Improving Cranial Vault Remodeling for Unilateral Coronal Craniosynostosis-Introducing Automated Surgical Planning

Abstract

Study design: Cranial vault remodeling (CVR) for unicoronal synostosis is challenging due to the asymmetric nature of the deformity. Computer-automated surgical planning has demonstrated success in reducing the subjectivity of decision making in CVR in symmetric subtypes. This proof of concept study presents a novel method using Boolean functions and image registration to automatically suggest surgical steps in asymmetric craniosynostosis.

Objective: The objective of this study is to introduce automated surgical planning into a CVR virtual workflow for an asymmetric craniosynostosis subtype.

Methods: Virtual workflows were developed using Geomagic Freeform Plus software. Hausdorff distances and color maps were used to compare reconstruction models to the preoperative model and a control skull. Reconstruction models were rated as high or low performing based on similarity to the normal skull and the amount of advancement of the frontal bone (FB) and supra-orbital bar (SOB). Fifteen partially and fully automated workflow iterations were carried out.

Results: FB and SOB advancement ranged from 3.08 to 10.48 mm, and -1.75 to 7.78 mm, respectively. Regarding distance from a normal skull, models ranged from .85 to 5.49 mm at the FB and 5.40 to 10.84 mm at the SOB. An advancement of 8.43 mm at the FB and 7.73 mm at the SOB was achieved in the highest performing model, and it differed to a comparative normal skull by .02 mm at the FB and .48 mm at the SOB.

Conclusions: This is the first known attempt at developing an automated virtual surgical workflow for CVR in asymmetric craniosynostosis. Key regions of interest were outlined using Boolean operations, and surgical steps were suggested using image registration. These techniques improved post-operative skull morphology.

Keywords: automation; craniofacial; craniosynostosis; pediatrics; virtual surgical planning.

Figure 1.

Virtual workflow for a unilateral FOA procedure in a patient with UCS is shown. This is one example of how the bone components can be manipulated to approximate a normal head shape. A) An STL skull is imported into Geomagic Freeform Plus software. B) The SOB ROI is delineated. C) The FB ROI is delineated. D) The FB and SOB ROIs are isolated to allow their individual manipulation. E) The FB ROI is removed simulating the FB osteotomy. F) The SOB is removed simulating the SOB osteotomy. G) The SOB ROI is reshaped using a bending tool. H) The SOB ROI is repositioned on the skull. I) The FB ROIs are rotated and exchanged from left to right and vice versa. J) The virtual surgery is finished and the reconstructed skull model is complete.

Figure 2.

Points A and B represent the right and left SOBs of the skull model, and points C and D represent the right and left frontal bones, respectively. These represent the 4 points where maximum Hausdorff surface distance measurements were taken to analyze the shape of the virtually reconstructed skull. The warm hues in the color map denote areas in the pre-operative skull model that were advanced more than areas in cooler hues. In the model below, it is clear that the right frontal bone on the synostotic side was advanced more than the left frontal bone, to achieve a more symmetric forehead shape.

Figure 3.

A Boolean subtraction function can be used to automatically delineate the FB ROI in the UCS skull. In panel A, the grey skull is the UCS skull and the beige skull is the matched normal skull. The right frontal bone area on the UCS skull is not visible, due to the flattened and retropositioned frontal bones on the synostotic side. Panel B highlights the area in blue that is automatically calculated as the frontal bone ROI using a Boolean subtraction operation, taking advantage of intersecting areas between the 2 skulls. Panel C shows the UCS skull, now in beige, with the frontal bone ROI isolated. The normal skull model was subtracted from the UCS skull to highlight the region that is flattened in the UCS skull. Panel D shows the frontal bone ROI highlighted in green. Panel E demonstrates how the ROI can be modified to respect surgical boundaries and maintain a separate ROI for the supra-orbital bar.

Figure 4.

A Boolean subtraction function can be used to automatically delineate the component of the SOB ROI in the UCS skull that requires reshaping and advancement. A) The UCS SOB (beige) and the control SOB (grey) are registered and shown from a bird’s eye view. B) The intersection between where the UCS and normal SOB overlap and begin to diverge is demonstrated. This is what will be automatically highlighted with the Boolean subtraction function. C) The control SOB is subtracted from the UCS SOB using a Boolean function. D) This image demonstrates the problematic resulting UCS SOB with the various areas of missing bone where areas overlapped. E) The “normal” part of the SOB ROI is highlighted in green and subtracted from the UCS SOB to reduce the amount of bone loss. F) Only the SOB component that requires advancement and reshaping is isolated.