Edge-Guided Deep Learning Model to Predict Fetal Brain Age Using MRI.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Fetal Development

Background:

• Deep learning models for fetal brain age prediction often overlook local edge details, potentially limiting accuracy.
• Existing methods may not fully capture the nuances of fetal brain development from MRI scans.

Purpose of the Study:

• To develop a novel deep learning model that integrates global edge information for enhanced fetal brain age prediction.
• To improve the accuracy and reliability of fetal age assessment from MRI data.

Main Methods:

• A retrospective dataset of 1630 fetal brain MRI scans (22-38 weeks gestation) was utilized.
• A neural network incorporating global edge information was trained and optimized using mean absolute error (MAE) and R-squared (R²).
• The dataset was split into training (4/5) and testing (1/5) sets for model validation.

Main Results:

• The edge-guided deep learning model achieved high accuracy, with a mean absolute error (MAE) of 0.79 weeks and an R² value of 0.94.
• The model demonstrated superior performance compared to existing fetal age prediction methods.
• The approach enhances the stability and reliability of regression models for fetal age estimation.

Conclusions:

• The proposed edge-guided deep learning model significantly outperforms existing methods for fetal brain age prediction.
• This novel approach offers a valuable tool for accurate clinical assessment of fetal brain development.