Graduation Semester and Year
Summer 2026
Language
English
Document Type
Dissertation
Degree Name
Doctor of Philosophy in Biomedical Engineering
Department
Bioengineering
First Advisor
George Alexandrakis
Second Advisor
Christos Papadelis
Third Advisor
Elizabeth Davenport
Fourth Advisor
Hanli Liu
Abstract
Transcranial photobiomodulation (tPBM) is a non-invasive neuromodulation technique that delivers near-infrared (NIR) light to the human brain to enhance mitochondrial metabolism and modulate cortical activity. Despite growing evidence that tPBM improves cognition and supports brain health, its mechanism of action remains incompletely understood because it has been studied piecemeal across biological scales, with gaps remaining on how to integrate information offered by different methods. Systems-level tPBM research has progressed largely in isolation across three types of studies: (1) Behavioral studies, which have repeatedly shown improvements in working memory, attention, and reaction time but cannot identify the neural substrate driving these gains; (2) Hemodynamic studies using functional near-infrared spectroscopy (fNIRS), which confirm that tPBM increases cortical oxygenation and oxidative metabolism but capture only slow hemodynamics changes; and (3) Electroencephalography (EEG) and Magnetoencephalography (MEG) sensor-level studies, which report tPBM-induced increases in neuronal power bands but, because most tPBM recordings are resting-state with inherently low signal-to-noise ratio (SNR), source reconstruction has either been avoided or limited to electromagnetic source imaging (EMSI) analyses that report only diffuse activation over large cortical areas. At the molecular level, a parallel gap persists: it remains unclear whether cytochrome c oxidase (CCO) is the dominant photoacceptor across all intervention wavelengths, or how near-infrared light modulates the electron-transfer interaction between CCO and its electron donor, cytochrome c (Cc). These unaddressed gaps, spanning from molecular photon–protein interactions to macroscopic network reorganization, have constrained tPBM's advancement from an empirically dosed intervention toward a mechanistically informed neuromodulation strategy.
Here, I present three research projects of my dissertation that have the overall aim of integrating information about tPBM effects from the brain tissue level down to the molecular scale: (i) to develop and validate a global-optimization algorithm to achieve MEG-EEG dipole source localization with few-mm accuracy even under the low-SNR conditions typical of resting-state recordings; (ii) to apply this algorithm to enable demonstrating for the first time how tPBM reorganizes cortical source dynamics in oscillatory hierarchies, where dipole sources move around the human brain in different patterns and frequencies depending on the MEG-EEG power band; and (iii) to contribute to the understanding of the molecular basis of tPBM by quantifying the binding kinetics between Cc and CCO, a key step in the Complex IV electron transfer process, at the single-molecule level for a wavelength range spanning that used in human subject tPBM studies (820 nm, 905 nm 1064 nm). The key result obtained from integrating the results of projects (i)-(iii) is that even when using longer wavelengths (1064 nm) to penetrate deeper into the human brain there is still enough light activation for CCO molecules to perform electron transfer and, by extrapolation, contribute to the downstream generation of ATP. Therefore, this work proposes that the 1064 nm wavelength can still produce tPBM effect by the same mechanism as shorter NIR wavelengths in addition to any possible heating effects on ion channels that other studies have proposed.
To delve into a little more detail in the results obtained per project:
The first project showed that a hybrid algorithm combining simulated annealing with gradient descent (hybrid-SA) outperformed conventional dipole-scanning and gradient-descent methods on a realistic pediatric head phantom, improving single-dipole localization accuracy by up to 45% and two-dipole accuracy by up to 50%, with the largest gains at low SNR (3–6) and for sources deeper than 10–17 mm, using EEG, MEG, and combined EEG-MEG data.
The second project showed that acute 1064-nm tPBM induced frequency-specific, reproducible reorganization of resting-state cortical networks in 25 healthy adults, moving beyond the diffuse activations reported previously: alpha oscillations engaged a recurring fronto-visual-sensorimotor loop, beta oscillations recruited a slower fronto-parietal-temporal loop emphasizing higher-order executive regions, and source imaging localized a shift from default mode network to central executive network dominance with strengthened directed (phase transfer entropy) connectivity between specific Brodmann areas. Both loops were hierarchically nested within infra-slow (< 0.1 Hz) phase-amplitude coupling, with no such dynamics present after sham stimulation.
The third project used a plasmon-enhanced nanopore (SANE) sensor to resolve individual Cc–CCO binding events in real time, revealing three kinetically and structurally distinct populations: short-duration binding events consistent with productive electron transfer; long-duration, binding events consistent with correctly docked but non-productive complexes; and a heterogeneous population reflecting transient, non-specific contact. Under normoxia, the productive fraction was statistically similar at 820 and 905 nm but fell statistically at 1064 nm, while the dwell time of productive events and the association kinetics of non-productive events remained wavelength-invariant - indicating that wavelength governs the probability that a docked complex completes electron transfer. Allosteric inhibition and hypoxia each suppressed the productive fraction to a much lower proportion at all three wavelengths, but through distinct mechanisms: inhibition redistributed failed encounters into the well-defined non-productive population; hypoxia redistributed them into the heterogeneous population, consistent with perturbing the Cc–CCO docking interface directly. The short-duration affinity metric increased from approximately 34–43 nM under normoxia to approximately 150–198 nM under hypoxia and 179–232 nM under allosteric inhibition, indicating substantially reduced realization of the short-duration interaction under both perturbations.
Overall, the dissertation proposes: (i) a globally optimized dipole-localization framework as a validated tool for resolving deep, low-SNR cortical sources; (ii) the first region-specific, mechanistic account of how tPBM reorganizes large-scale, frequency-specific cortical networks around infra-slow temporal scaffolds; and (iii) single-molecule evidence that the probability of completing electron transfer once Cc and CCO are correctly docked — not the probability of docking itself, and not merely bulk chromophore absorption — governs wavelength-dependent tPBM responses at the molecular level. Together, these three aims establish a multiscale methodological framework connecting single-molecule photon–protein interactions to whole-brain network reorganization, providing a mechanistic foundation for optimizing tPBM stimulation parameters in future cognitive-enhancement and neurotherapeutic applications.
Keywords
Transcranial photobiomodulation, Photobiomodulation, Cytochrome c oxidase, EEG-MEG source localization, Dipole source localization, Electromagnetic source imaging, Single-molecule sensing, Nanopore sensing, Functional connectivity
Disciplines
Bioelectrical and Neuroengineering
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
Recommended Citation
Bastola, Subrat, "Multimodal Methodology Development for Improved Understanding of Photobiomodulation and its Influence on Neuronal Activity Localization" (2026). Bioengineering Dissertations. 7.
https://mavmatrix.uta.edu/bioengineering_dissertations2/7