Predicting tumor location from prone to supine breast MRI using a simulation of breast deformation

Abstract

Breast cancer is one of the biggest killers to women, and early diagnosis is essential for improved prognosis. The shape of the breast varies hugely between the scenarios of magnetic resonance (MR) imaging (patient lies prone, breast hanging down under gravity) and ultrasound (patient lies supine). Matching between such pairs of images is considered essential by radiologists for more reliable diagnosis of early breast cancer. In this paper, a method to predict tumor location by simulating the breast deformation from breast in the prone position to the compressed breast in the supine position was developed, which is based on a 3-D patient-specific breast model constructed from MR images with the use of the finite-element method and nonlinear elasticity. The performance was assessed by the mean distance between corresponding lesion locations for three cases. A mean accuracy of 4.94mm in Euclidean distance was achieved by using the proposed method. Experiments using actual images show that the method gives good predictions which can be used to find exact correspondences between tumors location in prone and supine breast images.