A multi-tissue mass-spring model for computer assisted breast surgery

The aim of this work was to develop and validate a 3D female breast deformation model for computer assisted breast surgery. Magnetic resonance (MR) image data of a patient undergoing breast biopsy, were acquired using two different protocols with the patient in prone position: (i) uncompressed breast and (ii) compressed breast, with lateral single breast compression, realized with a movable slab. The acquired images were then segmented using a semi-automatic procedure and from the extracted volumes of interest tetrahedral meshes representing skin, fat and mammary glands were generated. Tissue deformation was ruled by a mass-spring model: first, an iterative approximation algorithm was implemented to estimate the spring's rest length and stiffness, accounting for gravity force; then the resulting parameters were used to deform the uncompressed breast model in order to reach the real compressed one (ground truth). Results showed that gravity force applied to the mesh was properly compensated by the internal elastic forces, leading to a distance between the deformed mesh and the reference data of 0.036 ± 0.092 mm (median ± inter quartile range). The point to mesh residual distance between the deformed mesh and the ground truth was 1.224 ± 2.202 mm (median ± inter quartile range). Further investigation on a larger patient dataset is required for a more robust confirmation of model accuracy in predicting breast deformations.