Efficient Breast Deformation Simulation

Abstract

Breast surgery might benefit from image guidance, if position and extend of the disease could be visualized in the same breast deformation as the surgery is performed. Such visualizations, however, are challenging to obtain, since the positioning for image acquisition in MRI is prone, while the patient lies supine for surgical procedures, causing a considerable deformation of the breasts between the two states. In our contribution, we outline a set of novel algorithms and methods to improve an efficient simulation of breast deformation between the prone image and the supine surgery positioning. In particular, we propose several extensions to a highly efficient dynamic corotated finite element method (FEM), namely non-linear material properties, the sliding of the breast tissue on the chest wall, and a fine-tuning step to align the breast model to a measured surface. All extensions are carefully designed to keep the efficiency and stability of the approach, and thus to allow their application in clinical routine. We explore all novel techniques using synthetic and volunteer prone and supine breast MRI data and assess their feasibility towards accurate, yet efficient simulation of large breast deformations.