eduzhai > Applied Sciences > Engineering >

An Unsupervised Approach to Ultrasound Elastography with End-to-end Strain Regularisation

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 10 pages

Abstract: Quasi-static ultrasound elastography (USE) is an imaging modality thatconsists of determining a measure of deformation (i.e.strain) of soft tissue inresponse to an applied mechanical force. The strain is generally determined byestimating the displacement between successive ultrasound frames acquiredbefore and after applying manual compression. The computational efficiency andaccuracy of the displacement prediction, also known as time-delay estimation,are key challenges for real-time USE applications. In this paper, we present anovel deep-learning method for efficient time-delay estimation betweenultrasound radio-frequency (RF) data. The proposed method consists of aconvolutional neural network (CNN) that predicts a displacement field between apair of pre- and post-compression ultrasound RF frames. The network is trainedin an unsupervised way, by optimizing a similarity metric be-tween thereference and compressed image. We also introduce a new regularization termthat preserves displacement continuity by directly optimizing the strainsmoothness. We validated the performance of our method by using both ultrasoundsimulation and in vivo data on healthy volunteers. We also compared theperformance of our method with a state-of-the-art method called OVERWIND [17].Average contrast-to-noise ratio (CNR) and signal-to-noise ratio (SNR) of ourmethod in 30 simulation and 3 in vivo image pairs are 7.70 and 6.95, 7 and0.31, respectively. Our results suggest that our approach can effectivelypredict accurate strain images. The unsupervised aspect of our approachrepresents a great potential for the use of deep learning application for theanalysis of clinical ultrasound data.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...