Open Source Imaging Initiative Builds Sub-$70,000 3D-Printed MRI as AI Sharpens 50mT Scans
Updated
Updated · Tom's Hardware · Jul 26
Open Source Imaging Initiative Builds Sub-$70,000 3D-Printed MRI as AI Sharpens 50mT Scans
1 articles · Updated · Tom's Hardware · Jul 26
Summary
Under $70,000 buys the OSI2 ONE, an open-source portable MRI with a 3D-printed core that has already been replicated globally as a far cheaper alternative to hospital scanners.
At 50mT, the device runs far below the 1.5T to 3T field strength of full-size MRIs, leaving weaker resolution and signal quality that supporters say specialized AI reconstruction can improve.
Brian Roemmele said models trained on high-field MRI data or physics-informed systems can denoise images, correct field inhomogeneity and adapt scan sequences in real time; synthetic data could help where patient datasets are unavailable.
New conventional MRI units typically cost $1.1 million to $3 million, while even refurbished machines start near $100,000 and require specialized rooms, making the portable system aimed at lower-resource hospitals and regions.
Regulatory hurdles remain for clinical use in tightly controlled markets, but the project argues a lower-resolution scanner could still expand diagnostic access if AI-assisted accuracy proves reliable.
Could AI-enhanced, 3D-printed MRI scanners democratize global healthcare, or will dangerous AI hallucinations lead to a crisis of misdiagnoses?
When medical imaging shifts to open-source software, who bears the legal liability if the AI reconstructs a non-existent tumor?
Breaking the MRI Barrier: How the OSI² ONE’s €25,000 Open-Source 50mT Scanner Is Disrupting Global Medical Imaging
Overview
The OSI² ONE is transforming medical imaging by replacing expensive, complex high-field MRI systems with a lightweight, ultra-low-field (50 mT) open-source scanner. Its innovative permanent magnet design eliminates the need for heavy infrastructure and costly helium cooling, making it portable and affordable for resource-limited settings. Although lower field strength reduces image quality, advanced AI techniques reconstruct and enhance scans to meet clinical needs. Open-source licensing and detailed documentation empower local teams to build, customize, and maintain scanners, reducing reliance on global supply chains. As of 2026, OSI² ONE is technically mature and moving toward medical certification, promising sustainable, accessible MRI worldwide.