Understanding bone responses in B-mode ultrasound images and automatic bone surface extraction using a Bayesian probabilistic framework

Ameet Kumar Jain, Russell H. Taylor

Research output: Contribution to journalConference articlepeer-review

59 Scopus citations


The registration of preoperative CT to intra-operative reality systems is a crucial step in Computer Assisted Orthopedic Surgery (CAOS). The intra-operative sensors include 3D digitizers, fiducials, X-rays and Ultrasound (US). Although US has many advantages over others, tracked US for Orthopedic Surgery has been researched by only a few authors. An important factor limiting the accuracy of tracked US to CT registration (1-3mm) has been the difficulty in determining the exact location of the bone surfaces in the US images (the response could range from 2-4mm). Thus it is crucial to localize the bone surface accurately from these images. Moreover conventional US imaging systems are known to have certain inherent inaccuracies, mainly due to the fact that the imaging model is assumed planar. This creates the need to develop a bone segmentation framework that can couple information from various post-processed spatially separated US images (of the bone) to enhance the localization of the bone surface. In this paper we discuss the various reasons that cause inherent uncertainties in the bone surface localization (in Bmode US images) and suggest methods to account for these. We also develop a method for automatic bone surface detection. To do so, we account objectively for the high-level understanding of the various bone surface features visible in typical US images. A combination of these features would finally decide the surface position. We use a Bayesian probabilistic framework, which strikes a fair balance between high level understanding from features in an image and the low level number crunching of standard image processing techniques. It also provides us with a mathematical approach that facilitates combining multiple images to augment the bone surface estimate.

Original languageEnglish (US)
Article number16
Pages (from-to)131-142
Number of pages12
JournalProgress in Biomedical Optics and Imaging - Proceedings of SPIE
Issue number27
StatePublished - 2004
EventMedical Imaging 2004 - Ultrasonic Imaging and Signal Processing - San Diego, CA, United States
Duration: Feb 18 2004Feb 19 2004


  • Automatic
  • Bone
  • CT
  • Image Registration
  • Segmentation
  • Ultrasound

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Biomaterials
  • Atomic and Molecular Physics, and Optics
  • Radiology Nuclear Medicine and imaging


Dive into the research topics of 'Understanding bone responses in B-mode ultrasound images and automatic bone surface extraction using a Bayesian probabilistic framework'. Together they form a unique fingerprint.

Cite this