August 2014
Volume 14, Issue 10
Free
Vision Sciences Society Annual Meeting Abstract  |   August 2014
Locally-Optimized Inter-Subject Alignment of Functional Cortical Regions
Author Affiliations
  • Marius Cătălin Iordan
    Department of Computer Science, Stanford University
  • Armand Joulin
    Department of Computer Science, Stanford University
  • Diane M. Beck
    Beckman Institute and Department of Psychology, University of Illinois at Urbana-Champaign
  • Li Fei-Fei
    Department of Computer Science, Stanford University
Journal of Vision August 2014, Vol.14, 714. doi:https://doi.org/10.1167/14.10.714
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Marius Cătălin Iordan, Armand Joulin, Diane M. Beck, Li Fei-Fei; Locally-Optimized Inter-Subject Alignment of Functional Cortical Regions. Journal of Vision 2014;14(10):714. https://doi.org/10.1167/14.10.714.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Inter-subject cortical registration is necessary in functional imaging (fMRI) studies for making inferences about equivalent brain function across a population. Most state-of-the-art alignment methods attempt to preserve anatomical landmarks, cortical curvature, or functional connectivity between cortical volumes (Yeo, 2010; Sabuncu, 2010; Conroy, 2013). However, these methods have difficulty aligning high-level visual areas across subjects, mainly due to large variability in anatomical position. Consequently, we propose a locally optimized registration method that directly predicts the location of a seed region of interest (ROI) on a separate target cortical sheet by maximizing the correlation between voxel-level functional responses in the two maps. A key advantage of our method is allowing for non-smooth local deformations in the mapping. We reason that peak functional contrast points (where ROIs are centered) share similar function between subjects, yet functional gradients of selectivity surrounding the peaks may not be spatially organized identically across subjects (Huth, 2012). We test our method by aligning a difficult to match, functionally defined, object-selective ROI (lateral occipital complex, LOC) between subjects using a passive-viewing fMRI experiment where participants were shown 1,024 images of objects from 32 categories. Our method vastly outperforms two canonical baselines (anatomical-landmark-based AFNI alignment and cortical-curvature-based FreeSurfer alignment) in overlap percentage between predicted region and ground truth LOC (i.e. defined via standard localizer procedures): baselines 10-11%, ours 24%. Furthermore, our predicted maps are more consistent across subjects than both baselines (overlap of region commonly mapped from 3+ subjects: baselines 9-11%, ours 26%). Therefore, our technique improves the quality and reliability of matching and transferring the location of functional ROIs across subjects, an important step towards obviating the need for additional or impossible to obtain localizer scans. Moreover, our method can be used to investigate the complex relationship between anatomy, functional contrast peak (ground-truth ROI), and cortical computation (BOLD response).

Meeting abstract presented at VSS 2014

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×