Characterization of BOLD activation in multi-echo fMRI data using fuzzy cluster analysis and a comparison with quantitative modeling

• Published in: NMR in biomedicine Vol. 14; no. 7-8; pp. 484 - 489
• Main Authors: Barth, Markus, Windischberger, Christian, Klarhöfer, Markus, Moser, Ewald
• Format: Journal Article Conference Proceeding
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 01.11.2001; Wiley
• Subjects: Biological and medical sciences; BOLD; Brain - metabolism; Cluster Analysis; EDA; functional MRI; fuzzy clustering; Fuzzy sets; high field; human brain; Humans; Image analysis; Investigative techniques, diagnostic techniques (general aspects); Magnetic Resonance Imaging; Mathematical models; Medical imaging; Medical sciences; motor cortex; multiple gradient-echo; Nervous system; Oscillations; Oxygen - blood; Radiodiagnosis. Nmr imagery. Nmr spectrometry; spiral imaging; Cluster analysis; Human; Central nervous system; Nuclear magnetic resonance imaging; Modeling; Gradient echo; Image analysis; Fuzzy algorithm; Motor cortex; High field; Oxygenation; Comparative study; Quantitative analysis; Brain (vertebrata)
• ISSN: 0952-3480; 1099-1492
• DOI: 10.1002/nbm.737

A combination of multiple gradient‐echo imaging and exploratory data analysis (EDA), i.e. fuzzy cluster analysis (FCA), is proposed for separation and characterization of BOLD activation in single‐shot spiral functional magnetic resonance imaging (fMRI) experiments at 3 T. Differentiation of functional activation using FCA is performed by clustering pixel signal changes (ΔS) as a function of echo time (TE). Further vascular classification is supported by the localization of activation and the comparison with a single‐exponential decay model. In some subjects, an additional indication for large vessels within a voxel was found as oscillation of the fMRI signal difference vs echo time (TE). Such large vessels may be separated from small vessel activation and, therefore, our proposed procedure might prove useful if a more specific functional localization is desired in fMRI. In addition to the signal change ΔS, ΔT2*/T2* is significantly different between activated regions. Averaged over all eight subjects ΔT2* is 1.7 ± 0.2 ms in ROIs with the highest signal change characterized as containing large vessels, whereas in ROIs corresponding to microvascular environment average ΔT2* values are 0.8 ± 0.1 ms. Copyright © 2001 John Wiley & Sons, Ltd.