Modeling Human Embryonic Brain Development

Brain development during the periconceptional period influences the current, future, and life-long
neurodevelopmental health of the developing embryo. Early brain development is monitored based
on manual measurements and visual inspection of standard planes in three-dimensional (3D)
ultrasound imaging. Manual measurements and visual inspection of standard planes are timeconsuming, operator-dependent, and only performed in two-dimensional planes. To save time,
reduce inter-rater variability, and fully employ 3D imaging, this thesis presents automatic AI
methods to model the brain development of the human embryo.
AI-based methods were developed for automatic alignment to a standard orientation, segmentation,
and for creating the 4D Human Embryonic Brain Atlas. Automatic alignment to a standard
orientation enables the derivation of standard planes used in clinical practice to monitor
development. Segmentation of the embryo, or other organs and structures, enables automatic
volumetric measurements. Finally, the 4D Human Embryonic Brain Atlas provides a template of the
normal brain per gestational day, which gives a unique, detailed, and accurate picture of brain
development during the first trimester. Using the atlas, the influence of periconceptional paternal
and maternal conditions and lifestyle behaviors can be studied. The presented methods were shown
to be accurate and time-saving. Implementing the presented methods can ultimately lead to earlier
and better detection, prevention, and treatment of neurodevelopmental disorders.