Self-Supervised Ultrasound Strain Elastography using SMURF

Displacement tracking is an integral part of ultrasound strain elastography (USE) and various approaches have been proposed in the literature. However, limitations exist in the estimation of the lateral displacement field in several of these algorithms for in vivo imaging. In this study, we explore the use of a Self-Teaching Multi-Frame Unsupervised Recurrent All-Pairs Field Transform (SMURF), an optical flow approach for USE, focusing on estimation of strain tensor estimates in both axial and lateral directions under an unsupervised setting using four-dimensional (4D) cost volumes. By retraining both the RAFT and SMURF network on B-mode images from a simulation dataset, and the SMURF network on B-mode in vivo datasets, we demonstrate that the fine-tuned RAFT model achieves comparable displacement and strain estimation accuracy and precision, especially in the lateral direction. In addition, the fine-tuned SMURF model showed comparable results to traditional window based, and optimization based methods for in vivo cases. Our results demonstrate the potential of using self-supervised techniques for in vivo datasets, which may be the best approach for estimating displacement fields for USE since it can utilize the large number of unlabeled datasets available with clinical studies and can extend to Lagrangian strain estimation in real-time.