@inproceedings{yang_towards_2022,
    author = {Yang, L and Grosenick, D and Wabnitz, H and von L{\"u}hmann, Alexander},
    address = "Boston, USA",
    title = "Towards the integration of {CW} {fNIRS} and absolute oximetry: {A} proof of concept",
    copyright = "All rights reserved",
    abstract = "Introduction: Monitoring relative and absolute tissue oxygenation changes simultaneously and in real-time can be advantageous in (neuro)physiological research. Recent works have shown the feasibility to estimate absolute oxygen saturation in human brain based on wearable continuous wave (CW) NIRS [1,2]. Robust estimation of absolute optical properties requires higher precision when compared to relative oximetry (i.e., optical and temperature drifts, geometric set-up) and there is currently no fNIRS-based imager available that provides both: concurrent regular (whole-head) fNIRS brain-imaging and absolute oximetry. Here we present a first proof of concept for a combined solution. Methods: A freely configurable NIRSport2 CW Imager (NIRx Medizintechnik GmbH, Germany) with 16 LED sources (760 nm and 850 nm) and 16 SiPD detectors and Aurora recording software were used for signal acquisition and probe calibration. Absolute StO2 measurements were based on rectangular patches with 2 sources and 2 detectors in a quasi-symmetric geometry (Fig. 1B). Data were streamed via LabStreamingLayer (LSL) to a physiological SRS model in Matlab (Mathworks) with a practical amendment in approximation to the classical self-calibrating probe [3], which compensates minor imprecisions in the free probe placement. For performance verification, precision, drifts and noise of estimated StO2 were quantified using 4 tissue-mimicking phantoms with known optical properties based on previous time-domain experiments. The phantom’s nominal (pseudo) StO2 values were calculated from the absorption coefficient at the source wavelengths, assuming a (virtual) 75\\% water content in the model. Results: Phantom tests yielded accurate StO2 estimation at {\textgreater} 1 Hz update rate with a small deviation from the nominal StO2 values: Phantom 1 (nominal 43.08\\%): 43.23\\%±0.30\\%; Phantom 2 (42.34\\%); 42.18\\%±0.24\\%, Phantom 3 (49.13\\%): 52.72\\%±0.30\\%; Phantom 4 (48.88): 48.58\\%±0.31\\%) and negligible drifts ({\textless}5x10-5 /min). Standard deviation in the StO2 estimation resulting from discrepancies in the quasi-symmetric geometry (source rotation → variation in multi-wavelength LED die distances to detectors), was determined to be ≤5\\% In vivo StO2 measurements (e.g., vascular occlusion tests), yielded values and time courses similar to literature (not part of this study). Fig 1. NIRSport2 16x16 with concurrent regular fNIRS (A) and StO2 (B) measurement setup on a phantom. After calibration, raw intensity data is streamed from Aurora to a SRS-based selfcalibrating model in Matlab via LSL Conclusion: The presented work provides first evidence that simultaneous and precise measurements of fNIRS and absolute oximetry can be performed robustly and in real-time with the same CW imager, when hardware and an appropriate physical model / geometry are correctly combined. Concurrent ΔhbO, ΔHbR and StO2 measurements in a customizable and scalable setup appear to be within reach.",
    language = "en",
    booktitle = "Proc. {Biennial} {Meeting} of the {Society} for {fNIRS} 2022",
    publisher = "SfNIRS",
    month = "October",
    year = "2022",
    pages = "1"
}