Optimal Method for Fetal Brain Age Prediction Using Multiplanar Slices From Structural Magnetic Resonance Imaging

Area of Science:

• Neuroimaging
• Developmental Biology
• Artificial Intelligence

Background:

• Accurate fetal brain age prediction via MRI aids in detecting abnormalities and developmental risks.
• Gestational age (GA) assessment is vital for monitoring fetal development and health outcomes.

Purpose of the Study:

• To develop and validate a novel method for predicting fetal brain age using multiplanar magnetic resonance imaging (MRI) slices.
• To assess the accuracy and efficiency of a 2D single-channel convolutional neural network (CNN) for fetal brain age prediction.

Main Methods:

• A 2D single-channel CNN was trained on MRI data from 220 healthy fetuses (15.9–38.7 weeks GA).
• Multiplanar MRI slices were utilized without interslice motion correction, with brain age determined by the mode of multiple slice predictions.
• Saliency maps were generated to identify key anatomical features contributing to age prediction.

Main Results:

• The proposed method achieved a mean absolute error (MAE) of 0.125 weeks (0.875 days) on the primary dataset.
• Multiplanar slice utilization significantly reduced prediction error compared to single slices or stacks.
• The 2D single-channel CNN with multiplanar slices outperformed 2D multi-channel and 3D CNNs in stack-wise MAE (0.304 weeks vs. 0.979 and 1.114 weeks, respectively).
• External validation on 21 fetuses yielded an MAE of 0.508 weeks, with the 2D single-channel CNN outperforming other CNN architectures.

Conclusions:

• The developed 2D single-channel CNN method accurately predicts fetal brain age using multiplanar MRI slices.
• This approach offers a simplified, computationally efficient alternative to complex 3D CNNs and extensive preprocessing.
• The method holds promise for improved identification of fetal brain abnormalities and developmental monitoring.