Through a series of methodological studies, I have addressed key challenges that limit fNIRS reproducibility and interpretability:
Functional Connectivity and Network Dynamics — Demonstrated that resting-state fNIRS captures stable and biologically meaningful connectivity patterns across time and subjects, identifying key bilateral hubs and symmetrical inter-hemispheric links in spontaneous brain activity.
Motion Artifact Correction for Speech Tasks — Characterized how jaw and facial movements distort fNIRS data during speech protocols and introduced a validated hybrid correction pipeline combining spline interpolation and wavelet filtering, eliminating >90% of artifacts without inducing spurious responses.
Spatial Accuracy and Reproducibility — Integrated neuronavigation and anatomical modeling to ensure consistent probe placement across sessions, markedly increasing within-subject reproducibility during motor and cognitive tasks.
Systemic Physiology Control — Quantified and mitigated the effects of cardiac, respiratory, and vascular signals on resting-state connectivity, demonstrating that fNIRS can reproduce canonical resting-state networks comparable to fMRI when systemic physiology is accounted for.
Image-Guided Probe Placement — Developed a GPU-accelerated system that uses MRI-based cortical reconstructions and electromagnetic digitization to achieve accurate, reproducible optode placement across operators and sessions.
Spatiotemporal Optimization and Brain Fingerprinting — In simultaneous fNIRS-fMRI studies, established the spatial and temporal conditions required for subject-level identification (“brain fingerprinting”) with near-fMRI accuracy, emphasizing the power of optimized fNIRS acquisition for individual neuroscience.
Together, these developments move fNIRS from a largely exploratory tool toward a validated, anatomically precise, and reproducible imaging modality—capable of advancing both basic and translational neuroscience.