fNIRS-QC: Crowd-Sourced Creation of a Dataset and Machine Learning Model for fNIRS Quality Control

Despite technological advancements in functional Near Infra-Red Spectroscopy (fNIRS) and a rise the application of fNIRS neuroscience experimental designs, processing data remains characterized by high number heterogeneous approaches, implicating scientific reproducibility interpretability results. For example, manual inspection is still necessary to assess quality subsequent retention collected signals for analysis. Machine Learning (ML) approaches are well-positioned provide unique contribution automating standardizing methodological control, where ML models can produce objective reproducible However, any successful grounded high-quality dataset labeled training data, unfortunately, no such currently available signals. In this work, we introduce fNIRS-QC, platform designed crowd-sourced creation control dataset. particular, (a) composed 4385 signals; (b) created web interface allow multiple users manually label signal 510 10 s segments. Finally, (c) subset used develop proof-of-concept model automatically The developed serve as more efficient check that minimizes error from need expertise with control.