MultiAtlas Library for Eliminating Normalization

Authors: Joseph A Maldjian Carol A Shively Michael A Nader David P Friedman Christopher T Whitlow

Publish Date: 2015/12/07

Volume: 14, Issue: 2, Pages: 183-190

PDF Link

Abstract

Current tools for automated skull stripping normalization and segmentation of nonhuman primate NHP brain MRI studies typically demonstrate high failure rates Many of these failures are due to a poor initial estimate for the affine component of the transformation The purpose of this study is to introduce a multiatlas approach to overcome these limitations and drive the failure rate to near zero A library of studyspecific templates SST spanning three Old World primate species Macaca fascicularis M mulatta Chlorocebus aethiops was created using a previously described unbiased automated approach Several modifications were introduced to the methodology to improve initial affine estimation at the studyspecific template level and at the individual subject level These involve performing multiple separate normalizations to a multiatlas library of templates and selecting the best performing template on the basis of a covariance similarity metric This template was then used as an initialization for the affine component of subsequent skull stripping and normalization procedures Normalization failure rate for SST generation and individualsubject segmentation on a set of 150 NHP was evaluated on the basis of visual inspection The previous automated template creation procedure results in excellent skull stripping segmentation and atlas labeling across species Failure rate at the individualsubject level was approximately 1  however at the SST generation level it was 17  Using the new multiatlas approach failure rate was further reduced to zero for both SST generation and individual subject processing We describe a multiatlas library registration approach for driving normalization failures in NHP to zero It is straightforward to implement and can have application to a wide variety of existing tools as well as in difficult populations including neonates and the elderly This approach is also an important step towards developing fully automated highthroughput processing pipelines that are critical for future high volume multicenter NHP imaging studies for studies of drug abuse and brain healthThis study was supported in part by R01DA025120 R37DA010584 PI MAN R01HL087103 PI CAS and AA014106 PI DPF Support for this research was also provided by NIH grants NS0075107 and NS082453 PIJAM The authors would also like to thank Ben Wagner for programming assistance and the Center for Biomolecular Imaging

Keywords: