Micro-CT-informed Monte Carlo and experimental dosimetry of transcranial photobiomodulation devices

Transcranial photobiomodulation (tPBM) is being investigated as a noninvasive strategy for modulating neural metabolism, neurovascular signaling, inflammation, and network activity; however, the field still lacks a quantitative framework that links nominal device parameters to delivered intracranial dose. Here, we combined micro-CT-informed Monte Carlo light transport with ex vivo phantom measurements to determine how wavelength, skull anatomy, cerebrospinal fluid (CSF) thickness, commercial device design, delivery route, and thermal behavior govern tPBM dosimetry. Two human skull models were embedded in four-compartment scalp-skull-blood-filled diploic pore-CSF models and simulated at 23 wavelengths from 610 to 1267 nm using an LED-like source geometry. Experimental benchmarking used a porcine skin-human skull-artificial CSF gelatin phantom measured regionally with a spectroradiometer for three commercial helmet devices, complemented by intranasal-style 810 nm irradiance mapping, brain-gelatin transmission measurements, and infrared thermal imaging. Monte Carlo results showed that transmitted photon weight was highly sensitive to skull thickness, CSF depth, and diploic microstructure, and generally favored the 1064–1070 nm range under controlled source-detector conditions, whereas 1267 nm decreased due to increased long-wavelength absorption. In contrast, commercial-device phantom measurements showed that delivered post-stack output was not determined by wavelength alone: the 810 nm helmet produced substantially higher region-resolved transmittance than the two longer-wavelength devices, while a paired 810 versus 1070 nm comparison showed higher instantaneous post-stack irradiance/transmittance for 810 nm but comparable detector-plane fluence after accounting for the longer 1070 nm exposure duration. Intranasal mapping demonstrated route-dependent redistribution of irradiance, and thermal measurements revealed marked differences in device self-heating despite modest skin-surface temperature changes. These findings show that meaningful tPBM evaluation requires wavelength, anatomy, source geometry, irradiance, exposure time, transmission, fluence, and thermal response to be reported as distinct but interdependent parameters.