An Adaptive Landmark Scheme for Modeling Brain Deformation in Diffusion-Based Tumor Growth

Abstract

In the present work a novel model to simulate mass effect caused by brain tumors is described. The progression of the tumor is modeled by means of a deterministic reaction- diffusion equation, which is discretized on a highresolution voxel grid. This model does not inherently account for the mass effect of the tumor. Therefore, the computed tumor cell density is linked to a parametric deformation model.

More precisely, this deformation model is based on a thinplate spline interpolation strategy. Correspondence during the spatio-temporal progression of the tumor is defined by tracking landmarks, which are attached to the boundary of the gross tumor volume (GTV). To suppress deformation of rigid structures, i.e. the skull, fixed shielding landmarks are introduced into the model. An adaptive landmark scheme is used that allows for introducing new landmarks into the model as the tumor progresses.

The present work has to be considered as a feasibility study. First qualitative results demonstrate the capability of the described method, which allows for plausibly approximating the mass effect caused by diffusive brain tumors.