Single-distance, phase-only frequency-domain NIRS for vascular oxygenation and tissue metabolism: a Monte Carlo...

Frequency-domain near-infrared spectroscopy (FD-NIRS) currently enables absolute hemoglobin quantification but requires multidistance measurements of both amplitude attenuation and phase shifts. Notably, existing FD-NIRS approaches have not demonstrated reliable quantification of differential redox-state concentrations of cytochrome c oxidase ( ${\mathrm{CCO}}_{\text{redox}}$ ), a critical metabolic marker.

We aimed to develop a novel optimization-based algorithm for single source-detector (S-D), phase-only FD-NIRS that achieves accurate quantifications of hemoglobin parameters (HbO and Hb) and ${\mathrm{CCO}}_{\text{redox}}$ .

Our computational framework implemented both forward modeling and inverse reconstruction. For the modeling, we first defined chromophore concentration sets (HbO, Hb, ${\mathrm{CCO}}_{\text{redox}}$ ), followed by calculations of wavelength-dependent optical properties for two- or eight-wavelength configurations. Next, time-domain photon propagation was generated via Monte Carlo (MC) simulations, and FD-NIRS parameters (modulation amplitude, phase) were extracted through Fourier analysis. In the inverse computation, nonlinear optimization with edge-barrier regularization was employed for the recovery of chromophore concentrations. Both the multiseparation method and the single S-D, phase-only algorithm were used to reconstruct chromophore concentrations.

Respective performances evaluated for the two methods were compared through their concentration recovery accuracy. In either the two- or eight-wavelength configuration, our new algorithm outperformed the conventional method for the S-D separations up to 3 cm for all three chromophores. In particular, ${\mathrm{CCO}}_{\text{redox}}$ estimation was improved markedly from a mean relative error of 34.1% with the conventional method to just 5.1% using our algorithm.

These results validate single-separation phase-only FD-NIRS as an accurate method for multichromophore quantification (including ${\mathrm{CCO}}_{\text{redox}}$ ), enabling simpler, cost-effective systems without compromising metabolic imaging capability. The approach achieves $<10\%$ error in hemoglobin quantification while eliminating traditional multidistance requirements.