Introduction
Magnetic Resonance Imaging (MRI) is a non-invasive radiologic examination which does not produce ionizing radiation like X-ray
imaging and Computed Tomography (CT). Due to superior soft tissue contrast and image resolution, MR is an important part of clinical
practice in making accurate diagnosis, monitoring therapeutic response, and follow-ups. Despite these advantages, however, MR exams
typically require long scan times – limiting its use compared to other modalities. To this effect, numerous research and technical
developments are now focused on reducing the scan time to increase scanner efficiency and improving the patient experience.
SwiftMR™ is FDA 510(k)-cleared* deep learning (DL)-based software medical device developed by AIRS Medical. SwiftMR™ reduces
image noise and increases sharpness of MR images based on its vast training dataset of high-quality MR images.
The following clinical cases were collected from research conducted at Seoul National University Hospital (Seoul, Korea). This study
was approved by the institutional review board (IRB) and informed consents were received from all enrolled patients. The purpose of
this study was to clinically evaluate the quality of spine (C-spine, L-spine) images processed with SwiftMR™, where input images were
acquired at were acquired under reduced scan time compared to the institutional standard. To do this, images were acquired from
enrolled patients using both standard and faster scan protocol, then images were compared after applying SwiftMR™ to the latter.
* FDA 510(k)-cleared for Standard of Care (SOC) brain, spine(C/T/L) and MSK (Knee, ankle, shoulder, and hip) MR images (K220416).