FFplus Interview with Panagiotis Papadimitroulas, Bioemtech

 

 

In this interview, we speak with Panagiotis Papadimitroulas, Associate Professor of Biomedical Informatics at the University of Thessaly and co-founder and software advisor at Bioemtech. Bioemtech is a Greek specialised organisation that provides services to assist pharmaceutical and biotechnology companies. 

Panagiotis explained how Bioemtech is accelerating preclinical research, towards clinical translation for promising drugs, through our high-quality services and products. He also shares insights from their successful participation in the FF4EuroHPC Business Experiment and the newest FFplus Innovation Study. 

 

Panagiotis Papadimitroulas, Associate Professor of Biomedical Informatics at the University of Thessaly 

and co-founder and software advisor at Bioemtech (Copyrights: Bioemtech)

 

 

Panagiotis, thank you for being with us today. To start, could you briefly introduce Bioemtech? 

It is a great pleasure for me to have the opportunity to introduce BIOEMTECH and our activities within the EuroHPC community.

BIOEMTECH is a Greek SME operating in the biotechnology sector, with a focus on molecular imaging, dosimetry, and biomedical engineering. The company is structured into three main units, each dedicated to a distinct area of activity: Design and construction of screening imaging systems – the "eyes" series.

We have developed and launched three imaging devices currently available on the market, with nearly 40 installations worldwide. These include: the g-eye and b-eye, for imaging SPECT and PET radioisotopes respectively, and the f-eye, for optical imaging.

BIOEMTECH Laboratories provides a unique infrastructure for preclinical studies, ranging from in vitro experiments to in vivo testing. Our "one-stop-shop" laboratory platform offers comprehensive services to the pharmaceutical industry, including:

·       in vitro assays,

·       radiochemistry controls,

·       animal model development and protocol design,

·       imaging and quantification,

·       and ex vivo analysis.

Software and Simulation Unit, which is highly active in European R&D projects. We have outstanding expertise in Monte Carlo simulations for medical imaging and dosimetry. Our team includes experienced engineers working on AI-based prediction models and Decision Support Systems for both clinical and preclinical applications.

More recently, we have expanded our efforts to include Gen-AI-based tools, explainable AI, and automated image processing and analysis.

BIOEMTECH is an interdisciplinary company composed of 50 experts from a wide range of fields, including mechanical and electrical engineering, software development, radiochemistry, biology, physics, veterinary science, and medical technology.

 

 

From your perspective, what are the most transformative trends currently shaping the future of the pharmaceutical industry in the preclinical research phase? 

The pharmaceutical industry is undergoing a major transformation in how preclinical research is conducted. Several key trends are driving innovation, increasing efficiency, and enhancing the predictive value of preclinical studies. From my perspective, the most impactful developments include:

The preclinical research landscape is being reshaped by technologies and practices that improve translational value while upholding high ethical standards. One significant trend is the integration of advanced imaging and in vivo techniques, which allow real-time, non-invasive monitoring of drug distribution and efficacy. This not only enhances data quality but also contributes meaningfully to the reduction of animal use, in line with the 3Rs principles (Replacement, Reduction, and Refinement).

Artificial intelligence (AI) and machine learning are also revolutionising data analysis—automating image processing, uncovering complex patterns, and accelerating decision-making. In particular, Generative AI models for dosimetry prediction are paving the way for personalised imaging and therapeutic protocols. These tools can shorten study timelines and reduce the need for animal sacrifice by more precisely predicting optimal dosing.

In parallel, the adoption of personalised and translational models—such as organoids and humanised mouse models—is significantly improving the predictive accuracy of preclinical research.

Finally, increasing regulatory and ethical pressures are compelling the industry to adopt more humane and efficient methods. Companies like BIOEMTECH are at the forefront of this shift, offering benchtop, high-resolution imaging solutions that make advanced technologies more accessible and impactful for preclinical research.

 

BIOEMTECH develops advanced imaging systems for preclinical research. How are you integrating AI into your imaging workflows, and what impact does it have on the speed, accuracy, or interpretation of biomedical data?

 At BIOEMTECH, AI is becoming a core enabler across our imaging workflows, enhancing both the efficiency and precision of data acquisition and analysis.

One key integration already in place is the use of synthetically generated X-ray images for anatomical mapping within our eyes-series imaging systems. This innovation enables users to localise radiotracer signals without the need for physical X-ray hardware, significantly streamlining the workflow. The feature has been well-received by our clients, particularly for its clarity, speed, and ease of use.

In parallel, we are developing automated, AI-based image analysis tools for our in-house processes, with a focus on image registration and segmentation. These tools are designed to reduce manual intervention and variability, making quantitative imaging more robust, reproducible, and accessible to users with varying levels of expertise.

We are also advancing a Generative AI-powered feature for 3D dosimetry prediction at the organ level, using 2D imaging data acquired from our systems. Thanks to the Fortissimo FFplus Innovation Study (DosimetrEYE) and the use of HPC resources, this tool will enable real-time, organ-level dosimetry during image acquisition. The goal is to accelerate preclinical protocols, support personalised treatment planning, and reduce animal use by minimising the need for multiple scans or invasive procedures.

Together, these AI-driven innovations reflect our commitment to smarter, faster, and more ethical preclinical imaging.

 

Let's now talk a bit about your experience with the participation in FORTISSIMO actions, such as the FF4EuroHPC experiment. Could you briefly introduce your use case and the innovative solution you developed?

At BIOEMTECH, our journey with FF4EuroHPC marked our first step into leveraging HPC resources to address a critical clinical challenge: achieving personalised pediatric dosimetry in nuclear medicine.

In current clinical practice, dose estimations are typically generalised, with children of varying ages and anatomies often receiving the same radiopharmaceutical dose based on standard protocols. However, given their increased radiosensitivity, personalised dosimetry is especially crucial for pediatric patients. While Monte Carlo (MC) simulations offer gold-standard accuracy, they are computationally intensive and therefore impractical for routine clinical use.

Through access to HPC resources provided by FF4EuroHPC, we built a large-scale MC simulation database for commonly used radiopharmaceuticals, using a diverse digital phantom library representing pediatric populations aged 3 to 15 years. This rich dataset enabled us to train an AI-based prediction model capable of estimating absorbed organ doses without the need to run new simulations for each patient.

In collaboration with IknowHow, a Greek SME specialising in clinical software and PACS systems, we developed a graphical user interface (GUI)-based clinical decision support system (DSS). As a pilot, the PEDIDOSE DSS was integrated into the EVORAD software platform, allowing clinicians to input key anatomical parameters and receive real-time, organ-level dose estimates. This empowers healthcare professionals to make more informed decisions and, when necessary, adjust administered doses—supporting safer, more personalised care for pediatric patients.

 

You were also successful with the application for the FFplus Open Call for Innovation Studies. How does this case differ from the one you developed under FF4EuroHPC?

Following the successful development of the PEDIDOSE clinical decision support system during the FF4EuroHPC project, we identified an opportunity to extend the concept of AI-driven personalised dosimetry into the preclinical imaging space, leveraging our own imaging technology.

The FFplus Innovation Study allowed us to advance our HPC and AI capabilities by developing a real-time dose prediction feature for our EYES-series preclinical imaging systems. Unlike FF4, which focused on clinical applications for pediatric patients, FFplus was dedicated to building a fully embedded dosimetry solution for research use—based entirely on 2D images generated during routine imaging acquisitions.

This was a technically ambitious project. To build a reliable training dataset, we collected:

·       3D SPECT/CT images

·       2D SPECT images from our EYES systems

We then used the 3D data to perform Monte Carlo simulations for dose mapping across a wide range of biodistributions.

Thanks to access to HPC resources, we were able to generate a large number of high-accuracy 3D dose maps with minimal statistical uncertainty. These maps were then used to train a Generative AI model capable of predicting organ-level 3D dose distributions directly from 2D images in real time.

In this project, we collaborated with ALETHIA, a Polish SME specialising in explainable AI. Their contribution ensured that our model not only produces accurate predictions but also delivers interpretable outputs, enhancing trust and usability within the preclinical research community. Although the project is still under development, we are excited by the very promising initial results and confident that this solution will enable more personalised, efficient, and ethical preclinical studies using our systems.

 

As a newcomer to HPC, could you share some insights into that experience? How valuable was this opportunity for your company in terms of innovation, development, and collaboration with other organisations?

Entering the world of HPC through the FF4EuroHPC and FFplus projects was a transformative experience for BIOEMTECH. Initially, as a company with no prior exposure to HPC, we viewed it as a complex and resource-intensive field. However, the structured support we received—both technical and collaborative (especially from GRNET, the Greek National Competence Center, NCC)—made our entry smooth and productive.

From an innovation perspective, HPC unlocked possibilities that were simply not feasible with standard computing resources. In both projects, we relied on large-scale Monte Carlo simulations to generate detailed and statistically robust dose maps—an essential component in developing our AI-based dosimetry tools. Without HPC, such simulations would have taken months or been computationally prohibitive.

From a development standpoint, access to HPC resources allowed us to move beyond the proof-of-concept stage and create production-level datasets and AI models, dramatically accelerating our R&D timeline. It enabled us to test at scale, iterate quickly, and build tools with real clinical and preclinical impact.

Equally important was the collaborative aspect. Through FF4EuroHPC, we partnered with IknowHow, a clinical software provider, and through FFplus, we worked with ALETHIA, a company specialising in explainable AI. These partnerships brought complementary expertise and helped extend the reach and usability of our solutions—clinically through user-friendly GUIs, and ethically through interpretable AI models. Our collaboration with all partners was efficient and continued seamlessly throughout the projects.

Overall, HPC access not only enhanced our technical capabilities—it expanded our innovation mindset and opened doors to future cross-disciplinary collaborations. It served as a real catalyst for growth and differentiation in the competitive field of biomedical imaging and personalised dosimetry.

 

In your HPC journey, you aimed to use precise dosimetry software tools such as “PEDIDOSE” and “DOSIMETREYE”. What business benefits did this bring, and how has the experience continued to support your company after the project ended?

Developing PEDIDOSE and DOSIMETREYE through our HPC-enabled projects has had a significant and lasting impact on BIOEMTECH—both in terms of business growth and technological advancement.

From a business perspective, these tools have allowed us to expand our portfolio with AI-powered, high-precision dosimetry solutions that address key market needs in both clinical and preclinical imaging. PEDIDOSE has opened new opportunities in the clinical decision support space, particularly in paediatric nuclear medicine, where personalisation is critical. In contrast, DOSIMETREYE differentiates our EYES-series imaging systems by adding unique value through real-time dose prediction capabilities—an entirely innovative feature in compact preclinical devices.

These innovations have enhanced our competitiveness, attracted new clients, and fostered collaborations with clinical software providers, AI developers, and research institutions. They also position us strongly in markets where precision, efficiency, and ethical compliance (through the reduction of animal use) are becoming standard expectations.

In addition, we have upskilled our team in areas such as parallel computing, simulation scaling, and AI model training, and have established long-term partnerships—most notably with GRNET, our National Competence Centre (NCC), whose ongoing support remains pivotal. The knowledge gained, access to infrastructure, and integration into a broader innovation ecosystem have become strategic assets as we explore future AI–HPC integrations in imaging and dosimetry.

In summary, HPC didn’t just help us build two innovative tools—it elevated our R&D capabilities, expanded our business horizons, and continues to support our evolution as a technology-driven biomedical company.

 

Let’s wrap up the interview with some final thoughts. What advice would you give to other SMEs that haven’t yet explored the use of HPC and AI in their business?

My advice to other SMEs is simple: don’t hesitate to try HPC and AI in your daily operations—embrace them as tools that can unlock innovation far beyond your current limits.

When we started, we had no prior expertise in HPC. However, through our participation in EU initiatives such as FORTISSIMO and EUROHPC, and with the invaluable guidance of partners like GRNET, we quickly realised that these technologies are not reserved for large corporations or academic institutions. They are accessible, scalable, and extremely valuable—especially for SMEs looking to stay competitive and innovative.

HPC, combined with emerging AI trends, can boost industrial automation, enhance decision-making, solve complex problems, and even enable the development of entirely new features for customers and global markets.

More importantly, getting involved in this ecosystem opens doors to collaborations, funding opportunities, and knowledge exchange—all of which are vital for growth, particularly in high-tech fields. I would strongly recommend such initiatives to other SMEs, with one important caveat: clearly identify your computing needs in advance. It’s essential to understand what you require before requesting HPC resources in order to make the most of their potential.

So, my message is this: don’t wait for the “perfect” moment or full in-house expertise to start using HPC and AI. Begin with a well-defined use case, leverage the support networks available, and stay open to learning. The return—in terms of innovation, capability, and market differentiation—is well worth the effort.

 

Panagiotis, thank you so much for this insightful interview. It was a pleasure having you with us! Wishing you all the best in your future endeavours.

Learn more about the Success Story A Pediatric Simulated Dosimetry Platform for Clinical Use here.