Download the presentation slides.
Webinar Q&A:
Question |
Answer |
Will this initiative collaborate with HE-Cluster Health for the 2 calls on GenAI (in case they work on cancer)? |
We encourage the submitting projects to the pathfinder challenge to seek collaborations with HE-cluster call. There is no obligation or structured collaboration in place. |
Are all cancer types referred to going to be funded under this PATHFINDER call? |
Yes. Please look at the guidelines too. |
Is there a way we can know (and support) connections of projects? e.g.: less frequent tumour types may be less covered (unless we pool some rare cancer types)… data augmentation projects may combine well with other projects but may be less clinical… |
There is no possibility to know upfront. Only when we have all the submission we will see what we have and create a portfolio. |
Is it possible to apply new technologies acquired for the studies? I got a Brain Computer Interface to work directly with neurological data. |
If your use of the BCI can enhance generative AI agents in a way aligned with the challenge (e.g., for personalized medicine, adaptive decision support, or human-AI alignment), then yes, it fits. |
Is it possible to join EUCAIM this year? |
Joining the EUCAIM consortium is not possible anymore. However, sharing data and requesting data is possible as a ‘normal’ data holder or data users. |
How could we access? I applied to EUCAIM, but I have not been accepted yet to access datasets. |
All the proposals to access datasets will be answered. We are finalizing the Access Committee pathway. Thanks for your willingness to collaborate. |
From the previous and today’s sessions, it sounds like the proposals that have the diagnostics component have a higher chance of being funded. Our project currently focuses on personalized treatment prediction and prognosis. Should we add the diagnostics part in the project to increase the chance of being funded? |
The clinical focus of the challenge can be either diagnostic or therapeutic, and proposals that address only one of the two — including just treatment prediction and prognosis — are fully within scope. That said, proposals that integrate both diagnostics and therapeutics in a coherent way may offer a more comprehensive impact and could be particularly compelling. However, we strongly advise not to add a diagnostic component artificially — it should only be included if it brings genuine scientific and clinical value to the project. |
Is it necessary (or recommended) to reserve additional personnel (e.g., a (partial) PhD student) for joint activities within the portfolio, beyond the mandatory allocation of person-months for portfolio/project management OR is dedicating only management time sufficient for the application? |
Each project will be requested to add a WP on portfolio activities for 10 man/months. if you do not do it at submission, you will still be asked to do it at the Grant Agreement signature. |
Does this project include the use of liquid biopsy for early cancer detection followed by imaging of cancer lesion ? |
The EUCAIM project focuses on diagnostic imaging at the moment. Expanding to other imaging modalities is something we’re evaluating and hope to implement in the future. |
Are startups automatically always falling under “commercial use” for the data? |
Answered live: 1:04:00 |
Following the requirements in EUCAIM, is that possible to access the EUCAIM data for a PhD student in Informatics? The PhD project is funded, and supervised. |
Yes, The Atlas is prepared for that. He/she has to search the dataset and ask for data access after meeting the requirements. |
What is the importance you give to the involvement of technology transfer offices of the involved organizations in these projects? They are usually the ones who drive the commercialization and impact work packages. |
It is not mandatory but it is positively evaluated. |
Could I provide and work with brain waves data if this is extracted from my own brain (BCI)? |
You would need to ask your ethical board. |
Multi-modal, is that meaning “different image modality” or it is also including EHR or OMICS? |
Answered live: 1:08:34 |
Our project would focus on multimodal data, but the main/only imaging modality would be digital pathology. If I’m correct, EUCAIM does not include digital pathology and so our use of it would be limited. I assume this is severely limiting the acceptability of our project idea? |
Answered live: 1:11:39 |
Are these tools “open source” ?If they are not perfect for our needs, can they be tuned for our needs? Or should be used as-is? |
This depends on the tool and the conditions set by the tool/software provider. Some EUCAIM tools are open-source (which any user can employ as it is, or can fine-tune), while others are licensed. However, as a general rule, in all cases, tool and software providers are open to discussions regarding collaboration and improvement of tools and solutions. |
For the accepted proposal, what will be the tier level of access to the data? I assume Tier-2 – full access? |
Answered live: 1:13:59 |
I have a question regarding the negotiation request. We need to provide a document titled “Objectives of the application, study design and work plan”: does the work plan refer to the timeline of our particular project? If so, should we consider the timeline as starting from our entry into EUCAIM (January 2025), or from the beginning of the negotiation phase (so now, June 2025)? |
The timeline should be defined in the future, considering from the data access is granted onwards. |
Again on the negotiation process. Regarding the document “Sources of funding” document: our project is 50% funded by our institution and 50% by EUCAIM. Should we explicitly mention this breakdown in the document? |
We have not defined this level of detail so far, but we can come back to you if there is any information missing (the negotiation application can take several iterations until the information is complete). |
I am gathering terabytes of BCI data to study advancements in Strokes and rehab. I am a medical doctor, computer scientist and data analyst in Ireland. How could I validate and provide this to EUCAIM? |
Answered live: 1:20:07 |
Can a US university be a partner on this grant? |
Answered live: 1:05:20 |
Can a UK university lead an application? |
Answered live: 1:19:26 |
Taking into account the specific Pathfinder call, how do you envisage the involvement of EUCAIM in a specific proposal? |
Answered live: 1:06:14 |
To what extent are patient trajectories included in the dataset (like comorbidities, concurrent drug use, outcomes?) |
It depends on the dataset. Some datasets have additional data from the one on the CDM. You should identify the datasets and get in contact with the dataset contact point (listed in the catalogue) to request the specific variables collected. |
How can we leverage the available data from EUCAIM to align with the current Pathfinder Challenge call? Is there a specific direction or approach we should consider? |
Answered live: 1:18:22 |
What happens in the event that EUCAIM data are not available or accessible to a Pathfinder proposal after it has been approved for funding? |
EUCAIM is aiming to constitute an EDIC during the next year, which will define the sustainability paths to ensure the preservation of the data. If you request access and you specify the timeline, we will commit to have it during this period. |
Can a single institution submit a grant on a smaller scale than a consortium if they e.g. have very specific experience in image synthesis for treatment selection or are you prioritising consortia? |
Answered live: 1:10:18 |
Do you envisage funding one project for each cancer e.g. one for lung cancer etc or would you end up funding multiple projects for the same cancer? |
Answered live: 1:17:09 |
What is the expected duration of an EIC Pathfinder project? |
Answered live: 1:15:26 |
On the other hand, is there any upper limit for the number of partners in the consortium that you envision? |
There’s no formal upper limit. Consortia should be reasonable and realistic. |
in the challenge, we are also addressing the challenges for integrating Multimodal Multidimensional Data. What I can see you are also evolving inside of EUCAIM. What would be key recommendations to avoid duplicated effort? Is there any clear view around the main gaps we should be more focused on? |
This is a recent activity we are developing to align the infrastructures, not for the integration of multimodal data. It will facilitate discovering data that are in different repositories due to come from different modalities. We do not see a major risk of overlap. |
From the perspective of incorporating EUCAIM in the project we would need to navigate deeper in the structure, the gaps and details of the implementation. Is there any specific link to get this info or should we schedule a call with support team? Is there any specific email or person we should reach out as next step? |
Sure! You can make a request in our dashboard’s helpdesk or explore our website. We have public documents with the architecture and user guides. |
It was mentioned that for every €1billion only about 11.5 products were launched. Is this funding directly to SME or are you also including all the funding, including academia? |
SME, companies and investor funding |
I would like to ask if you can fund prospective studies that involve scanning new cancer patients (not funding the scanner but funding the scans) or are we encouraged to use EUCAIM or retrospective data? |
Answered live: 1:25:28 |
Did I understand correctly that you also provide GPU computer capabilities? We have figured out that time-aware multimodal transformers requires at least H100 GPUs for mammography due to the high imaging resolution. Is this kind of GPU chips available? |
Answered live: 1:22:54 |
Do we have to use generative AI techniques? Can we not use traditional supervised learning techniques with CNNs and methods using vision transformers? Please can you define what you mean by generative AI in this context as you did for autonomous agents from the NVIDIA website? |
Answered live: 1:25:54 |
🗓️ June 10, 2025 | 🕘 09:00 – 10:30 CEST | 📍 Online
Are you applying to the EIC Pathfinder Challenge on Generative AI for Cancer Diagnosis and Treatment?
Join EUCAIM and Federica Zanca (EIC Programme Manager for Medical Imaging & AI in Healthcare, European Commission) for a deep dive into how the Cancer Image Europe platform can empower innovators building multimodal AI tools.
This 90-minute session will:
- Introduce the EUCAIM infrastructure and vision
- Showcase EUCAIM’s tools for federated learning, benchmarking, and data governance
- Explore how imaging can anchor multimodal AI model design
- Present concrete use-cases and show how to engage with EUCAIM during proposal preparation
- Offer a 45-minute interactive Q&A and discussion with the community
This is your chance to discover how EUCAIM can support the next generation of generative AI tools for cancer diagnosis and treatment, and how you can include this in your proposal for the upcoming EIC Pathfinder Challenge call.
🔗 Register here:
https://us02web.zoom.us/webinar/register/WN_e0pkOJ6LQGqxTq71O_7MzQ#/registration
📘 Webinar Agenda:
- Introduction to the EIC Pathfinder Challenge (7 min) – F. Zanca
-
- Scope and goals of the 2025 Challenge on Generative AI for Cancer
-
- Overview of EUCAIM and the Cancer Image Europe Platform (12 min) – L. Marti-Bonmatí
-
- Vision of EUCAIM under the European Cancer Imaging Initiative
-
- Federated infrastructure, clinical partners, and data modalities covered
-
- FAIR principles, data anonymization, and legal/ethical frameworks
-
- What EUCAIM Offers to Early-Stage Innovators applying for the pathfinder (12 min) – I. Blanquer
-
- Use of EUCAIM as a cancer imaging data reference platform
-
- Tools for annotation, federated training, and metadata curation
-
- How imaging can anchor multimodal model design (e.g., radiology + pathology)
-
- EUCAIM’s Broader Contribution to Multimodal AI Research (12 min) – L. Cerdá-Alberich
-
- Insights into data interoperability and architecture
-
- Contribution to benchmarking efforts required by the Pathfinder call
-
- Synergies with genomics/pathology/molecular datasets in federated environments
-
- Use Cases & Engagement Opportunities for Innovators (12 min) – A. Alberich-Bayarri
-
- Concrete scenarios where EUCAIM can support early experimentation
-
- How to engage with the infrastructure during and after project design
-
- Discussion & Q&A (35 min)
-
- Addressing participant questions
- Exploring alignment, feasibility, and collaboration opportunities
-
Speakers:
- Federica Zanca – 𝘌𝘐𝘊 𝘗𝘳𝘰𝘨𝘳𝘢𝘮𝘮𝘦 𝘔𝘢𝘯𝘢𝘨𝘦𝘳, 𝘌𝘶𝘳𝘰𝘱𝘦𝘢𝘯 𝘊𝘰𝘮𝘮𝘪𝘴𝘴𝘪𝘰𝘯
- Luis Martí-Bonmati – 𝘌𝘜𝘊𝘈𝘐𝘔 𝘚𝘤𝘪𝘦𝘯𝘵𝘪𝘧𝘪𝘤 𝘊𝘰𝘰𝘳𝘥𝘪𝘯𝘢𝘵𝘰𝘳, 𝘏𝘜𝘓𝘈𝘍𝘌
- Leonor Cerdá-Alberich – 𝘌𝘜𝘊𝘈𝘐𝘔 𝘈𝘐 𝘳𝘦𝘴𝘦𝘢𝘳𝘤𝘩𝘦𝘳, 𝘏𝘜𝘓𝘈𝘍𝘌
- Ignacio Blanquer – 𝘌𝘜𝘊𝘈𝘐𝘔 𝘛𝘦𝘤𝘩𝘯𝘪𝘤𝘢𝘭 𝘊𝘰𝘰𝘳𝘥𝘪𝘯𝘢𝘵𝘰𝘳, 𝘜𝘗𝘝
- Angel Alberich Bayarri – 𝘍𝘰𝘶𝘯𝘥𝘦𝘳 𝘢𝘯𝘥 𝘊𝘌𝘖 𝘘𝘶𝘪𝘣𝘪𝘮