Evaluation of a similarity-based elastography technique using four similarity metrics

This study uses a similarity-based elastography method to attempt to reconstruct elastic modulus values in a simulation of cancerous tumors in the breast. Four similarity metrics were compared as to their ability to resolve a single near field tumor, far field tumor and a multiple tumor scenario. Mutual Information provided the best reconstruction for near field tumors while Gradient Correlation provided the best reconstruction for the far field and multiple tumor models. In all elasticity images, good tumor localization was achieved; however, the limited contrast resolution needs further investigation.