The LowGAN model improves lesion visibility and brain volume analysis using Hyperfine’s 64-mT system.
Article Summary:
A recent study published in Radiology highlights how deep learning could elevate the clinical potential of portable low-field MRI in multiple sclerosis (MS) care. Researchers developed a generative adversarial network model, LowGAN, to enhance image quality from 64-mT scans by synthesizing views closer to those of conventional 3T MRI.
Tested on 63 patients with MS, the LowGAN-enhanced images demonstrated improved white matter lesion detection, better image quality across T1, T2, and FLAIR sequences, and more accurate brain volume measurements—including thalamic atrophy, a key marker for neurodegeneration. This AI-powered approach also preserved critical lesion characteristics while reducing background noise.
Although the study’s dataset was relatively small, the findings suggest LowGAN could help bridge the gap between portable and high-field MRI—offering new possibilities for remote or resource-limited MS care.
Want to learn more?
Click below to explore the original article
Go back to News & Updates page.
Let’s Stay in Touch
We’d love to hear from you!
For any questions or inquiries, please fill out the form below or use our contact details. Our team is ready to respond promptly and assist you on your journey.