Meeting Banner
Abstract #1407

Accurate 3D T2 Relaxometry with Simultaneous High-Resolution Structural Imaging using Deep Learning

Akshay S Chaudhari1, Arjun D Desai1, Zhongnan Fang2, Eric M Bultman1, Jin Hyung Lee3, Garry E Gold1, and Brian A Hargreaves1

1Radiology, Stanford University, Palo Alto, CA, United States, 2LVIS Corporation, Palo Alto, CA, United States, 3Neurology, Stanford University, Palo Alto, CA, United States

Rapidly obtaining high-resolution structural magnetic resonance images (MRI) and generating quantitative biomarkers, such as the T2 relaxation time, using a single sequence is useful for musculoskeletal imaging. However, high-resolution is at odds with high signal-to-noise ratio (SNR) in MRI, which makes it challenging to simultaneously optimize for image quality and quantitative accuracy. In this study, we demonstrate how deep-learning-based super-resolution can create high-resolution images with accurate T2 values using a prospectively-sampled 5-minute quantitative double-echo steady-state sequence. We validate this method using high-SNR reference sequences for T2 accuracy and high-resolution reference sequences and a reader study for image quality assessment.

This abstract and the presentation materials are available to members only; a login is required.

Join Here