Predicting Regional Cerebral Blood Flow Using Voxel-Wise Resting-State Functional MRI

Area of Science:

• Neuroimaging
• Neuroscience
• Biomarkers

Background:

• Regional cerebral blood flow (rCBF) is a potential biomarker for neuropsychiatric disorders like major depressive disorder (MDD).
• Current methods for measuring rCBF can be resource-intensive.

Purpose of the Study:

• To develop and validate a method for predicting rCBF from resting-state functional MRI (rsfMRI).
• To assess the utility of rsfMRI-derived rCBF in identifying MDD-related cerebral changes.

Main Methods:

• rsfMRI data were analyzed using a support vector machine algorithm to predict voxel-wise rCBF, correcting for partial volume averaging (PVA) artifacts.
• The method was validated using three independent datasets (Amish Connectome Project, UK Biobank, Amen Clinics Inc.) with ASL and SPECT data.

Main Results:

• PVA-corrected rCBF predicted from rsfMRI showed significant correlations with ASL-measured rCBF.
• Significant regional cerebral blood flow deficits were identified in the MDD group within the UK Biobank dataset.
• The pattern of MDD-related hypoperfusion from rsfMRI showed high agreement with SPECT findings.

Conclusions:

• Cerebral blood flow (CBF) can be reliably computed from widely available rsfMRI data.
• This rsfMRI-based approach provides a scalable method for investigating cerebral neurophysiology in neuropsychiatric disorders like MDD.