The fundamental principle of NIRS relies on the interaction of near-infrared light with biological tissues [11]. Near-infrared light, typically in the wavelength range of 700 to 1100 nanometers, exhibits the property of penetrating tissue several centimeters due to relatively low absorption and dominant scattering by tissue components [12]. Within this nearinfrared window, key chromophores present in the brain, namely oxygenated hemoglobin (HbO2), deoxygenated hemoglobin (HHb), and cytochrome c oxidase (CCO), exhibit distinct absorption spectra. This unique spectral signature allows NIRS to differentiate and quantify changes in the concentration of these chromophores, providing valuable information about tissue oxygenation and metabolic state. While light scattering in tissue complicates the direct measurement of absorption, it is also this scattering that enables the near-infrared light to travel through several centimeters of tissue, making non-invasive assessment of deeper tissues like the brain possible [13]. Broadband NIRS systems further exploit this principle by measuring changes in light attenuation across a wide range of wavelengths, enabling the retrieval of information not only about hemoglobin oxygenation but also about the redox state of CCO, which serves as an indicator of cellular energy utilization [14].

NIRS measurements are typically performed using one of three main spectroscopic techniques: Continuous wave (CW) spectroscopy, frequency domain (FD) spectroscopy, and time-resolved spectroscopy (TRS), also known as time domain (TD) spectroscopy [15]. CW spectroscopy represents the simplest and most widely used technique, involving the steady illumination of tissue with light and the detection of the transmitted light intensity [16]. While CW spectroscopy cannot directly differentiate between the effects of absorption and scattering, it allows for the reliable assessment of relative changes in blood volume and oxygenation, making it particularly useful for functional neuroimaging applications where changes over time are of primary interest. CW systems are generally cost-effective but rely on the assumption of homogenous tissue and a fixed path length of light through the tissue [17]. FD spectroscopy represents a more advanced technique where the light source is intensity-modulated at a high frequency (typically in the radio frequency range) [18]. By measuring both the attenuation of light intensity and the phase shift of the modulated signal as it passes through the tissue, FD spectroscopy allows for the distinction between absorption and scattering properties of the tissue and holds the potential for absolute quantification of chromophore concentrations [19]. However, FD systems are more complex and costly than CW systems and may exhibit a greater degree of noise in the recovered parameters [20]. TRS or TD spectroscopy utilizes extremely short pulses of light (on the order of picoseconds) and measures the distribution of arrival times of photons that have traveled through the tissue [21]. This technique provides the richest information about the optical properties of the tissue, enabling the quantification of both absorption and scattering coefficients with improved spatial resolution [22]. Despite its advantages, TD spectroscopy is often associated with cumbersome and expensive instrumentation, limiting its widespread clinical adoption. Comparison of NIRS measurement principles are listed in Table 1.

From these NIRS measurements, several key parameters can be derived, providing insights into cerebral oxygenation and hemodynamics [23]. ScO2, also referred to as regional cerebral oxygen saturation (rSO2), tissue oxygenation index (TOI) is a primary parameter that reflects the overall oxygen saturation within the cerebral tissue, encompassing a mixture of arterial, venous, and capillary blood. This parameter indicates the balance between oxygen delivery to the brain and oxygen consumption by the brain tissue [24]. It is important to note that different NIRS device manufacturers may use different terms and proprietary algorithms to calculate this value. Cerebral blood volume (CBV), which represents the total volume of blood in the cerebral tissue, can also be derived from NIRS measurements, often reflecting changes in the total hemoglobin concentration [25]. More advanced time-resolved NIRS techniques can even provide absolute measurements of CBV [25]. Furthermore, NIRS allows for the monitoring of changes in the concentrations of HbO2 and HHb based on their differential absorption of light at specific wavelengths [26]. Another derived parameter, fractional tissue oxygen extraction (FTOE), can be estimated by combining cerebral oxygenation measurements with peripheral oxygen saturation. FTOE provides an indication of the proportion of delivered oxygen that is extracted by the brain tissue, offering a surrogate marker of cerebral metabolism [27].

While NIRS offers a non-invasive window into cerebral oxygenation, it is important to consider its limitations in terms of depth of penetration and spatial resolution [28]. For brain imaging applications, the typical penetration depth of NIRS is approximately 3 centimeters [29], although this can vary depending on the wavelength of light used [30]. This depth limitation means that NIRS primarily allows for the assessment of the superficial cortical regions of the brain [29]. The spatial resolution achievable with NIRS is generally in the range of 5 to 10 millimeters, although some studies have reported resolutions of 2 to 3 centimeters [31]. More advanced techniques like time-domain near-infrared optical tomography have demonstrated the potential to achieve resolutions up to a depth of 30 millimeters in controlled liquid phantoms [32]. The sensitivity of NIRS to localized changes in light absorption can also vary depending on the specific monitoring device used [3]. However, advancements in high-density wearable NIRS systems are showing promise in achieving spatial resolutions comparable to functional magnetic resonance imaging (MRI). These considerations regarding penetration depth and spatial resolution are crucial for understanding the anatomical scope of NIRS measurements and for interpreting the resulting data in various clinical contexts. Having explored the fundamental principles of NIRS and how this non-invasive technology utilizes near-infrared light to measure key parameters like ScO2, also known as rSO2 or TOI, providing valuable insights into brain tissue oxygenation and metabolism, the succeeding section now delve into the specific clinical contexts where NIRS is applied.

During neurosurgical procedures, maintaining adequate cerebral oxygenation is critical to prevent intraoperative ischemic injury. NIRS has emerged as a valuable tool for intraoperative neuromonitoring in several specific neurosurgical contexts.

NIRS plays an important role in the postoperative period, particularly in the neurosurgical intensive care unit (ICU), for the continuous monitoring of cerebral oxygenation and the early detection of complications. Continuous NIRS monitoring is feasible and valuable in the neurosurgical ICU for critically ill patients [53]. In patients who have experienced a SAH, NIRS can aid in the early detection of DCI [54]. Studies have shown significant differences in rSO2 levels between patients who develop DCI and those who do not, particularly between days 7 and 9 post-hemorrhage [40]. A decline of more than 12.7% in the rSO2 rate has demonstrated high sensitivity and specificity for identifying DCI, indicating that NIRS is a feasible method for the real-time detection of this serious complication [55]. NIRS is also utilized for monitoring cerebral oxygenation in patients with traumatic brain injury (TBI) in the postoperative period [41]. It has long been recognized as a promising non-invasive tool for monitoring cerebral tissue oxygenation and perfusion in the context of TBI [13] and has the potential for earlier initiation of monitoring with minimal interoperator variability [45]. NIRS can detect intracranial bleeding and assess brain tissue oxygenation and cerebral perfusion in TBI patients [43]. Research suggests that higher rSO2 levels are associated with survival in TBI patients, whereas values below 68% may indicate a critical threshold for increased mortality risk [45].

In the critical care management of TBI, NIRS serves as a valuable tool for continuous monitoring of cerebral oxygenation and hemodynamics [45]. While some studies have explored the correlation between NIRS parameters and ICP as well as CPP [44], the sensitivity of NIRS in reliably detecting or predicting changes in ICP remains limited, although some temporal relationships have been observed [48]. NIRS can also be used to detect secondary insults such as ischemia and hypoxia in TBI patients [52], and it has shown sensitivity in detecting severe cerebral hypoxia (PbrO2 < 12 mm Hg) but may be less effective in milder cases [56]. Furthermore, NIRS plays a role in guiding management strategies aimed at optimizing cerebral oxygenation in TBI patients, with proposed algorithms for clinical decision-making based on cerebral desaturation episodes [57].

NIRS has potential applications across the spectrum of stroke care. In the acute phase of ischemic stroke, it has been investigated for its ability to aid in the early detection of the penumbra and the monitoring of collateral circulation [57]. However, one study found that NIRS monitoring during endovascular treatment for acute ischemic stroke did not detect significant differences in rSO2 between hemispheres or changes within a hemisphere over time [50]. NIRS also has a role in monitoring cerebral oxygenation during and after thrombolysis or thrombectomy procedures [52]. In the management of hemorrhagic stroke, particularly subarachnoid hemorrhage, cerebral oximetry using NIRS has demonstrated utility in the diagnosis of secondary hypoperfusion and cerebral vasospasm [52].

As previously mentioned, continuous NIRS monitoring is valuable for the detection of DCI in patients with SAH. Changes in NIRS parameters, such as rSO2, have been shown to correlate with the occurrence and severity of angiographic vasospasm, a major contributor to DCI [58]. This suggests that NIRS can potentially serve as a tool to guide prophylactic or therapeutic interventions aimed at preventing or treating DCI in this high-risk population, with proposed decision-making algorithms based on cerebral desaturation [59].

Cerebral oxygenation monitoring is of paramount importance in neonates and children who are neurologically compromised. NIRS has found widespread applications in this population, particularly in conditions such as hypoxic-ischemic encephalopathy (HIE), congenital heart disease, and other critical illnesses [49]. In infants with HIE, NIRS can provide continuous bedside information about brain hemodynamics, oxygenation, and metabolism. It is also widely used in pediatric cardiac surgery to monitor tissue oxygenation [38]. Furthermore, NIRS is utilized to monitor the effects of various interventions designed to improve cerebral oxygenation in this vulnerable population [60]. Studies have indicated that specific NIRS-derived parameters, such as cerebral ScO2, can significantly predict adverse outcomes in infants with HIE within the first 24 to 72 hours of life in some instances [60].

NIRS monitoring plays a significant role in detecting cerebral hypoperfusion during cardiopulmonary bypass (CPB) in patients undergoing cardiac surgery [61]. While the correlation between intraoperative NIRS desaturations and postoperative neurological outcomes is still under investigation, NIRS-guided strategies are being explored to optimize cerebral perfusion during these complex procedures [62]. For instance, the determination of cerebral pressure-flow autoregulation limits is now possible using processed oximetry signals in relation to arterial pressure [42]. Notably, recent research suggests that NIRS monitoring during cardiac surgery with CPB can help reduce the incidence of postoperative delirium, a common complication in this patient population [63].

The accuracy of NIRS in identifying cerebral ischemia has been evaluated in numerous studies, often compared against gold standard methods such as microdialysis and imaging techniques [64]. One study proposed a correlation between a novel NIRS sensor and a brain tissue oxygen probe in detecting ischemic events. However, the reliability of NIRS measurements can be influenced by several factors, in cluding extracranial contamination from superficial tissues and inherent physiological variability [65]. The rSO2 signal can be affected by anatomical confounders such as the thickness of extracranial tissues, the skull, and the cerebrospinal fluid, as well as dynamic factors like anemia, acid-base status, and changes in arteriovenous blood partitioning [66]. The presence of intermediate structures like skin, skull, meninges, and cerebrospinal fluid can significantly influence NIRS results [67]. Erroneous NIRS values may also arise due to altered tissue boundary conditions during surgery or inadequate penetration depth of the sensors [68]. Furthermore, the use of different NIRS devices and proprietary algorithms by various manufacturers can lead to non-comparable absolute oxygenation values, posing challenges for standardization and interpretation across different studies and clinical settings [67].

Establishing clear thresholds for significant changes in NIRS parameters that are indicative of cerebral ischemia is crucial for clinical decision-making. Several studies have proposed and investigated various thresholds for different neurological conditions [69]. For instance, in carotid endarterectomy, a 12.3% decrease in rSO2 has been identified as an optimal threshold for indicating intraoperative hypoperfusion [39], while different thresholds may be necessary for symptomatic and asymptomatic patients [1]. In the context of subarachnoid hemorrhage, a reduction in rSO2 greater than 14.7% has been associated with delayed cerebral ischemia [2], and a decline of more than 12.7% in the SO2 rate has shown high sensitivity and specificity for DCI detection in another study [14]. During aneurysm clipping, a 20% decrease in rSO2 from baseline predicted neurological compromise with good accuracy [22]. A widely used general paradigm for defining cerebral desaturation is a reduction of more than 20% from the patient's baseline value or an absolute cerebral oxygen saturation value of less than 50%. Given the inherent variability in baseline cerebral oxygen saturation values among individuals, monitoring trends in NIRS data is often considered more clinically useful than relying on isolated absolute measurements [38]. Integrating NIRS data with other physiological monitoring modalities, such EEG in neonates, can provide a more comprehensive assessment of the patient's neurological status and improve the interpretation of NIRS findings [41].

Real-time NIRS monitoring has been successfully used to guide clinical decision-making in various neurosurgical and critical care scenarios [48]. In carotid endarterectomy, NIRS monitoring facilitates the implementation of selective shunting strategies, where a significant drop in cerebral oxygen saturation triggers the placement of an intraluminal shunt [50]. During cardiovascular surgery, NIRS can be used to determine the limits of cerebral pressure-flow autoregulation, allowing for the optimization of blood pressure management [58]. Furthermore, NIRS monitoring during cardiac surgery with cardiopulmonary bypass has been shown to help reduce the incidence of postoperative delirium [49]. Based on NIRS readings, clinicians can make adjustments to blood pressure, ventilation, and other physiological parameters to maintain adequate cerebral oxygenation [60]. For instance, in cases of cerebral desaturation detected by NIRS, interventions such as increasing mean arterial pressure, adjusting arterial partial pressure of oxygen, or modifying the fraction of inspired oxygen may be implemented [42]. There is growing evidence suggesting that NIRS-guided management can lead to improved neurological outcomes in certain clinical settings, such as the association of NIRS-guided selective shunting with a lower rate of perioperative stroke in patients undergoing carotid endarterectomy [65].