Performance comparison of several published tissue near-infrared spectroscopy algorithms

Area of Science:

• Biomedical Optics
• Physiological Monitoring
• Spectroscopy Analysis

Background:

• Near-infrared (NIR) spectroscopy is utilized for non-invasive tissue chromophore analysis.
• Accurate quantification of chromophore concentration changes is crucial for physiological monitoring.

Purpose of the Study:

• To compare multiwavelength NIR attenuation spectra from human, rat, and piglet tissues.
• To evaluate the consistency of published algorithms for chromophore concentration derivation.
• To identify discrepancies in algorithm performance across different physiological systems and with simulated data.

Main Methods:

• Collected NIR attenuation spectra from human forearm muscle, adult rat head, and newborn piglet head.
• Applied four published algorithms to derive chromophore concentration changes.
• Utilized simulated data based on hemoglobin absorption spectra and a diffusion model for light transport.

Main Results:

• Significant differences observed in chromophore concentration estimates, particularly for cytochrome oxidase (cyt-aa3) redox, among the four algorithms.
• Algorithm performance varied when applied to different physiological datasets, indicating potential transferability issues.
• One algorithm (Piantadosi) produced notably different results for cyt-aa3 and HbO2 compared to others using simulated data.

Conclusions:

• Published NIR spectroscopy algorithms exhibit variability in estimating chromophore concentrations, especially cyt-aa3 redox changes.
• Challenges exist in transferring algorithms between diverse physiological systems.
• Discrepancies in results suggest potential inaccuracies in hemoglobin spectra, tissue models, or specific algorithm implementations.