Meeting Banner
Abstract #2530

Fetal brain age estimation and anomaly detection using attention-based deep ensembles with uncertainty

Wen Shi1,2, Guohui Yan3, Yamin Li2,4, Haotian Li1, Tintin Liu1, Yi Zhang1, Yu Zou3, and Dan Wu1
1Key Laboratory for Biomedical Engineering of Ministry of Education, Department of Biomedical Engineering, College of Biomedical Engineering & Instrument Science, Zhejiang University, Hangzhou, China, 2Key Laboratory of Biomedical Information Engineering of Ministry of Education, Department of Biomedical Engineering, School of Life Science and Technology, Xi'an Jiaotong University, Xi'an, China, 3Department of Radiology, Women's Hospital, School of Medicine, Zhejiang University, Hangzhou, China, 4School of Biomedical Engineering, Shanghai Jiaotong University, Shanghai, China

Accurate estimation of the brain age is important for the evaluation of brain development, especially in the fetal stage when little diagnostic tools are available. This study designed attention-based deep ensembles to estimate brain age in the normal developing fetus, based on axial T2-weighted in-utero MRI images from routine clinical scans. Mean absolute error of 0.803 week was achieved, and the attention maps highlighted the regions of interest associated with the estimation. Predictive uncertainty was simultaneously quantified, and together with the proposed prediction confidence, we were able to detect several types of anomalies, including small head circumference, malformations, and ventriculomegaly.

This abstract and the presentation materials are available to 2020 meeting attendees and eLibrary customers only; a login is required.

Join Here