Weighted local mutual information for 2D-3D registration in vascular interventions

  • Cai Meng*
  • , Qi Wang
  • , Shaoya Guan
  • , Yi Xie
  • *Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In this paper, a new similarity measure, WLMI (Weighted Local Mutual Information), based on weighted patch and mutual information is proposed to register the preoperative 3D CT model to the intra-operative 2D X-ray images in vascular interventions. We embed this metric into the 2D-3D registration framework, where we show that the robustness and accuracy of the registration can be effectively improved by adapting the strategy of local image patch selection and the weighted joint distribution calculation based on gradient. Experiments on both synthetic and real X-ray image registration show that the proposed method produces considerably better registration results in a shorter time compared with the conventional MI and Normalized MI methods.

Original languageEnglish
Title of host publicationStructural, Syntactic, and Statistical Pattern Recognition - Joint IAPR International Workshop, S+SSPR 2018, Proceedings
EditorsEdwin R. Hancock, Tin Kam Ho, Battista Biggio, Richard C. Wilson, Antonio Robles-Kelly, Xiao Bai
PublisherSpringer Verlag
Pages376-385
Number of pages10
ISBN (Print)9783319977843
DOIs
StatePublished - 2018
EventJoint IAPR International Workshops on Structural and Syntactic Pattern Recognition, SSPR 2018 and Statistical Techniques in Pattern Recognition, SPR 2018 - Beijing, China
Duration: 17 Aug 201819 Aug 2018

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11004 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceJoint IAPR International Workshops on Structural and Syntactic Pattern Recognition, SSPR 2018 and Statistical Techniques in Pattern Recognition, SPR 2018
Country/TerritoryChina
CityBeijing
Period17/08/1819/08/18

Keywords

  • 2D-3D registration
  • Gradient weighted
  • Local patch
  • Mutual information

Fingerprint

Dive into the research topics of 'Weighted local mutual information for 2D-3D registration in vascular interventions'. Together they form a unique fingerprint.

Cite this