2014 Summer Project Week:MR-Ultrasound Registration for Prostate Interventions

From NAMIC Wiki
Jump to: navigation, search
Home < 2014 Summer Project Week:MR-Ultrasound Registration for Prostate Interventions

Key Investigators

  • Andriy Fedorov, SPL
  • Chenxi Zhang, SPL/FudanU
  • Andras Lasso, Queens
  • Steve Pieper, Isomics

Project Description

Prostate cancer is a kind of common cause of cancer in men in many countries. Transrectal ultrasound (TRUS) is the most commonly used modality for guiding needle biopsies and minimally-invasive surgical interventions for the treatment of prostate cancer. However,it is difficult to accurately distinguish tumors in B-mode TRUS images. Magnetic resonance (MR) is a very promising technique for prostate biopsies and prostate interventions. Disadvantages of MR-guided prostate biopsies and prostate interventions are that MRI is an expensive method and not widely available. MR-Ultrasound registration can improve TRUS imaging and take advantages of both modalities. The object of this project it to develop a method of MR-Ultrasound image registration for prostate intervention.


  • Investigate a method of MR-Ultrasound images registration for prostate intervention.

Approach, Plan

  • Continue development and evaluation of a method based on non-rigid registration of distance maps of the gland segmentation
    • Extract prostate from MR image and Ultrasound image
    • Calculate distance maps for segmented MR image and Ultrasound image
    • Do registration for distance maps of MR image and Ultrasound image using Slicer module of "BRAINSFit"
  • Investigate alternative approaches available within the group
  • We made a test dataset publicly available. If you have a method that works on this dataset, let us know. We have expert annotations that we can use to calculate registration error and evaluate your approach.


  • Presented this registration problem at the Prostate IGT breakout session
  • Discussed alternative approaches: applied LandmarkRegistration module of Slicer for manually guided registration
  • Steve added sample dataset has been added to the Slicer SampleData module