Publication Info

• Authors
  Peter Brößner
  B. Hohlmann
  K. Radermacher 
• Conference
  BVM
  
• Publisher
  Springer
• Date
  2021
• Link
  https://link.springer.com/chapter/10.1007/978-3-658-33198-6_8
• PDF „BVM 2021“

Abstract

For minimally-invasive surgery of the scaphoid, navigation based on ultrasound images instead of fuoroscopy could reduce costs as well as prevent exposure to ionizing radiation. We present a machine learning based two-stage approach that tackles the tasks of image segmentation and point cloud registration individually. For this, Deeplabv3+ as well as the PRNet architecture were trained on two newly generated datasets. An evaluation on in-vitro data results in an average surface distance error of 1.1mm and a mean rotational deviation of 6:2 with a processing time of 9 seconds. We conclude that near real-time navigation is feasible.

Intra-operative procedure for the registration of therapeutical object and virtual object: the US image slices acquired by a 3D US probe are fed into a segmentation network. The resulting masks are thinned and sampled to a partial point set, which is registered to the point set obtained from CT.