Finite element model of breast compression during mammography

Abstract

Diagnosis of the breast cancer involves the reconstruction of images taken during mammography, typically involving cranio-caudal (CC, or head-to-toe) and medio-lateral oblique (MLO, or shoulder-to-opposite hip) compression. Current algorithms to associate the two images are not appropriate for image registration due to the large nonhomogenous deformation of the breast during a mammographic procedure. A finite element model of the breast which can simulate the breast mechanics under compression will help to establish the relations between the two images, making more effective tracking of tumours possible. We performed compression experiments on a cylindrical silicon gel phantom and compared experimental deformations with predictions from our computational framework, which incorporates contact mechanics with finite deformation non-linear elasticity. Using the known material properties of the gel, compression predictions matched closely with experimental results. We then performed compression experiments on a gel phantom moulded into a realistic anatomical breast shape to mimic mammography, and compared deformations with predictions using a patient specific finite element model. The combination of tissue mechanics with compression modelling has the potential to provide a virtual tool to track breast tumours, and to assist with surgical guidance for clinical procedures such as biopsy and radiotherapy