Ultrasound estimation of breast tissue biomechanical properties using a similarity-based non-linear optimization approach

The in vivo estimation of tissue elasticity parameters is important for realistic tissue deformation modelling and diagnosis tasks such as cancer mass detection and characterization. Elastography (strain imaging) provides non-quantitative information about tissue stiffness and is becoming well established clinically. Acoustic radiation force imaging, supersonic shear wave imaging and (Young's) modulus imaging are all evolving as quantitative methods. This paper concerns the latter. The established approach to Young's modulus reconstruction involves solving the so-called inverse elasticity problem of recovering elastic parameters by comparing the displacement field from the measurements and the theoretical tissue deformation modelling. In this paper, a new modulus imaging pathway is proposed in which the inverse problem of elasticity reconstruction has been converted into a global optimization problem involving an image similarity measure. This approach is applied to ultrasound images and the tissue elasticity distribution is recovered by adaptively adding new regions based on the local mismatch of the images using a multiscaled split-and-merge approach. For this experimental evaluation of the performance of the proposed method, first synthetic images are used with different Gaussian noise levels. The reconstructed elasticity shows good agreement with the ground-truth. A comparison of the proposed method with a conventional displacement-based method is made using a gelatine phantom. The results using both methods agree remarkably well with the theoretical prediction. A further in vivo study on 19 breast cancer masses was performed. The present method estimated an elasticity contrast of 6.41 ± 0.98 between cancerous tissue and normal tissue, which is consistent with values reported elsewhere in the literature.