Royal Philips has entered into a collaboration with NVIDIA to develop a foundational AI model for MR, with the goal of enhancing image quality, accelerating scan times, and streamlining diagnostic workflows.

The initiative will leverage NVIDIA’s AI computing platform to power a large-scale deep learning model trained on diverse MR datasets. The model is intended to support a range of clinical applications by enabling automation and interactivity in image acquisition and interpretation.

According to Philips, the AI system will allow for zero-click scan planning across anatomical regions, aiming to reduce workflow complexity and boost throughput. Radiologists will also have access to tools for real-time image enhancement, including denoising and super-resolution, and the ability to preview and adjust scan parameters prior to acquisition.
“The benefits for patients and healthcare providers could be enormous,” said Dr. Ioannis Panagiotelis, business leader of MR at Philips. “By partnering with NVIDIA to build an MR foundational model, we’re pioneering a new frontier for medical imaging.”

Philips will incorporate NVIDIA’s VISTA-3D and MAISI technologies into the model. VISTA-3D is an interactive foundation model built for 3D medical imaging, while MAISI is designed to generate synthetic images with or without anatomical labels. The companies intend to integrate the resulting AI model into existing MR systems and workflows.

Philips, headquartered in Amsterdam, has emphasized the scalability of the approach, positioning the foundational model as a platform for future AI-driven applications in MR.

The announcement was made during the 2025 International Society for Magnetic Resonance in Medicine (ISMRM) meeting.

No timeline was given for commercial deployment, and regulatory pathways were not discussed in the announcement.

MRI Homepage