A method for determining venous contribution to BOLD contrast sensory activation.

Area of Science:

• Neuroimaging
• Biophysics
• Medical Physics

Background:

• Blood-oxygen-level-dependent (BOLD) contrast in fMRI reflects both neural activity and venous blood volume.
• Large veins contribute significantly to the BOLD signal, often with delayed responses, complicating the localization of neural activation.
• Identifying venous contributions is crucial for accurate interpretation of fMRI activation maps.

Purpose of the Study:

• To develop and validate a method for predicting the location of large veins using intrinsic properties of fMRI functional responses.
• To assess if functional response measures can reliably down-weight the venous contribution to BOLD-based activation maps.

Main Methods:

• Combined 3 Tesla fMRI with high-resolution anatomical imaging and MR venography.
• Analyzed functional response time courses using a gamma fit to extract measures like magnitude and delay.
• Employed logistic regression to discriminate between venous and gray matter tissue types based on functional response properties.

Main Results:

• Significant differences in mean response magnitude and delay were observed across tissue types (CSF, veins, gray matter, white matter).
• Veins showed distinct characteristics, with higher magnitude than gray matter and white matter, and delayed responses.
• A logistic regression model achieved 72% accuracy in discriminating veins from gray matter without prior macroscopic vessel information.

Conclusions:

• Functional response properties, specifically magnitude and delay, can be used to predict vein location in fMRI data.
• Weighting the T contrast by predicted probabilities effectively reduced the venous component in the activation map for the tested subject.
• This method offers a promising approach to improve the specificity of BOLD-based neuroimaging by accounting for venous signal contamination.