Charting neuroimaging-based head circumference and cranial shape across development

A new automated imaging framework called CranioTrace has been developed to streamline the measurement of head circumference and cranial shape during childhood and adolescence. This method offers a more consistent and scalable approach compared to the existing manual techniques, which are prone to variability and anatomical limitations.

CranioTrace employs topology-constrained segmentation, contour regularization, and geometry-based quality control to automatically extract relevant data from various MRI and CT scans, eliminating the need for manual intervention. The framework was tested on a dataset of 9,685 scans spanning from birth to early adulthood across 25 different cohorts.

The results demonstrated that CranioTrace's imaging-derived head circumference measurements aligned well with traditional tape measurements, showing high reproducibility (intraclass correlation up to 0.99) and consistent performance across different imaging modalities. Moreover, CranioTrace goes beyond conventional head circumference by also quantifying cranial shape and asymmetry, which are crucial for capturing developmental dynamics that cannot be observed through standard measurements.

By integrating these advanced cranial phenotypes into population models, CranioTrace reveals distinct patterns in head growth. Rapid expansion occurs during early life, followed by a nonlinear deceleration and eventually a stable pattern of differences between males and females. This information can be particularly valuable when applied to neurogenetic cohorts, where CranioTrace has identified consistent shifts in growth trajectories.

For example, individuals with neurofibromatosis type 1 and 16p11.2 deletion exhibited higher head circumference, while those with 22q11.2 deletion and 16p11.2 duplication showed lower values.

CranioTrace effectively transforms clinical imaging archives into standardized, population-referenced markers of cranial development. By providing scalable cranial phenotypes linked to probabilistic reference models, this framework offers a solid foundation for imaging-based growth charting, integrated skull-brain phenotyping, and precise assessment of neurodevelopment.