Evaluation of low-dose limits in 3D-2D rigid registration for surgical guidance

A. Uneri, A. S. Wang, Y. Otake, G. Kleinszig, S. Vogt, A. J. Khanna, G. L. Gallia, Z. L. Gokaslan, J. H. Siewerdsen

Research output: Contribution to journalArticle

Abstract

An algorithm for intensity-based 3D-2D registration of CT and C-arm fluoroscopy is evaluated for use in surgical guidance, specifically considering the low-dose limits of the fluoroscopic x-ray projections. The registration method is based on a framework using the covariance matrix adaptation evolution strategy (CMA-ES) to identify the 3D patient pose that maximizes the gradient information similarity metric. Registration performance was evaluated in an anthropomorphic head phantom emulating intracranial neurosurgery, using target registration error (TRE) to characterize accuracy and robustness in terms of 95% confidence upper bound in comparison to that of an infrared surgical tracking system. Three clinical scenarios were considered: (1) single-view image + guidance, wherein a single x-ray projection is used for visualization and 3D-2D guidance; (2) dual-view image + guidance, wherein one projection is acquired for visualization, combined with a second (lower-dose) projection acquired at a different C-arm angle for 3D-2D guidance; and (3) dual-view guidance, wherein both projections are acquired at low dose for the purpose of 3D-2D guidance alone (not visualization). In each case, registration accuracy was evaluated as a function of the entrance surface dose associated with the projection view(s). Results indicate that images acquired at a dose as low as 4 μGy (approximately one-tenth the dose of a typical fluoroscopic frame) were sufficient to provide TRE comparable or superior to that of conventional surgical tracking, allowing 3D-2D guidance at a level of dose that is at most 10% greater than conventional fluoroscopy (scenario #2) and potentially reducing the dose to approximately 20% of the level in a conventional fluoroscopically guided procedure (scenario #3).

Original languageEnglish (US)
Pages (from-to)5329-5345
Number of pages17
JournalPhysics in medicine and biology
Volume59
Issue number18
DOIs
StatePublished - Sep 21 2014

    Fingerprint

Keywords

  • 3D-2D registration
  • dose
  • image-guided surgery
  • neurosurgery
  • surgical navigation
  • x-ray fluoroscopy

ASJC Scopus subject areas

  • Radiological and Ultrasound Technology
  • Radiology Nuclear Medicine and imaging

Cite this