Meet: Professor Anita Grigoriadis

Early detection of cancer significantly improves patient outcomes, allowing medical professionals to intervene much sooner. For Professor Anita Grigoriadis, the fight against cancer starts with identifying indicative disease patterns that could signal danger early on: tiny changes in molecules within a cell, and subtle shifts in the way cells arrange themselves in normal tissue. Her team has now received unprecedented levels of government funding to support this mission.

Anita Grigoriadis is Professor of Molecular and Digital Pathology at King’s College London, where she heads the Cancer Bioinformatics group in the School of Cancer & Pharmaceutical Sciences (Faculty of Life Sciences & Medicine) and is group lead of the Spatial Biology Network at Guy’s Cancer Centre. Her interdisciplinary team works on all aspects of AI-powered cancer research, including digital pathology, molecular analysis, spatial transcriptomics, and importantly, building the right kind of datasets to train AI models for early detection.

Building AI tools capable of recognising such patterns quickly enough can save patients’ lives. "My research is really about identifying patterns in DNA, RNA and histology that are associated with treatment response or treatment or disease progression, either at the primary tumour site or in nearby lymph nodes of breast cancer patients", she explains.

From wet lab bench to cancer bioinformatics

Anita’s route into AI-powered cancer care didn’t begin with computer science. She trained as a cancer biologist at the Ludwig Institute for Cancer Research, working on the molecular side of breast tumours. Then came a pause: five years focused on raising her children, combined with part-time work that nudged her, step by step, towards the computer lab.

It was exactly the moment bioinformatics was taking off. "In those five years I became a computer scientist", she recalls, swapping pipettes for Python. What began as a modest patient donation to support a single PhD student in her team grew, over two decades, into a large research programme on lymph-node patterns in breast cancer. That initial gift now underpins a growing team of researchers, with support from major funders, in close collaboration with Dr Dinis Caldao at the Francis Crick Institute.

Mobilising AI models in the fight against cancer requires thorough and comprehensive training data. Anita’s team has built the GRAPE repository, containing more than 1,500 digitised whole slide images of lymph nodes from nearly 200 breast cancer patients treated at Guy’s and St Thomas Foundation Trust and other hospitals across the globe. Next to GRAPE sits OASIS, a retrospective collection of nearly 1,500 whole slide images of normal breast tissue for the purpose of comparison, and built to improve cancer risk assessment in women with germline mutations.

Foundational work at this level lets other teams ask new questions. A lab in another country that has no archive of its own can now explore how immune cells cluster in lymph nodes or how normal breast tissue differs in high-risk groups, because Anita’s team has already done the hard work of curating, digitising and sharing. As Anita puts it, there is a deliberate shift from individual projects to 'shared assets' that can keep generating insights long after the original papers in a research project have been published.

Patterns at national scale

If GRAPE and OASIS are specialist libraries, Anita’s current flagship project is a whole new library network that connects them and opens the doors to everyone. PharosAI is a national platform that will allow researchers and companies to build and test AI models for cancer diagnosis, treatment selection and drug discovery on decades of NHS and biobank data, without shipping sensitive patient data around. In February 2025, the UK government’s Research Ventures Catalyst programme awarded PharosAI an unprecedented £18.9 million in public funding, matched by £24.7 million in co-investment from partners including charities, NHS trusts, AI developers, pharmaceutical and technology companies.

PharosAI is jointly led by King’s College London, Queen Mary University of London, Guy’s and St Thomas’ NHS Foundation Trust and Barts Health NHS Trust, and is part of the UK Health Data Research Alliance. Visiting ministers have highlighted it as one of three exemplar ventures showing how AI can transform cancer care. The idea is to "democratise cancer AI by refining decades of oncology data into deep multimodal datasets" so that models can be developed, compared and validated in one place. Rather than building yet another silo, it offers what Anita calls an "ecosystem to accelerate the path to AI-powered precision medicine", advancing knowledge through collaboration.

In this together

Anita pushes back strongly against any narrative that centres on a lone PI.

She emphasises that PharosAI only became possible because four large institutions with very different cultures chose to align around a shared goal.

The project has brought many researchers together. "To be honest, it has been absolutely amazing. Over the period of even putting the grant application together, I have never worked with so many people across King’s than in these last two years, and I have been at King’s since 2008", she laughs. Research managers, finance and contracts staff, IT specialists, data governance experts, clinicians and patient representatives all had to be in the room. These are, as she puts it, "the people that sit in the back and actually make the engine work", and they played a part in helping to connect dots, suggest contributors and solve problems as the project began taking shape.

The King’s Institute for Artificial Intelligence was one of the key connectors in this process, helping Anita to find collaborators beyond the usual cancer and bioinformatics circles and making sure the project could tap into the right expertise in law, ethics, risk management and public engagement in a timely way. For Anita, this is what a modern AI institute should do: actively help turn ambitious ideas into operational platforms that can change clinical practice, and ultimately, people’s lives.

From one to many

As PharosAI accelerates, the day-to-day work of the Cancer Bioinformatics group continues to push the boundaries of what is possible in focused ways. A recent preprint on normal breast tissue classifiers shows how convolutional neural networks trained on OASIS can accurately classify large scale tissue compartments, providing a new tool for understanding risk in BRCA mutation carriers and available for others to build on. Earlier work highlighted by Breast Cancer Now uses patterns in immune cells in lymph nodes to predict whether aggressive triple negative breast cancer is likely to spread, giving clinicians another piece of evidence when planning treatment.

For all these important publications, large numbers and national platforms, Anita’s narrative keeps circling back to the wellbeing of individual patients. When asked about the origins of her lymph node work, she does not start with a grant reference, but with the story of a single woman treated at Guy’s who decided to donate money for Anita to get a PhD student onboard. That gift, she explains, "enabled me to employ the first person to really explore lymph nodes in depth", and over time it grew into the programme that now underpins GRAPE and related projects.

And when asked if she still thinks about that first donor, she smiles and responds they are still in touch to this day and meet for lunch. It is a neat image of what Anita’s work attempts to do at every scale: connect the patterns in data and tissue back to the individual receiving treatment, whether that is a single patient or, through PharosAI, thousands of people whose data now help to shape better cancer care. Anita’s work shows how, in cancer research, far from pushing people apart, AI can help clinicians, scientists and patients come together and amplify their combined efforts to really make a difference to patient outcomes.