The truth is, for many patients, cancer therapy today is hardly better than it was 100 years ago. On top of that, chemotherapy medicines, especially the newest advanced drugs, leave bereaved families with staggering medical bills they have no ability to pay. Doctors universally agree the best way to slay the dragon of cancer is by early detection. I have advocated finding cancer before it finds us for three decades. My ideas about prevention finally gained traction because recent discoveries revealed the ways early and even recurring cancers originate. We believe our approach to cancer prevention will be affordable worldwide. Although I practice medicine in New York, in treating cases in my native Pakistan I saw patients so poor they could not afford medicine without letting their children go hungry. They could choose medicine or food, but not both. Preventing cancer would make such vile choices go away.
To prove our approach, we focus on detecting The First Cell Stage of cancer in patients who have already survived cancer once. About 20% of new cancers are diagnosed in people who have already had a different type of cancer before. We will monitor Cancer Survivors using highly sophisticated technology to catch the very earliest cells and bodily signals associated with cancer inception. Technology is evolving in our labs to neutralize these early cells and signals so cancer can be prevented altogether. Our long-term goal is to develop implantable biosensors for continuous, 24/7 monitoring from birth to death, from womb to tomb, making real-time intervention possible at the First Cell Stage rather than waiting for it to develop even to Stage 1 cancer.
Tong Ding
Harvard Medical School, Boston, USA
13 Jan 2025
Title: Multimodal Whole Slide Foundation Model for Pathology
Abstract: The field of computational pathology has been transformed with recent advances in foundation models that encode histopathology region-of-interests (ROIs) into versatile and transferable feature representations via self-supervised learning (SSL). However, translating these advancements to address complex clinical challenges at the patient and slide level remains constrained by limited clinical data in disease specific cohorts, especially for rare clinical conditions. We propose TITAN, a multimodal whole slide foundation model pretrained using 335,645 WSIs via visual self-supervised learning and vision-language alignment with corresponding pathology reports and 423,122 synthetic captions generated from a multimodal generative AI copilot for pathology. Without any finetuning or requiring clinical labels, TITAN can extract general-purpose slide representations and generate pathology reports that generalize to resource-limited clinical scenarios such as rare disease retrieval and cancer prognosis. We evaluate TITAN on diverse clinical tasks and find that TITAN outperforms both ROI and slide foundation models across machine learning settings such as linear probing, few-shot and zero-shot classification, rare cancer retrieval and cross-modal retrieval, and pathology report generation. The model is publicly accessible at https://github.com/mahmoodlab/TITAN.
Prof Lee Cooper
Northwestern Medicine, Chicago, USA
20 Jan 2025
Title: TBC
Abstract: TBC
Prof Kun-Hsing Yu
Harvard Medical School, Boston, USA
3 Feb 2025
Title: A pathology foundation model for cancer diagnosis and prognosis prediction
Abstract: TBC
Dr Lucy Godson
National Pathology Imaging Co-operative (NPIC), Leeds, UK
17 Feb 2025
Title: TBC
Abstract: TBC
Dr Danielle Belgrave
GSK, London, UK
TBC 2025
Title: TBC
Abstract: TBC
Dr Ming Lu
Harvard Medical School, Boston, USA
TBC 2025
Title: A Multimodal Generative AI Copilot for Human Pathology
Abstract: TBC