MENLO PARK — FastMRI, a collaborative initiative between Facebook AI and NYU Langone Health, has achieved a major milestone by demonstrating that AI-accelerated MRI scans can produce images that are diagnostically equivalent to traditional MRIs. The fastMRI technology uses artificial intelligence to generate high-quality MRI images with just a quarter of the data typically required. This innovation drastically reduces scan times and could enhance patient experiences, expand access to MRIs, and create new applications for the technology.
In a clinical study, radiologists were unable to differentiate between AI-generated images and those from conventional MRIs, and they even rated the AI-generated scans as higher in quality. The AI-powered system, which relies on PyTorch, enables MRI scans to be completed nearly four times faster than traditional methods.
The initiative also aims to share the technology openly, encouraging collaboration and future advancements in AI-driven medical imaging. FastMRI not only has the potential to revolutionize MRI practices but could also reduce the need for radiation-heavy alternatives like X-rays and CT scans, benefiting both patients and the healthcare system.
