TU/e unveils open-source AI model to enhance medical diagnosis

Researchers at Eindhoven University of Technology (TU/e) have developed a powerful AI model that can rapidly analyze medical imaging data, enabling earlier disease detection, such as cancer. Trained on over 250,000 computed tomography (CT) scans, the model provides a valuable tool to augment doctors' expertise.

The AI model can detect tumors, forecast disease progression, and pinpoint patterns within medical imaging data. According to the researchers, it accurately distinguishes between healthy kidney tissue and tumors, mirroring the diagnostic capabilities of a skilled physician. However, the model is designed to enhance data analysis rather than replace the crucial role of doctors, who remain essential for interpreting the AI's findings.

Associate Professor Fons van der Sommen, who leads the research team, emphasizes the collaborative potential of this technology, stating, "We provide the stem from which others can grow their own medical AI models. This lowers the threshold for innovation and collaboration in healthcare".

Open source AI medical diagnosis

TU/e's AI model will be made available to the broader medical community. This decision will enable hospitals, research institutions, and companies to develop customized versions tailored to their specific needs. Van der Sommen explains the rationale behind this open-source approach: "Previously, a new AI model was viewed as a goose that lays golden eggs – something you wouldn't want to share. But this model can lay so many golden eggs that we cannot possibly handle them all ourselves. By sharing it, everyone can move forward".

The new supercomputer contribution

The development of this AI model was made possible by the formidable computing power of SPIKE-1, TU/e's new supercomputer. Put into service in 2024, SPIKE-1 consists of four NVIDIA DGX B200 systems, each equipped with eight powerful Blackwell GPUs.

This project marked SPIKE-1's first practical deployment, with researchers actively involved in configuring both the hardware and software to ensure optimal performance. SPIKE-1 is hosted in a sustainable data center located in Finland. TU/e was among the first universities to have access to the NVIDIA DGX B200 system.

TU/e researchers plan to continue publishing their findings to broaden awareness and enhance the international visibility of their research and the AI model. Today, November 13, 2025, the AI Summit Brainport event in Eindhoven will showcase SPIKE-1 and related AI research, featuring several TU/e researchers as speakers. Furthermore, TU/e is actively participating in the OpenEuroLLM project, a collaborative effort involving 20 leading European research institutions and companies focused on developing next-generation open-source language models. This project aims to advance European AI capabilities through transparent AI applications.