Motivation: Previous published work has focused on technical feasibility of automated fetal brain biometry methods. Enabling reliable automated biometry directly during MRI examinations could greatly add to clinical utility. Goal(s): To automate fetal brain biometry measurements and centile calculations from 3D motion-corrected T2w MRI. Approach: Automated extraction of fetal brain biometry measurements using deep learning localisation of anatomical landmarks. Results: Our study automates 11 routinely reported fetal brain measurements trained on a large cohort of control subjects, across a wide range of gestational ages, field strengths and scanning parameters. Impact: The benefits of automating the time-consuming manual biometry method include improved diagnostic accuracy, confidence and reliability of derived measurements, enabling precise quantification of fetal brain development, as well as improved workflow efficiency and turnaround time for radiology reports.
Luis et al. (Tue,) studied this question.