Computational Patient Avatars for Surgery Planning

In this paper a new method is described for the generation of computational patient avatars for surgery planning. By “patient avatar” a computational, patient-specific, model of the patient is meant, that should be able to provide the surgeon with an adequate response under real-time restrictions, possibly including haptic response. The method is based on the use of computational vademecums (F. Chinesta et al., PGD-based computational vademecum for efficient design, optimization and control. Arch. Comput. Methods Eng. 20(1):31–59, 2013), that are properly interpolated so as to generate a patient-specific model. It is highlighted how the interpolation of shapes needs for a specialized technique, since a direct interpolation of biological shapes would produce, in general, non-physiological shapes. To this end a manifold learning technique is employed, that allows for a proper interpolation that provides very accurate results in describing patient-specific organ geometries. These interpolated vademecums thus give rise to very accurate patient avatars able to run at kHz feedback rates, enabling not only visual, but also haptic response to the surgeon.