Enhanced Statistical Shape Model: A Statistical-Based Tool to design Custom Orthopaedic Devices

Abstract

Dealing with the design of personalized medical devices, mass production is not an option that can be hypothesized. Indeed, a cumbersome production process must be considered in such cases, mainly to account for a delicate design phase that needs to take into consideration, as input, an anatomy that vary each time. This article discusses the development of a statistical tool able to support the design of patient-specific devices. By expanding the classical formulation of the Statistical Shape Model (SSM) with the introduction of multiple levels of information within the same model, the authors have experimented with the concept of an “enhanced SSM”. While the traditional SSM only provides information on the variations that a class of shapes can manifest, the eSSM may include more levels of information. The article discusses two possible mathematical formulations of such statistical tool. Its application to the design of custom-made pelvic implants is discussed. Such application scenario is described starting from the generation of the eSSM for the pelvis. The features of interest considered in this paper are the centers of the acetabular regions of the pelvis, the segmentation of the anatomy in a series of semantical regions that must be considered when developing a load-bearing implant. Finally, the conclusions of this research are drawn and discussed together with possible future development of eSSMs.