2015 Summer Project Week:LinearFeatureRegistration

• 

  Projects List

• 

  Screenshot of LinearFeatureRegistration module

Key Investigators

• Matthew Holden
• Nicole Aucoin, BWH (fiducials)
• Eugenio Marinetto, UC3M Madrid (Line-Based Registration, Reg. Algorithm)

Project Description

Objective

• Add module for registration using linear features

Approach, Plan

• Investigate module use cases:
  • Ultrasound calibration for PLUS
  • ...
• Use Fiducial Registration Wizard module as a template
• Investigate whether linear features can be treated in the same way as fiducials

Progress

• See: LinearObjectRegistration
• Discussed generic module for collecting/defining/acquiring parametric features:
  • Use data structure to store a set of 0D/1D/2D parametric features
  • For each feature, store raw point data, as well as basic information (i.e. centroid, variance, etc.)
  • Individual registration algorithms are implemented as modules using the data structure
  • Individual algorithm is responsible for converting 0D/1D/2D data to points/lines/planes (if applicable)
  • Methods for acquisition: tracking, models, fiducials, segmentation
• Use cases:
  • Phantom registration for ultrasound calibration
  • Line feature based image to tracker registration
  • Fiducial frame registration
• Began implementation of generic parametric features