eduzhai > Applied Sciences > Engineering >

Deep-Learning Based Adaptive Ultrasound Imaging from Sub-Nyquist Channel Data

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 9 pages

Abstract: Traditional beamforming of medical ultrasound images requires sampling ratessignificantly higher than the actual Nyquist rate of the received signals. Thisresults in large amounts of data to store and process, translating to big,expensive and power-hungry devices. In light of the capabilities demonstratedby deep learning methods over the past years across a variety of fields,including medical imaging, it is natural to consider their ability to recoverhigh-quality ultrasound images from partial data. Here, we propose an approachfor deep-learning based reconstruction from temporally and spatiallysub-sampled data. We begin by considering sub-Nyquist sampled data,time-aligned in the frequency domain and transformed back to the time domainwith no additional recovery steps. This results in low resolution andcorruption due to loss of frequencies and aliasing. The data is further sampledspatially to emulate acquisition from a sparse array. It is then given as inputto an encoder-decoder convolutional neural network which is trained separatelyfor each rate reduction, with targets generated from minimum-variance (MV)beamforming of the fully-sampled data. Our approach yields high-quality B-modeimages, with higher resolution than previously proposed reconstructionapproaches (NESTA) from compressed data as well as delay-and-sum beamforming(DAS) of the fully-sampled data. In terms of contrast to noise ratio, it iscomparable to MV beamforming of the fully-sampled data, thus enabling betterand more efficient imaging than is mostly used in clinical practice today.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×