Machine learning enables cheaper, safer, low-power MRI

Machine learning enables cheaper and safer low-power magnetic resonance imaging (MRI) without sacrificing accuracy, according to a new study. According to the authors, these advances pave the way for affordable, patient-centered, deep learning-powered ultra-low field (ULF) MRI scanners to fill unmet clinical needs in diverse medical settings around the world. They say they can address your needs.

Magnetic resonance imaging (MRI) has revolutionized medicine by providing non-invasive, radiation-free imaging. There are great expectations for advances in medical diagnosis using artificial intelligence. However, despite 50 years of development, MRI remains largely inaccessible, especially in low- and middle-income countries. This is primarily due to the high costs associated with standard superconducting MRI scanners and the specialized infrastructure required for their operation. These scanners are typically located in specialized radiology departments or large imaging centers, limiting their availability to smaller medical facilities. Additionally, the need for radiofrequency (RF) shielded rooms and high power consumption further limit access to MRI technology.

To address MRI accessibility challenges, Yujiao Zhao et al. developed a low-power, highly simplified ULF MRI scanner that operates from a standard wall power outlet and without the need for RF or magnetic shielding. announced. This scanner uses a compact 0.05 Tesla (T) magnet (most MRI machines use 1.5 T magnets, but some MRI machines use up to 7 T magnets) and eliminates electromagnetic interference signals. It incorporates active sensing and deep learning to address and improve image quality. Additionally, the device only consumed 1800 watts (W) during the scan, compared to more than 25000 W consumed by conventional MRI.

Zhao other. performed imaging on healthy volunteers and showed that the device can produce images as clear and detailed as the high-power MRI machines currently used in clinics. In a related perspective, Udunna Anazodo and Stefan du Plessis note the limitations and challenges that need to be addressed before low-field MRI can be widely adopted for clinical use. “Low-field MRI is not yet mature enough to allow cost-effective access to medical images,” they write. “Having low-field MRI available to more communities around the world without any hindrance will prove its potential as an essential and environmentally sustainable health technology.”