Automated Planning with Multivariate Shape Descriptors for Fibular Transfer in Mandibular Reconstruction

Abstract

Objective: This paper introduces methods to automate preoperative planning of fibular segmentation and placement for mandibular reconstruction with fibular flaps. Methods: Preoperative virtual planning for this type of surgery has been performed by manual adjustment of many parameters, or based upon a single feature of the reconstruction. We propose a novel planning procedure formulated as a non-convex minimization problem of an objective function using the multilateral shape descriptors. Results: A retrospective study was designed and 120 reconstruction plans were reproduced using computed tomography images with oral surgeons. The proposed automated planning model was quantitatively compared with both the existing model and the surgeons’ plans. Conclusion: The results show that the developed framework attains stable automated planning that agrees with the surgeons’ decisions. Significance: This method addresses trade-off problems between symmetric reconstruction and restoration of the native contour of the mandible.