Premium
Calculating potential error in sodium MRI with respect to the analysis of small objects
Author(s) -
Stobbe Robert W.,
Beaulieu Christian
Publication year - 2018
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.26962
Subject(s) - imaging phantom , voxel , observational error , magnetic resonance imaging , nuclear magnetic resonance , signal (programming language) , physics , noise (video) , nuclear medicine , mathematics , computational physics , algorithm , computer science , statistics , optics , medicine , artificial intelligence , radiology , image (mathematics) , programming language
Purpose To facilitate correct interpretation of sodium MRI measurements, calculation of error with respect to rapid signal decay is introduced and combined with that of spatially correlated noise to assess volume‐of‐interest (VOI) 23 Na signal measurement inaccuracies, particularly for small objects. Methods Noise and signal decay–related error calculations were verified using twisted projection imaging and a specially designed phantom with different sized spheres of constant elevated sodium concentration. As a demonstration, lesion signal measurement variation (5 multiple sclerosis participants) was compared with that predicted from calculation. Results Both theory and phantom experiment showed that VOI signal measurement in a large 10‐mL, 314‐voxel sphere was 20% less than expected on account of point‐spread‐function smearing when the VOI was drawn to include the full sphere. Volume‐of‐interest contraction reduced this error but increased noise‐related error. Errors were even greater for smaller spheres (40–60% less than expected for a 0.35‐mL, 11‐voxel sphere). Image‐intensity VOI measurements varied and increased with multiple sclerosis lesion size in a manner similar to that predicted from theory. Correlation suggests large underestimation of 23 Na signal in small lesions. Conclusions Acquisition‐specific measurement error calculation aids 23 Na MRI data analysis and highlights the limitations of current low‐resolution methodologies. Magn Reson Med 79:2968–2977, 2018. © 2017 International Society for Magnetic Resonance in Medicine.