The 4D Human Embryonic Brain Atlas: Spatiotemporal atlas generation for rapid anatomical changes.

Area of Science:

• Neuroscience
• Medical Imaging
• Developmental Biology

Background:

• Early brain development is critical for lifelong neurodevelopmental health.
• Clinical understanding of normal embryonic brain anatomy during rapid development is limited.
• Ultrasound imaging captures dynamic changes in the embryonic brain between gestational weeks 8 and 12.

Purpose of the Study:

• To create a detailed 4D Human Embryonic Brain Atlas for insights into normal embryonic brain development.
• To identify deviations from normal development using advanced computational methods.
• To improve the detection and understanding of prenatal neurodevelopmental disorders.

Main Methods:

• Utilized a deep learning-based approach for groupwise registration and spatiotemporal atlas generation.
• Developed a time-dependent initial atlas with penalization for deviations to maintain age-specific anatomy.
• Generated and validated the atlas using 831 3D ultrasound images from 402 subjects (gestational weeks 8-12).

Main Results:

• The deep learning approach successfully captured rapid anatomical development of the embryonic brain.
• An ablation study confirmed that time-dependent atlas and penalization are crucial for anatomical accuracy.
• Visual comparisons validated the anatomical accuracy against an existing ex-vivo embryo atlas.

Conclusions:

• The 4D Human Embryonic Brain Atlas provides unprecedented insights into early brain development.
• This atlas can significantly improve the detection, prevention, and treatment of prenatal neurodevelopmental disorders.
• The method offers a valuable tool for studying dynamic embryonic brain anatomy.