Cortical surface registration using spherical thin-plate spline with sulcal lines and mean curvature as features

Hyunjin Park, Jun Sung Park, Joon Kyung Seong, Duk L. Na, Jong Min Lee

Research output: Contribution to journalArticlepeer-review

13 Citations (Scopus)

Abstract

Analysis of cortical patterns requires accurate cortical surface registration. Many researchers map the cortical surface onto a unit sphere and perform registration of two images defined on the unit sphere. Here we have developed a novel registration framework for the cortical surface based on spherical thin-plate splines. Small-scale composition of spherical thin-plate splines was used as the geometric interpolant to avoid folding in the geometric transform. Using an automatic algorithm based on anisotropic skeletons, we extracted seven sulcal lines, which we then incorporated as landmark information. Mean curvature was chosen as an additional feature for matching between spherical maps. We employed a two-term cost function to encourage matching of both sulcal lines and the mean curvature between the spherical maps. Application of our registration framework to fifty pairwise registrations of T1-weighted MRI scans resulted in improved registration accuracy, which was computed from sulcal lines. Our registration approach was tested as an additional procedure to improve an existing surface registration algorithm. Our registration framework maintained an accurate registration over the sulcal lines while significantly increasing the cross-correlation of mean curvature between the spherical maps being registered.

Original languageEnglish
Pages (from-to)46-53
Number of pages8
JournalJournal of Neuroscience Methods
Volume206
Issue number1
DOIs
Publication statusPublished - 2012 Apr 30
Externally publishedYes

Keywords

  • Cortical surface
  • Mean curvature
  • Spherical thin-plate splines
  • Sulcal lines
  • Surface registration

ASJC Scopus subject areas

  • Neuroscience(all)

Fingerprint

Dive into the research topics of 'Cortical surface registration using spherical thin-plate spline with sulcal lines and mean curvature as features'. Together they form a unique fingerprint.

Cite this