Automated Processing of fNIRS Data-A Visual Guide to the Pitfalls and Consequences

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Functional near-infrared spectroscopy (fNIRS) is increasingly used in various research fields.
• A growing number of new users rely on commercial software for fNIRS data analysis.
• Concerns exist regarding potential biases introduced by suboptimal processing methods in fNIRS studies.

Purpose of the Study:

• To provide a visual reference illustrating the impact of different fNIRS processing methods.
• To guide researchers in establishing and evaluating robust fNIRS data processing pipelines.
• To highlight the importance of informed processing choices for accurate fNIRS results.

Main Methods:

• Systematic evaluation of various pre-processing techniques for fNIRS data.
• Assessment of different post-processing strategies applied to fNIRS signals.
• Comparative analysis of data processing pipelines using simulated or real fNIRS data.

Main Results:

• Demonstration of significant effects of pre- and post-processing choices on fNIRS outcomes.
• Visual evidence showcasing how different methods alter the interpretation of brain activity.
• Quantification of the impact of processing on the accuracy of activation site identification.

Conclusions:

• Processing choices critically influence the validity and reliability of fNIRS findings.
• Researchers must be aware of and carefully select processing steps to avoid introducing bias.
• Combining data from both oxygenated and deoxygenated hemoglobin is essential for accurate fNIRS-based activation inference.