Generation of normative pediatric skull models for use in cranial vault remodeling procedures

Abstract

Purpose

While the goal of craniofacial reconstruction surgery is to restore the cranial head shape as much towards normal as possible, for the individual patient, there is, in fact, no normal three-dimensional (3D) model to act as a guide. In this project, we generated a library of normative pediatric skulls from which a guiding template could be fabricated for a more standardized, objective and precise correction of craniosynostosis.

Methods

Computed tomography data from 103 normal subjects aged 8–12 months were compiled and a 3D computational model of the skull was generated for each subject. The models were mathematically registered to a baseline model for each month of age within this range and then averaged, resulting in a single 3D point cloud. An external cranial surface was subsequently passed through the point cloud and its shape and size customized to fit the head circumference of individual patients.

Results

The resultant fabricated skull models provide a novel and applicable tool for a detailed, quantitative comparison between the normative and patient skulls for preoperative planning and practice for a variety of craniofacial procedures including vault remodeling. Additionally, it was possible to extract the suprafrontal orbit anatomy from the normative model and fabricate a bandeau template to guide intraoperative reshaping.

Conclusions

Normative head shapes for pediatric patients have wide application for craniofacial surgery including planning, practice, standarized operative repair, and standardized measurement and reporting of outcomes.

Appendix A. 3D modeling methodology

Within each data class categorized by age, an iterative registration process was performed to achieve correspondence between the 3D models. We selected a mathematical algorithm for point set registration as opposed to manual registration that requires matching the models in 3D space via predefined anatomical markers on the skull surface; see, for example, Ref. [12]. The Coherent Point Drift (CPD) algorithm of Myronenko and Song was used for this purpose. The method considers the alignment of two point sets as a probability density estimation problem and fits the Gaussian mixture model (GMM) centroids representing an initial ‘base’ point set to the second point set by maximizing the likelihood. At the optimum, the two point sets become aligned and correspondence is achieved using the maximum of the GMM posterior probability for a given data point. The GMM centroids are then forced to move coherently as a group in order to preserve the topological structure of the point sets. The rigid option of the algorithm was used for this study’s skull model registration. For each data class, the ‘base’ model to which all others were registered was chosen as the skull featuring the highest level of anatomical detail in the region of interest, in our case being the forehead region and the back part of the cranium. Once point-based correspondence was achieved within each data class, a linear averaging step was performed to generate a single representative point cloud for the class. These resultant points formed the basis of a final triangulation step that allowed a surface describing the external boundary of the average 3D skull model to be reconstructed from the point cloud in stereolithography (STL) form. In addition, an average composite skull model was created from the total number of data sets available for this study (N = 103) in the same manner as above. This composite skull model was considered the most accurate estimate of what a normal child’s skull resembled in the cumulative age range of 8–12 months by consensus. It was then possible to alter and rescale the dimensions of the computational skull model to customize it for the age and size (e.g., head circumference) of the individual subject being assessed versus the norm. The ‘base’ data set for the composite skull was selected as the most detailed among the total available.