Improved fat suppression of head and neck MRI using 3D modified mDixon-TSE with deep learning constrained Compressed SENSE.

Motivation: Conventional 3D T1WI in head and neck MRI has difficulty in obtaining uniform fat suppression images within a feasible examination time. Goal(s): The purpose of this study is to realize a 3D MRI of the head and neck region with uniform fat suppression. Approach: We evaluated the feasibility of high-resolution MRI with uniform fat suppression in the orbit and larynx with 3D 3-point mDixon TSE using Deep Learning constrained Compressed SENSE (CS-AI). Results: The 3D 3-point mDixon TSE with CS-AI was able to provide high-resolution images of the orbit and larynx with robust fat suppression within a feasible scan time. Impact: 3D 3-point mDixon TSE with CS-AI provides high-resolution MRI images with uniform fat suppression. This technique is expected to accurately delineate small lesions and tissue structures in the head and neck region.