Melanin Bias in Cerebral fNIRS: A Monte Carlo Study Toward Equitable Optical Brain Sensing

Functional near-infrared spectroscopy (fNIRS) is an optical biosensing modality that transduces cortical hemodynamics into a measurable near-infrared signal, underpinning wearable brain monitors and brain–computer interfaces (BCIs). Because epidermal melanin is a strong near-infrared absorber, device sensitivity may depend on skin tone, raising an equity concern for optical neural biosensors. (1) Objective: We quantify, by Monte Carlo simulation, how epidermal melanin biases the cerebral sensitivity of continuous-wave fNIRS biosensors across the Fitzpatrick scale, and evaluate source-wavelength selection as a device-level mitigation. (2) Methods: Using a physically sourced melanin model and a two-stage design—an idealized layered slab for controlled sweeps and the Colin27 anatomical atlas for realistic-geometry confirmation—we computed gray-matter sensitivity for three skin tones, four wavelengths, and seven source–detector separations. (3) Results: Melanin acts as a surface gate that lowers absolute cerebral sensitivity by up to ~12.5-fold (at 690 nm) for dark skin—and by ~5-fold at the 850 nm optimum—while preserving depth selectivity; longer wavelengths reduce the penalty (dark-skin retained sensitivity rising from 8.0% at 690 nm to 29.6% at 940 nm) but never remove it; and 850 nm is a practical optimum for single-wavelength cerebral sensitivity, combining the highest absolute sensitivity with improved equity, whereas 940 nm gains relative equity only as absolute signal collapses through water absorption. Atlas simulations reproduce the 850 nm optimum in realistic anatomy under identical optical assumptions and show that idealized slabs overestimate sensitivity ~50-fold and that sensitivity varies ~19-fold with probe placement. (4) Conclusions: Skin tone and anatomy compound to disadvantage darker-skinned users of optical neural biosensors; wavelength selection and a sourced correction-factor framework offer concrete routes toward equitable device design.