Revisiting equivalent optical properties for cerebrospinal fluid to improve diffusion-based modeling accuracy in the...

Area of Science:

• Biomedical Optics
• Neuroimaging
• Computational Modeling

Background:

• Diffusion approximation (DA) is widely used in functional near-infrared spectroscopy (fNIRS) for brain imaging.
• Existing DA models have limitations due to cerebrospinal fluid (CSF) optical properties, often using empirically derived values.
• Previous recommendations for CSF optical properties were not optimized for minimizing DA modeling errors.

Purpose of the Study:

• To directly quantify the accuracy of DA solutions in brain models by comparing them to gold-standard mesh-based Monte Carlo (MMC) solutions.
• To derive updated recommendations for CSF optical properties to improve DA model accuracy.
• To minimize errors in fNIRS analysis caused by CSF optical property assumptions.

Main Methods:

• DA solutions were generated for 5-layer head and Colin27 atlas models.
• CSF absorption and reduced scattering coefficients were independently swept to obtain DA solutions.
• Errors were computed by comparing DA solutions to MMC solutions using literature CSF optical properties as a reference.
• Optimized CSF optical properties were identified by minimizing errors in surface fluence, total brain sensitivity, and brain energy-deposition.

Main Results:

• Previously recommended CSF properties led to significant errors (8.7%–52%) in tested metrics.
• Infinite solutions were found to match DA and MMC for single metrics by sweeping CSF absorption and scattering coefficients.
• Simultaneous minimization of multiple metrics at various source/detector separations identified optimized settings.
• A new recommendation for CSF optical properties was derived: setting absorption to a specific value while maintaining physiological scattering.

Conclusions:

• The new recommended CSF optical properties significantly reduce model mismatches between DA and MMC solutions across multiple metrics.
• These updated properties improve fNIRS analysis accuracy without sacrificing computational speed.
• It is possible to eliminate DA-MMC mismatches for specific metrics of interest by adjusting CSF optical properties.