Breast Image Fusion Using Biomechanics

Abstract

Breast cancer is a leading cause of cancer mortality in women worldwide. Biophysical mathematical models of the breast have the potential to aid in the diagnosis and treatment of breast cancer. This thesis presents research on the development and validation of biomechanical models of the breast subject to gravity and compressive loads. The finite element method was used to implement the theory of finite elasticity coupled with contact mechanics in order to simulate the large non-linear deformations of the breast tissues. Initially, validation studies were conducted using a breast phantom, which was placed in different orientations with respect to the gravity loading and compressed using a custom made device. A novel application of a block matching image processing method was used to quantitatively assess the accuracy of the biomechanics predictions throughout the entire phantom. In this way, systematic changes to the assumptions, parameters, and boundary constraints of the breast models could be quantitatively assessed and compared. Using contact mechanics to model the interactions between the ribs and breasts can improve the accuracy of simulating prone to supine deformations due to the relative sliding of the tissues, as was observed using MRI studies on volunteers. In addition, an optimisation framework was used to estimate the heterogeneous mechanical parameters of the breast tissues, and the improvements to the models were quantified using the block matching comparison method. A novel multimodality framework was developed and validated using MR and X-ray images of the breast phantom before being applied to clinical breast images. Using this framework, it was shown that the parameters of the model (boundary conditions, mechanical properties) could be estimated and the image alignment improved. The biomechanical modelling framework presented in this thesis was shown to reliably simulate both prone to supine reorientation, and prone to mammographic compression, deformations. This capability has the potential to help breast radiologists interpret information from MR and X-ray mammography imaging in a common visualisation environment. In future, ultrasound imaging could also be incorporated into this modelling framework to aid clinicians in the diagnosis and management of breast cancer.