Recommendations for motion correction of infant fNIRS data applicable to multiple data sets and acquisition systems

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Motion artifacts are a significant noise source in infant fNIRS data.
• Existing motion correction methods, optimized for adults, may not be suitable for infants due to differences in artifact characteristics.
• Recent research has begun investigating motion correction specifically for infant fNIRS.

Purpose of the Study:

• To evaluate the effectiveness of combined Spline interpolation and Wavelet filtering for motion artifact correction in infant fNIRS data.
• To compare the performance of this combined approach against individual methods and trial rejection.
• To determine optimal motion correction strategies for infant fNIRS across various conditions.

Main Methods:

• Comparison of motion correction techniques on infant semi-simulated data.
• Evaluation of Spline interpolation and Wavelet filtering, individually and combined, on real infant cognitive fNIRS data (3 datasets, ages 5-10 months, diverse tasks).
• Quantitative assessment using hemodynamic response recovery error, within-subject and between-subjects standard deviation, and trial survival rate.

Main Results:

• Correcting motion artifacts is superior to rejecting corrupted trials.
• Wavelet filtering, particularly when combined with Spline interpolation, effectively reduces both within- and between-subject variability.
• The combined Spline and Wavelet approach demonstrated optimal performance in semi-simulated data and recovered the most motion-affected trials across real datasets.

Conclusions:

• The combined use of Spline interpolation and Wavelet filtering is the most effective strategy for motion artifact correction in infant fNIRS data.
• This approach enhances data quality by minimizing noise and maximizing usable trial data, which is critical for infant research.
• Findings provide evidence-based recommendations for processing infant fNIRS data, advancing the field of developmental neuroscience.