The University of Szeged’s Department of Software Engineering is helping advance international training in clinical image interpretation through a project supported by a HUF 39 million grant. The project aims to develop an educational approach that strengthens professionals’ expertise and enhances the quality of image-based diagnosis – ultimately contributing to greater patient safety.
The “SZTE International Clinical Diagnostic Imaging Education – VM-Edu” project has secured HUF 39 million in funding from SZTE TTC Zrt. under the Proof-of-Concept grant program launched by Hungary’s National Research, Development and Innovation Office (NKFIH). Dr. István Juhos, a researcher at the University of Szeged’s Department of Software Engineering, explained what the project sets out to achieve.
For the past ten years, Juhos has led the international working group that helped the University of Szeged develop its own digital platform for examinations in clinical image interpretation. The platform supports international clinical quality assurance and university education. External quality assessment programs regularly evaluate professionals’ proficiency in image interpretation: participants receive a case to analyze and submit their findings for assessment. For the results to be comparable, a standardized examination requires every participant to receive exactly the same case – something that can be difficult to achieve when physical specimens are distributed individually.
This is where digital technology can make a difference. Tissue sections and blood smears can be digitized and made available online, giving everyone access to the same material. Images that are digital from the outset, such as those used in radiology, can be shared in the same way. However, organizations responsible for external quality assurance in clinical practice typically lack the resources needed to establish the necessary digital infrastructure for digitizing specimens, sharing large files, and collecting and processing structured examination data.
What was needed, Juhos explained, was a collaborative solution – software that would allow organizations to pool their resources and share not only images, but also expert analyses and, through them, their expertise. The Szeged team developed such a platform in collaboration with several external quality assessment organizations and their European umbrella organization, EQALM (European Organisation for External Quality Assurance in Laboratory Medicine).
Clinical image interpretation is facing growing pressures: rising caseloads, a shortage of specialists, an aging workforce, and a limited pipeline of new professionals. One potential way to ease some of these pressures is through artificial intelligence (AI), which can already assist with some image interpretation tasks. However, it cannot solve staffing shortages on its own or eliminate the need for continuing education. Its potential is also constrained by the availability of data needed to develop and evaluate AI systems reliably, particularly in rare cases, where high-quality, expert-verified data are scarce.
The University of Szeged’s project seeks to address these challenges through a “human-in-the-loop” approach – one that brings human expertise and AI together so that each can contribute to improving the other. In practice, expert assessments and structured examination data could support AI development and provide benchmarks for comparing AI systems, while AI-generated results could, in turn, contribute to the training and competency development of professionals. This interplay, however, does not diminish the role of specialists. In a high-stakes field such as clinical image interpretation, expert human oversight remains essential.
The examination platform previously developed at the University of Szeged supports both image sharing and the structured recording and assessment of participants’ responses. It has been successfully used in more than 15 countries for several years, albeit still on a pilot basis. Building on this experience, the researchers are now turning their attention to another area where the platform could make a difference: training in clinical image interpretation. Each examination generates a wealth of valuable data, including clinical images, expert reference assessments, and participants’ responses. In the longer term, the researchers hope to use these resources to build European databases supporting examinations, education, research, and AI applications, working in collaboration with professional partners and ensuring that the necessary usage rights are in place.
“We want to create a system that identifies strengths and gaps revealed by examination tasks, then offers targeted learning activities to help participants learn more effectively and retain knowledge longer – ultimately contributing to patient safety. This approach could, of course, also be used in university education. With appropriate methodology and data protection safeguards in place, the international initiative could also enable comparisons of aggregate results across countries, regions, and institutions. The goal is to help harmonize professional practices and improve the quality of image-based diagnosis across Europe,” Juhos said.
The international collaboration needed to pursue this vision is already taking shape. In 2025 and 2026, the EQALM Virtual Microscopy Interlaboratory Study brought together participants from 15 European countries, as well as Brazil and Mexico. Together, they completed several thousand case analyses – involving roughly 100,000 individual cell evaluations.
The new project takes this experience a step further. Its aim is to develop an educational approach that addresses an unmet need and to demonstrate its feasibility in an international setting. Rather than starting from scratch, the project builds on two strong foundations: EQALM’s international professional network and the University of Szeged’s existing digital examination platform for clinical image interpretation.
Original Hungarian article by Imre Vida-Szűcs
