An open-source MRI platform combines 3D-printed components, low-field magnetic imaging, and AI-assisted reconstruction to reduce the cost and complexity of medical imaging systems.
There is now an open-source MRI scanner called the OSI2 ONE developed by the Open Source Imaging Initiative (OSI2), which can be assembled for less than $70,000. The scanner features a 3D-printed core and is designed to provide a cheaper alternative to the currently existing MRI systems that can be used in research, education, and clinical settings.
While standard MRI scanners operate at magnetic field strengths between 1.5 T to 3 T, OSI2 ONE employs a much weaker magnetic field (50 mT) provided by a permanent magnet. This design eliminates the need for cryogenic cooling and many costly parts; however, the reduced strength of the magnetic field reduces spatial resolution and the signal-to-noise ratio.
To address the poor imaging characteristics, the project uses an AI-assisted image reconstruction technique. Technology analyst Brian Roemmele suggests that deep learning algorithms which are trained on data obtained through high-field MRI machines or physics-informed methods can reduce image noise, correct magnetic field inhomogeneity, improve spatial resolution, and adjust imaging sequences depending on the quality of signal in real time.
The MRI machine itself is developed to be an open-source hardware platform that allows researchers and engineers to inspect, manipulate, and even reproduce the device using publicly available documentation. Due to its modular design and its extensive use of 3D-printed components, it is relatively easy to produce, while the platform itself encourages further development.
Although OSI2 ONE is not intended to replace high-field MRI devices used in hospitals and clinics, this project demonstrates the benefits that open-source hardware, additive manufacturing, and AI-based image processing can improve access to medical diagnostics.
