25 Christopher Columbus Drive, STE 5403, Jersey City, NJ 07302 646-887-7984
General anesthesia induces sedation, analgesia, immobility, and amnesia and is an essential component of many surgeries. However, if managed poorly, anesthetic delivery carries risks, including intraoperative awareness, delayed recovery, organ damage, and increased mortality. Anesthesiologists monitor anesthesia through a variety of physiological indicators, but novel brain-state monitoring techniques are being increasingly used to enhance safety. Near-infrared spectroscopy (NIRS) is promising for anesthesia monitoring because it provides a noninvasive measure of real-time cerebral oxygenation, which provides a suitable proxy for brain activity. Unlike EEG and other methods, NIRS is resistant to electrical interference from operating room equipment. Previous studies using NIRS for anesthesia monitoring have analyzed time-domain (i.e., hemoglobin concentrations, sample entropy) and frequency-domain (i.e., signal power, phase-amplitude coupling) features.1
NIRS works by applying the Beer–Lambert law to near-infrared light, which can pass through biological tissue, with its absorption depending on the relative amounts of oxygenated and deoxygenated hemoglobin. Modern devices use reflected light and multiple detector spacings to distinguish superficial from deep tissue, allowing them to calculate local tissue oxygen saturation based on differences in hemoglobin absorption.2
Of course, NIRS has important limitations as a method of monitoring anesthesia depth. For starters, it measures light absorbance ratios rather than oxygen molecules, is highly sensitive to external light and probe positioning, and is best used as a trend monitor, since tissue composition causes large inter-individual variability in absolute values.2,3 Additionally, its readings reflect a mixed arterial–capillary–venous signal with unknown proportions, which means changes in blood volume or position can alter values without true changes in tissue oxygenation.2,4 Finally, low regional saturations do not always result in detectable neurologic injury and may miss ischemia occurring outside the sensor area, limiting NIRS’s sensitivity and specificity.2,5
In a clinical demonstration of NIRS anesthesia monitoring, a research group from China identified and classified three anesthesia stages: maintenance (MNT), emergence (EM), and consciousness (CON). MNT is the fully anesthetized stage after loss of consciousness (LOC); EM is the stage when anesthesia is discontinued, and the patient begins to regain consciousness; and CON is defined as the recovery stage after the patient has fully awakened (ROC). They collected NIRS data during anesthesia administration to 25 patients undergoing lower limb surgery and analyzed the NIRS signal pattern variations across MNT, EM, and CON to classify each stage.6 This study recorded NIRS signals, preprocessed the data to remove trends, motion artifacts, and noise, and then extracted time-, frequency-, and phase-amplitude-coupling (PAC)-based features from each 5-minute segment. Using feature-selection methods and a multi-class machine learning algorithm with leave-one-subject-out validation, they trained and tested a classifier to distinguish the three anesthesia stages.
As anesthesia deepened, the level of coupling between high-frequency and low-frequency signals increased.6 By demonstrating that features like PAC, power, and sample entropy reliably distinguish anesthesia stages, the work supports the development of objective, noninvasive brain-based monitoring tools that could improve intraoperative anesthesia management, enhance patient safety, and reduce reliance on subjective clinical judgment.
Near-infrared spectroscopy is a new imaging modality that offers a promising, clinically viable way to monitor brain activity during anesthesia by capturing real-time cerebral oxygenation changes. Although NIRS possesses notable limitations, recent work has shown advanced signal features on NIRS can meaningfully differentiate various stages of anesthesia depth. Continued refinement of this revolutionary technique may provide anesthesiologists with more reliable, objective tools for guiding anesthetic delivery and improving postoperative recovery.
References
1. Wang G, Liu Z, Feng Y, et al. Monitoring the Depth of Anesthesia Through the Use of Cerebral Hemodynamic Measurements Based on Sample Entropy Algorithm. IEEE Transactions on Biomedical Engineering. 2019;67(3):807-816. https://doi.org/10.1109/tbme.2019.2921362
2. Moerman A, Wouters P. Near-infrared spectroscopy (NIRS) monitoring in contemporary anesthesia and critical care. 2010;61(4):185-194.
3. Ajayan N, Thakkar K, Lionel KR, Hrishi AP. Limitations of near infrared spectroscopy (NIRS) in neurosurgical setting: our case experience. Journal of Clinical Monitoring and Computing. 2019;33(4):743-746. https://doi.org/10.1007/s10877-018-0209-1
4. Shaaban-Ali M, Momeni M, Denault A. Clinical and Technical Limitations of Cerebral and Somatic Near-Infrared Spectroscopy as an Oxygenation Monitor. Journal of Cardiothoracic and Vascular Anesthesia. 2021;35(3):763-779. https://doi.org/10.1053/j.jvca.2020.04.054
5. Tachtsidis I, Scholkmann F. False positives and false negatives in functional near-infrared spectroscopy: issues, challenges, and the way forward. Neurophotonics. 2016;3(3):031405. https://doi.org/10.1117/1.nph.3.3.031405
6. Liu Z, Si L, Shi S, et al. Classification of Three Anesthesia Stages Based on Near-Infrared Spectroscopy Signals. IEEE Journal of Biomedical and Health Informatics. 2024;28(9):5270-5279. https://doi.org/10.1109/jbhi.2024.3409163
