SciTransfer
DeepHealth · Project

AI Tools That Help Hospitals Diagnose Diseases Faster From Medical Images

healthPilotedTRL 7
deephealth-project.eu

Imagine you need a doctor to look at thousands of medical scans — X-rays, tissue slides, brain images — to spot signs of disease. That takes enormous time and expertise. DeepHealth built ready-to-use AI software libraries that let hospitals and medical software companies plug deep learning into their existing systems, so computers can help doctors read those images far more accurately and quickly. They tested it across 14 real medical scenarios including dementia, migraine, and digital pathology, and plugged it into 7 existing medical platforms including commercial ones from Philips and Thales.

By the numbers
14
medical use cases validated
7
existing biomedical platforms integrated
25
consortium partners
9
European countries involved
31
project deliverables produced
4
SME partners in the consortium
The business problem

What needed solving

Hospitals and medical software companies need AI that can analyze complex medical images — pathology slides, brain scans, clinical records — but building deep learning systems from scratch is expensive, slow, and requires rare HPC expertise. Most healthcare organizations lack the computing infrastructure and data science teams to develop and deploy these models at scale.

The solution

What was built

DeepHealth built two core software libraries: the European Distributed Deep Learning Library (EDDLL) and the European Computer Vision Library (ECVL), plus an HPC runtime system. These were validated across 14 medical use cases and integrated into 7 existing biomedical platforms including commercial systems from Philips and Thales.

Audience

Who needs this

Medical imaging software companies looking to add AI capabilitiesHospital IT departments deploying AI-assisted diagnosticsDigital pathology companies automating tissue analysisClinical decision support system vendorsHealth-tech startups building AI diagnostic tools
Business applications

Who can put this to work

Medical imaging software
enterprise
Target: Companies developing clinical decision support or diagnostic imaging platforms

If you are a medical software company struggling to add AI capabilities to your diagnostic platform — this project developed two open libraries (EDDLL for deep learning, ECVL for computer vision) that were validated in 14 medical use cases and integrated into 7 existing platforms including the Philips Clinical Decision Support System. Instead of building AI from scratch, you can plug these libraries into your product.

Digital pathology
mid-size
Target: Pathology labs and companies digitizing tissue analysis

If you are a pathology lab dealing with growing caseloads and slow manual slide review — this project built and tested AI models for digital pathology that run on hardware ranging from hospital servers to supercomputing centers. The tools were validated in CRS4's Digital Pathology platform across real clinical data.

Hospital IT and healthcare providers
enterprise
Target: Hospitals and health networks investing in AI-assisted diagnosis

If you are a hospital IT department trying to deploy AI diagnostics but struggling with computing requirements — DeepHealth built a solution compatible with HPC infrastructure ranging from supercomputing centers to hospital-grade servers. The 14 validated use cases cover areas like dementia, migraine, and depression, with models ready for integration.

Frequently asked

Quick answers

What would it cost to adopt these AI libraries?

The EDDLL and ECVL libraries were developed as European open-source tools. Licensing terms would need to be confirmed with the consortium, but as an EU-funded Innovation Action the intent is broad availability. Integration costs would depend on your existing platform and computing setup.

Can this scale to handle a full hospital's diagnostic workload?

Yes — the entire project was designed to scale. The solution works on HPC infrastructure ranging from supercomputing centers down to hospital-level servers. It was validated across 14 use cases with large and complex biomedical datasets, and the project specifically tracks time-to-model-in-production as its success metric.

Who owns the intellectual property?

The consortium of 25 partners across 9 countries jointly developed the technology. IP and licensing arrangements would need to be discussed with the coordinator (NTT DATA Spain). Given that commercial partners like Philips and Thales integrated the tools into their platforms, commercial licensing paths likely exist.

Is this certified for clinical use?

Based on available project data, the tools were validated in 14 medical use cases and integrated into 7 existing biomedical platforms. However, regulatory certification (CE marking, FDA clearance) for specific clinical applications would need to be confirmed with the consortium partners.

How long would integration take?

The libraries were specifically designed to plug into existing biomedical software platforms — 7 were successfully integrated during the project. The project tracked time-to-model-in-production as a key metric, suggesting the tools are optimized for fast deployment. Timeline depends on your current infrastructure.

Does this work with our existing medical software?

The project validated integration with 7 different biomedical platforms, both commercial (Philips Clinical Decision Support, Thales PIAF) and research-oriented (CEA ExpressIF, CRS4 Digital Pathology). The libraries were built to be flexible and compatible with heterogeneous computing environments.

Consortium

Who built it

The DeepHealth consortium brings together 25 partners from 9 European countries — a strong mix of 9 industry players (36%), 7 research organizations, 6 universities, and 3 other entities including 4 SMEs. What matters for business adoption: major commercial players like Philips and Thales already integrated the results into their platforms, meaning the technology has been stress-tested against real product requirements, not just academic benchmarks. The coordinator is NTT DATA Spain, a global IT services company, which signals a business-oriented project leadership rather than a purely academic effort. The presence of health organizations alongside tech companies means the tools were validated by the people who actually use diagnostic software daily.

How to reach the team

NTT DATA Spain coordinated this project. SciTransfer can facilitate an introduction to discuss licensing, integration support, or access to the trained models and libraries.

Order a report on this consortiumSee NTT DATA SPAIN, SL profile →
Next steps

Talk to the team behind this work.

Want to explore how DeepHealth's AI libraries could accelerate your medical imaging product? SciTransfer can connect you directly with the right consortium partner for your use case — whether that's the deep learning library team, a clinical validation partner, or the commercial integration leads.

Order a report on this topicMore in Health & Biomedical
More in Health & Biomedical
See all Health & Biomedical projects →
Automated Allergy Test Covering All Allergens in One Run, Ready to Ship
AllergyExplorer
Open →
Lab-Sized 3D Cell Microscope Replaces 12-Month Waits at Giant Synchrotron Facilities
LICENT
Open →
Blood Test Platform That Catches Rare Diseases Early Using AI-Powered Diagnostics
METAdiag
Open →