Philips teams with NVIDIA to build foundation model for MR AI

Philips and NVIDIA are collaborating to build a large, deep learning neural network foundation model to advance AI in MR imaging.

Announced during the 2025 International Society for Magnetic Resonance in Medicine meeting, the project lays the groundwork for a new generation of applications set to enhance MR image quality, accelerate scan times, and improve diagnostic workflow and accuracy across a wide range of clinical applications, Philips said.

The foundation model will enable zero-click planning of scans across different anatomies and has the potential to increase throughput, according to Philips. Radiologists will also be able to preview images and adjust quality and speed parameters before the actual scan. 

Additionally, the model will be able to facilitate automatic detection and interpretation of image findings, Philips noted. 

The project builds on NVIDIA's Vista-3D model for segmentation analysis and Medical AI for Synthetic Imaging (MAISI), the 3D model designed to generate synthetic images with or without anatomical annotations, according to the company.

Together, these technologies will create a domain-specific solution tailored to the challenges of MR imaging, Philips said.