Bigpicture Annual Meeting

The Bigpicture project is an initiative aimed at transforming the field of pathology through the creation of a comprehensive, high-quality digital repository of three million pathology images. Funded by the EU Innovative Medicines Initiative (IMI), Bigpicture brings together a consortium of leading academic institutions, research organisations, and industry partners from across Europe and beyond. The project’s mission is to develop an open-access platform that leverages artificial intelligence and machine learning to analyse vast amounts of pathology data, advancing research, diagnostics, and treatment in healthcare. By fostering collaboration and innovation, Bigpicture envisions a future where digital pathology becomes a cornerstone of precision medicine, enabling faster, more accurate diagnoses and personalised patient care.

The Bigpicture project is structured into six Work Packages (WPs), each focusing on a critical aspect of building and leveraging a large-scale digital pathology platform. Further information on the Bigpicture work packages can be found here.

How the TIA Centre is Contributing to Bigpicture

The TIA Centre is a key contributor to the consortium and the Bigpicture project predominantly focused around Work Package 4 (WP4), focusing on advancing AI-driven computational pathology. This work package develops tools for accessing, annotating, and analysing digital pathology slides. It leverages the Cytomine platform for WSI access and visualisation and focuses on creating task-agnostic AI models, federated learning strategies, and an integrated app store for AI algorithms. The goal is to enable efficient data mining and algorithm development. Learn more about WP4.

In particular, the TIA Centre’s focus is on:

• Domain generalisability research, ensuring AI models perform reliably across diverse datasets and institutions.
• Developing AI tools, such as automated Tumor-Infiltrating Lymphocytes (TILs) detection and nuclear instance segmentation/classification, to enhance diagnostic accuracy and efficiency.
• Integrating these tools into Cytomine, BigPicture’s platform, enabling seamless AI-powered slide analysis and annotation for users.
• Creating standardised metadata models (MSMom) to document AI models transparently, promoting reproducibility and interoperability.

By bridging cutting-edge AI research with practical platform integration, TIA Centre strengthens Bigpicture’s goal of transforming digital pathology into a scalable, standardised resource for global healthcare innovation.

Bigpicture Consortium Annual Meeting in Uppsala

Each year, the Bigpicture consortium gathers to share insights, solve challenges, brainstorm ideas, and plan future steps. After a year of biweekly online meetings, the annual event is also a chance to reconnect with colleagues in person, fostering collaboration and camaraderie. This year, the meeting was hosted by Uppsala University in the historic city of Uppsala, Sweden, from January 27 to 29, 2025.

The Venue and its Charm

The meeting took place in the University Main Building, a stunning historic venue that reflects Uppsala’s rich academic heritage, dating back to 1477. The city itself, with its cobblestone streets and medieval architecture, provided a picturesque backdrop for the event. Attendees were treated to a consortium dinner at the Uppsala Castle, a 16th-century landmark, and a guided tour of the Gustavianum Museum, home to an anatomical theatre from 1662. The blend of history and innovation perfectly mirrored Bigpicture’s mission to modernise pathology while respecting its roots.

The Annual Meeting

The three-day event was filled with a variety of activities, including:

· Demonstrations: On the first day, participants explored the latest AI developments in pathology through interactive laptop and poster presentations. Highlights included tools for TIL scoring, whole-slide image registration, and AI model benchmarking. We presented our work on benchmarking domain generalisation in computational pathology in a poster format. This was jointly led by Neda Zamanitajeddin and myself, where we investigated the performance of thirty different algorithms for domain generalisation over 3 different CPath tasks. Computer scientists at the event found this work very useful, especially one of our findings that modality specific methods such as stain augmentation could outperform the rest of the algorithms. Readers that are interested in this work can refer to our preprint.

· Workshops: Eight interactive sessions were held over the next two days, covering topics like data upload bottlenecks, AI model development, honest broker processes, and indirect access tools. These workshops aimed to gather input, align priorities, and address technical and policy challenges.

· Networking and Social Events: The consortium dinner at Uppsala Castle and the museum tour offered opportunities for informal discussions and team bonding.

Key Takeaways

Attending the Bigpicture consortium meeting in Uppsala was a truly enriching experience, both professionally and personally. Sweden, like other Scandinavian countries, has a serene and welcoming atmosphere, and Uppsala’s charm was undeniable. I particularly enjoyed peaceful walks along the Fyris river, soaking in the city’s tranquil beauty. The architecture of the University Main Building was breathtaking—its grandeur and historical significance made me wish our computer science department at the University of Warwick had a similarly inspiring space! And, of course, the famous Swedish meatballs were the perfect finishing touch to an already wonderful trip.

Reconnecting with consortium members was one of the highlights of the event. After a year of biweekly virtual meetings, it was fantastic to finally meet familiar voices in person and catch up with colleagues beyond their Bigpicture roles. I was fascinated to learn about the diverse side projects emerging within the consortium, such as developing foundation models for histology images, fine-tuning large language models for pathology tasks, and creating AI-based quality control tools for histology. These discussions highlighted the breadth of innovation happening alongside BigPicture’s core objectives.

On a professional level, the meeting was incredibly reassuring. Seeing the infrastructure for data uploads fully in place—complete with legal agreements, storage solutions, and metadata tools—was a major milestone. It was exciting to learn that some datasets are already accessible, with dedicated tools for searching the Bigpicture catalogue, requesting access, and downloading slides. However, as developers emphasised, these systems now need real-world testing, and the consortium is eager for collaborators to start utilizing them. While reaching the ambitious goal of three million uploaded slides remains a long-term challenge, the enthusiasm and motivation among partners were palpable. Key concerns, such as sustaining the project post-funding, are clearly being addressed by dedicated teams working hard to secure Bigpicture’s future.

Personally, I’m thrilled about the prospect of data sharing finally taking off. This will be instrumental for our large-scale, multi-centric project on automated Tumor-Infiltrating Lymphocyte (TIL) scoring in breast cancer, bringing us closer to impactful advancements in computational pathology. The Uppsala meeting not only reinforced the consortium’s collaborative spirit but also left me inspired and eager to contribute to the next phase of this transformative initiative.

Summary

Bigpicture’s Annual Meeting in Uppsala reinforced the consortium’s collaborative spirit, with participants contributing diverse perspectives to advance the project’s goals. From technical solutions for data uploads to strategies for AI model integration, the discussions and workshops laid the groundwork for the next phase of Bigpicture’s development.

The event concluded with a sense of accomplishment and renewed energy, as the consortium left Uppsala inspired to continue transforming digital pathology into a global resource for research and healthcare innovation.