We are developing a technique for freehand ultrasound elastography of breast lesions based on estimation of tissue deformation using Bayesian regularization. Displacements fields between RF fields acquired at rest and compressed states are estimated by minimizing a cost function consisting of echo amplitude conservation constraints and roughness penalty. Adaptations of a multi-scale deterministic optimization algorithm have been made to take into account the specific nature of RF data and improve and accelerate the convergence. Performances evaluated on gelatin phantoms demonstrate up to an order of magnitude reduction in image noise compared to conventionally used cross-correlation techniques without deterioration of contrast and resolution. Qualitative improvements are illustrated with a sequence acquired in vivo from a patient with fibrocystic changes and radial scars.