Open AccessA modality‐independent approach to spatial normalization of tomographic images of the human brain
Author(s) -
Lancaster Jack L.,
Glass Thomas G.,
Lankipalli Bhujanga R.,
Downs Hunter,
Mayberg Helen,
Fox Peter T.
Publication year - 1995
Publication title -
human brain mapping
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.005
H-Index - 191
eISSN - 1097-0193
pISSN - 1065-9471
DOI - 10.1002/hbm.460030305
Subject(s) - spatial normalization , normalization (sociology) , artificial intelligence , positron emission tomography , tomography , image resolution , computer vision , computer science , affine transformation , software , pattern recognition (psychology) , nuclear medicine , physics , mathematics , optics , voxel , medicine , geometry , sociology , anthropology , programming language
A modality‐independent approch for interactive spatial normalization of tomographic images of the human brain is described and its performance evaluated. Spatial normalization is accomplished using a nine‐parameter affine transformation to interactively align and adjust the shape of a subject brain to the reference brain detailed in the 1988 atlas of Talairach et al. A user‐friendly software application was developed using the X‐windows Motif environment to guide the user through this process. This software supports data types from a wide variety of tomographic imagers and produces output in spatially concise formats. The parameters used for spatial alignment and shape normalization are presented and methods to apply them discussed. Where normalization parameters cannot be obtained directly from the image, as with positron emission tomography (PET), methods for estimating them are given. Evaluation of a new four‐landmark method to fit the AC‐PC line in 16 magnetic resonance imaging (MRI) studies indicated an average difference assessed as the distance between the true and fitted AC‐PC line at four locations of 0.82 mm when using a 2‐D weighted fit. The same landmarks were evaluated using lower spatial resolution PET‐like images simulated from the 16 MRI studies. The difference between the PET and MR image volumes following alignment was minimal, with mean rotational differences of less than 0.2 deg and mean translational differences of generally less than 2 mm. Spatial normalization is illustrated for single photon emission computed tomography (SPECT), X‐ray computed tomography (CT), PET, and MR image volumes. Modality‐independent spatial normalization can be consistently and reliably performed with the methods and software presented. © 1995 Wiley‐Liss, Inc.