Comparison of Representation Learning Techniques for Tracking in time resolved 3D UltrasoundDownload PDF

Apr 06, 2021 (edited Jun 15, 2021)MIDL 2021 Conference Short SubmissionReaders: Everyone
  • Keywords: radiotherapy, deep learning, latent space, clustering
  • Abstract: 3D ultrasound (3DUS) becomes more interesting for target tracking in radiation therapy due to its capability to provide volumetric images in real-time without using ionizing radiation. It is potentially usable for tracking without using fiducials. For this, a method for learning meaningful representations with which recognizing anatomical structures in different time frames is capable would be useful. In this study, 3DUS patches are reduced into a 128-dimensional representation space using conventional autoencoder, variational autoencoder and sliced-wasserstein autoencoder. In the representation space, the capability of separating different ultrasound patches as well as recognizing similar patches is investigated and compared based on a dataset of liver images. Two metrics to evaluate the tracking capability in the representation space are proposed. It is shown that ultrasound patches with different anatomical structures can be distinguished and sets of similar patches can be clustered in representation space. The results indicate that the investigated autoencoders have different levels of usability for target tracking in 3DUS.
  • Paper Type: both
  • Primary Subject Area: Unsupervised Learning and Representation Learning
  • Secondary Subject Area: Application: Other
  • Paper Status: original work, not submitted yet
  • Source Code Url: The source code can be made available by contacting the corresponding author.
  • Data Set Url: The dataset used in this paper was first published in: Ipsen et al., "Towards automated ultrasound imaging - robotic image acquisition in liver and prostate for long-term motion monitoring", Physics in Medicine & Biology, 2021. The dataset is available by contacting the corresponding author of that paper.
  • Registration: I acknowledge that publication of this at MIDL and in the proceedings requires at least one of the authors to register and present the work during the conference.
  • Authorship: I confirm that I am the author of this work and that it has not been submitted to another publication before.
4 Replies

Loading