Premium
Improved precision in CHARMED assessment of white matter through sampling scheme optimization and model parsimony testing
Author(s) -
Santis S.,
Assaf Y.,
Evans C. J.,
Jones D. K.
Publication year - 2014
Publication title -
magnetic resonance in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.696
H-Index - 225
eISSN - 1522-2594
pISSN - 0740-3194
DOI - 10.1002/mrm.24717
Subject(s) - diffusion mri , computer science , sampling (signal processing) , monte carlo method , pipeline (software) , algorithm , tensor (intrinsic definition) , variance reduction , data acquisition , selection (genetic algorithm) , limiting , model selection , protocol (science) , reduction (mathematics) , white matter , diffusion , mathematical optimization , mathematics , artificial intelligence , statistics , magnetic resonance imaging , physics , alternative medicine , filter (signal processing) , pathology , engineering , operating system , geometry , computer vision , thermodynamics , radiology , programming language , medicine , mechanical engineering , pure mathematics
Purpose The composite hindered and restricted model of diffusion provides microstructural indices that are potentially more specific than those from diffusion tensor imaging. However, in comparison to diffusion tensor imaging, the acquisition time is longer, limiting clinical applications. Moreover, the model requires several parameters to be estimated whose confidence intervals can be large. Here, the composite hindered and restricted model of diffusion acquisition and data processing pipelines are optimized to extend the utility of this approach. Methods A multishell sampling scheme was optimized using the electrostatic repulsion algorithm, combined with optimal ordering. The optimal protocol, using as few measurements as possible, was determined through leave‐ n ‐out analyses. Parsimonious model selection criteria were used to select between nested models, comprising up to three restricted compartments. The schemes were evaluated using both through Monte‐Carlo simulations and in vivo data. Results The optimization/model selection procedure resulted in increased accuracy and precision on the estimated parameters, allowing for a reduction in acquisition time and marked improvements in data quality. The final protocol provided whole brain coverage data in only 12 min. Conclusion Through careful optimization of the acquisition and analysis pipeline for the composite hindered and restricted model of diffusion, it is possible to reduce acquisition time for whole brain datasets to a time that is clinically applicable. Magn Reson Med 71:661–671, 2014. © 2013 Wiley Periodicals, Inc.