Investigation of BOLD signal dependence on cerebral blood flow and oxygen consumption: The deoxyhemoglobin dilution model

• Published in: Magnetic resonance in medicine 1999-11, Vol.42 (5), p.849-863
• Main Authors: Hoge, Richard D., Atkinson, Jeff, Gill, Brad, Crelier, Gérard R., Marrett, Sean, Pike, G. Bruce
• Format: Article
• Language: English
• Subjects: aerobic metabolism; BOLD contrast; Brain Mapping - methods; Cerebrovascular Circulation; Hemoglobins - metabolism; Hemoglobins - physiology; Humans; hypercapnia; Hypercapnia - diagnosis; Hypercapnia - physiopathology; Image Enhancement - methods; Magnetic Resonance Imaging; Models, Biological; Oxygen - blood; Oxygen Consumption; perfusion; Photic Stimulation; Sensitivity and Specificity; Visual Cortex - anatomy & histology; Visual Cortex - physiopathology
• ISSN: 0740-3194; 1522-2594
• DOI: 10.1002/(SICI)1522-2594(199911)42:5<849::AID-MRM4>3.0.CO;2-Z

The relationship between blood oxygenation level‐dependent (BOLD) MRI signals, cerebral blood flow (CBF), and oxygen consumption (CMRO2) in the physiological steady state was investigated. A quantitative model, based on flow‐dependent dilution of metabolically generated deoxyhemoglobin, was validated by measuring BOLD signals and relative CBF simultaneously in the primary visual cortex (V1) of human subjects (N = 12) during graded hypercapnia at different levels of visual stimulation. BOLD and CBF responses to specific conditions were averaged across subjects and plotted as points in the BOLD‐CBF plane, tracing out lines of constant CMRO2. The quantitative deoxyhemoglobin dilution model could be fit to these measured iso‐CMRO2 contours without significant (P ≤ 0.05) residual error and yielded MRI‐based CMRO2 measurements that were in agreement with PET results for equivalent stimuli. BOLD and CBF data acquired during graded visual stimulation were then substituted into the model with constant parameters varied over plausible ranges. Relative changes in CBF and CMRO2 appeared to be coupled in an approximate ratio of ∼2:1 for all realistic parameter settings. Magn Reson Med 42:849–863, 1999. © 1999 Wiley‐Liss, Inc.