Sparse shape model for fibular transfer planning in mandibular reconstruction.

In this paper, we propose an automated preoperative planning method that estimates a plan fitted to the data of a new patient using a planned dataset of previous patients. Although mandibular reconstruction with fibular segments needs preoperative planning for the precise placement of segments, recent interactive planning software cannot secure objectivity of the planning and time-consuming trial-and-error processes are required. The proposed method employs sparse shape modeling; in this modeling, we select a subset of the data from a prepared preoperative planning dataset to make an example or instance of reconstruction via a linear combination of the data. We conduct experiments using the dataset planned by medical doctors and compare the instance estimated by the proposed method to the manual placement by these doctors.