Premium
Intensity Inhomogeneity Correction Using N3 on Mouse Brain Magnetic Resonance Microscopy
Author(s) -
Lin Lan,
Wu Shuicai,
Bin Guangyu,
Yang Chunlan
Publication year - 2013
Publication title -
journal of neuroimaging
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.822
H-Index - 64
eISSN - 1552-6569
pISSN - 1051-2284
DOI - 10.1111/jon.12041
Subject(s) - bonferroni correction , spline (mechanical) , segmentation , coefficient of variation , nonparametric statistics , neuroimaging , magnetic resonance microscopy , white matter , multiple comparisons problem , magnetic resonance imaging , artificial intelligence , algorithm , computer science , medicine , pattern recognition (psychology) , statistics , mathematics , physics , spin echo , radiology , psychiatry , thermodynamics
BACKGROUND AND PURPOSE Small animal neuroimaging using magnetic resonance microscopy (MRM) has evolved significantly from understanding of imaging physics to widely use today as an important tool in computational neuroanatomy, while how to get optimal inhomogeneity correction for inhomogeneous mouse brain MRM has been given less attention. METHODS This present study investigates the ability of fine‐tuning the nonparametric nonuniform intensity normalization (N3) technique to get optimal inhomogeneity correction of mouse brain MRM. Six mice were scanned on a 7‐T scanner with a phased array surface coil of four elements. The N3 parameters such as stopping criteria, maximum iterations, down‐sampling ratio, full width at half maximum, spline distance, and brain mask have been tuned to get optimal correction result. We used coefficient of variation of the white matter and joint variation to ascertain quantitatively the correction. The data were analyzed by two‐way repeated measures analysis of variance and Bonferroni post hoc test. RESULTS The quantitative outcomes show that brain mask and spline distance have a significant influence on correcting performance. CONCLUSIONS The present study demonstrates the benefit of reducing the spline distance values to 25 and tighter mask. The finding can help researches to enhance precision in studies where mouse MRM need further registration or segmentation.