Automated Image Registration: II. Intersubject Validation of Linear and Nonlinear Models

TLDR

The study aimed to validate automated linear and nonlinear intersubject image registration using AIR 3.0 based on voxel intensities. The authors registered PET and MRI scans from 22 subjects to averaged brain atlases with multiple linear and nonlinear transformation models via an automated algorithm, validating accuracy with anatomically defined landmarks. Automated registration outperformed manual Talairach, with higher degrees of freedom improving accuracy; nonlinear models provided better alignment than linear ones but were slower, demonstrating practical and more precise intersubject registration.

Abstract

Purpose Our goal was to validate linear and nonlinear intersubject image registration using an automated method (AIR 3.0) based on voxel intensity. Method PET and MRI data from 22 normal subjects were registered to corresponding averaged PET or MRI brain atlases using several specific linear and nonlinear spatial transformation models with an automated algorithm. Validation was based on anatomically defined landmarks. Results Automated registration produced results that were superior to a manual nine parameter variant of the Talairach registration method. Increasing the degrees of freedom in the spatial transformation model improved the accuracy of automated intersubject registration. Conclusion Linear or nonlinear automated intersubject registration based on voxel intensities is computationally practical and produces more accurate alignment of homologous landmarks than manual nine parameter Talairach registration. Nonlinear models provide better registration than linear models but are slower.