Automatic planning of T2-weighted fetal scans at 0.55T using fetal brain landmark detection

Abstract

Motivation: Fetal MRI plays an important role in clinical and research settings. The variability of the fetal position and extensive fetal motion, however, create challenges limiting the use of fetal MRI mainly to specialist centres. Goal(s): Real-time fully automatic planning of true radiological fetal brain planes for anatomical TSE scans. Approach: Deep-learning based detection of key landmarks in the fetal brain on a whole-uterus EPI scan enables the subsequent automatic calculation of the radiological plane for the TSE scan. Results: Prospective results on three fetal MRI scans on a clinical low-field 0.55T MRI scanner illustrate the ability of the framework to perform diagnostic planning. Impact: Fully automated planning of radiological planes for low-field fetal MRI demonstrates time efficiency and carries the potential to significantly widen accessibility to fetal MRI beyond specialist centres.

Ähnliche Arbeiten