Overview
Functional near-infrared spectroscopy (fNIRS) is a promising technology for monitoring cognitive workload in naturalistic settings. However, signals are contaminated by systemic physiological noise. This project establishes best practices for preprocessing fNIRS data to reliably detect cognitive load from cortical activation.
Working Memory
N-back paradigm across multiple difficulty levels
fNIRS Technology
Non-invasive optical brain imaging
Short-Channel Regression
Removing systemic noise from cortical signals
fNIRS Preprocessing Pipeline
Figure 1: Complete fNIRS preprocessing pipeline showing raw optical density signals, motion artifact correction, short-channel regression, and hemodynamic response extraction.
My Role
- Experimental Design: Designed the N-back working memory paradigm with graded difficulty
- Data Collection: Conducted fNIRS acquisition sessions with human participants
- Signal Processing: Developed and compared preprocessing pipelines in MATLAB and Python
- Statistical Analysis: Validated short-channel regression methodology with rigorous statistics
- Supervision: Mentored student development of fnirsPy Python package
Methodology
Figure 2: (Left) N-back working memory task with 0-, 1-, 2-, and 3-back conditions. (Right) fNIRS optode configuration covering prefrontal cortex.
What's Innovative
This study provides systematic validation of short-channel regression for fNIRS-based cognitive load monitoring. Key findings:
By using short-channel measurements to regress out scalp hemodynamics, we achieve more reliable cortical activation estimates, establishing best practices for the cognitive monitoring field.
Results
Figure 3: Comparison of oxygenated hemoglobin responses to increasing N-back load with and without short-channel regression, showing improved effect sizes after correction.
Publication
Functional near-infrared spectroscopy short-channel regression improves cortical activation estimates of working memory load
Neurophotonics, 2025