Open AccessSingle‐subject independent component analysis‐based intensity normalization in non‐quantitative multi‐modal structural MRI
Author(s) -
Papazoglou Sebastian,
Würfel Jens,
Paul Friedemann,
Brandt Alexander U.,
Scheel Michael
Publication year - 2017
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.23615
Subject(s) - artificial intelligence , computer science , pattern recognition (psychology) , normalization (sociology) , independent component analysis , diffusion mri , unobservable , principal component analysis , statistics , mathematics , magnetic resonance imaging , econometrics , medicine , sociology , anthropology , radiology
Non‐quantitative MRI is prone to intersubject intensity variation rendering signal intensity level based analyses limited. Here, we propose a method that fuses non‐quantitative routine T1‐weighted (T1w), T2w, and T2w fluid‐saturated inversion recovery sequences using independent component analysis and validate it on age and sex matched healthy controls. The proposed method leads to consistent and independent components with a significantly reduced coefficient‐of‐variation across subjects, suggesting potential to serve as automatic intensity normalization and thus to enhance the power of intensity based statistical analyses. To exemplify this, we show that voxelwise statistical testing on single‐subject independent components reveals in particular a widespread sex difference in white matter, which was previously shown using, for example, diffusion tensor imaging but unobservable in the native MRI contrasts. In conclusion, our study shows that single‐subject independent component analysis can be applied to routine sequences, thereby enhancing comparability in‐between subjects. Unlike quantitative MRI, which requires specific sequences during acquisition, our method is applicable to existing MRI data. Hum Brain Mapp 38:3615–3622, 2017 . © 2017 Wiley Periodicals, Inc.