Automated instance segmentation and registration of spinal vertebrae from CT-Scans with an improved 3D U-net neural network and corner point registration.

Journal: Computers in biology and medicine
Published Date:

Abstract

This paper presents a rapid and robust approach for 3D volumetric segmentation, labelling, and registration of human spinal vertebrae from CT scans using an optimised and improved 3D U-Net neural network architecture. The network is designed by incorporating residual and dense interconnections, followed by an extensive evaluation of different network setups by optimising the network components like activation functions, optimisers, and pooling operations. In addition, the network architecture is optimised for varying numbers of convolution layers per block and U-Net levels with fixed and cascading numbers of filters. For 3D virtual reality visualisation, the segmentation output of the improved 3D U-Net network is registered with the original scans through a corner point registration process. The registration takes into account the spatial coordinates of each segmented vertebra as a 3D volume and eight virtual fiducial markers to ensure alignment in all rotational planes. Trained on the VerSe'20 dataset, the proposed pipeline achieves a Dice score coefficient of 92.38% for vertebrae instance segmentation and a Hausdorff distance of 5.26 mm for vertebrae localisation on the VerSe'20 public test dataset, which outperforms many existing methods that participated in the VerSe'20 challenge. Integrated with Singular Health's MedVR software for virtual reality visualisation, the proposed solution has been deployed on standard edge-computing hardware in medical institutions. Depending on the scan size, the deployed solution takes between 90 and 210 s to label and segment vertebrae, including the cervical vertebrae. It is hoped that the acceleration of the segmentation and registration process will facilitate the easier preparation of future training datasets and benefit pre-surgical visualisation and planning.

Authors

  • James Hill
    Department of Physical Therapy, University of Pittsburgh, Pittsburgh, PA.
  • Muhammad Rizwan Khokher
    CSIRO Data61, Marsfield, NSW, Australia.
  • Chuong Nguyen
    CSIRO Data61, Marsfield, NSW, Australia.
  • Matt Adcock
    CSIRO Data61, Marsfield, NSW, Australia.
  • Ron Li
    Division of Hospital Medicine, Stanford University, Stanford, California, United States.
  • Stuart Anderson
    School of Social and Political Science, University of Edinburgh, Edinburgh, United Kingdom.
  • Thomas Morrell
    Singular Health Group Limited, Subiaco, WA, Australia.
  • Tim Diprose
    Singular Health Group Limited, Subiaco, WA, Australia.
  • Olivier Salvado
    CSIRO Health and Biosecurity, The Australian e-Health & Research Centre, Herston, QLD, Australia. Electronic address: olivier.salvado@csiro.au.
  • Dadong Wang
    Quantitative Imaging, Data61 CSIRO, Sydney, NSW, Australia.
  • Guan K Tay
    Singular Health Group Limited, Subiaco, WA, Australia; Division of Psychiatry, UWA Medical School, The University of Western Australia, Perth, WA, Australia; School of Medical and Health Sciences, Edith Cowan University, Joondalup, WA, Australia; College of Medical and Health Sciences, Khalifa University of Science and Technology, Abu Dhabi, United Arab Emirates. Electronic address: guan.tay@uwa.edu.au.

Keywords

No keywords available for this article.