Near infrared spectroscopy is a non-invasive continuous technique to monitor regional tissue haemoglobin saturation and may be particularly useful for patients at risk for developing cerebral ischaemia
Patrick Schober MD PhD
Lothar A Schwarte MD PhD
Department of Anaesthesiology,
VU University Medical Center,
Amsterdam, The Netherlands
Brain function highly depends on oxidative metabolism of glucose, and inadequate cerebral oxygenation readily leads to neuronal injury. Maintenance of adequate cerebral oxygenation is a main therapeutic goal in many fields of medicine.(1) Herein, clinical neurologic assessment is often useful to detect impairments of cerebral oxygenation, but its usefulness is severely limited in sedated, anaesthetised or comatose patients. Therefore, especially in anaesthesiology and intensive care medicine, a continuous non-invasive technique for early detection of cerebral hypoxia is highly desirable. Until recently, however, routine bedside monitoring of cerebral oxygenation was not widely available because measurement techniques were quite invasive (for example, polarographic probes) or only provide indirect information and/or require specially trained operators (for example, jugular bulb oximetry, electroencephalography, transcranial Doppler ultrasound). Therefore, in routine clinical practice, parameters of systemic haemodynamics and oxygenation are often used as surrogates to estimate cerebral perfusion and oxygenation, and prevention of cerebral hypoxia usually aims at maximising oxygen delivery to the brain.
However, cerebral hypoxia can occur despite normal systemic oxygenation and haemodynamics, and by contrast, potential risks of hyperoxia are increasingly appreciated.1 Patient management should ideally aim at specific oxygenation targets while avoiding hypoxia and hyperoxia, and a suitable monitoring technique is essential to assess whether brain oxygenation is within the targeted range.
Near infrared spectroscopy (NIRS) is a non-invasive technique suitable for continuous monitoring of cerebral oxygenation. Triggered by technical refinements and increasing marketing of user-friendly and affordable commercial devices, NIRS has emerged as a technique that is increasingly used in clinical practice. This article reviews the technical basis as well as limitations and clinical applications of this monitoring technique.
What are we measuring?
NIRS is based on transmission of near infrared light (approximately 750–1400nm) through tissues and measurements of light attenuation after tissue transit. While NIR light easily passes through biological tissues, including the cranium, for several centimetres,(2) the distance is too short to transluminate the whole head in adults. Therefore, reflectance spectroscopy is used in clinical practice: because NIR light is partially reflected and scattered at intra- and extracellular boundaries, a small portion of light travels back from the tissue to the skin surface in a banana-shaped pathway, where it can be detected by a superficial sensor.(1,2)
Herein, the depth of the measured tissue volume is proportional to the distance between light emitter and detector,(3) allowing interrogation of the cerebral cortex when the light emitter and sensor are placed on the forehead in a distance of a few centimetres (typically 4–5cm, depending on the manufacturer). Commercial devices comprise the light emitter and one or more detectors in a single sticker, which stabilises the probe and restricts interferences by ambient light.(1)
As NIR light traverses the tissue of interest, photons are absorbed by molecules along the optical pathway.(4) Each substance has a characteristic absorption spectrum, which, for the case of molecules involved in oxygen transport (haemoglobin) or cellular respiration (for example, cytochrome c oxidase), depends on the oxygenation status. Deoxygenated haemoglobin shows a peak of light absorption at approximately 760nm, whereas oxyhaemoglobin peaks at about 900nm.(3) Oxidised cytochrome c oxidase has a peak around 830nm that disappears in the reduced state.(3) According to the Lambert-Beer law, the ratio of the intensity of emitted to detected light is a function of the molecule concentration, the length of the light pathway and a substance specific absorption coefficient.(2,4) Hence, these optical properties should ideally allow to determine attenuation of light at specific wavelengths and to calculate concentrations of oxygenated or deoxygenated molecules up to the mitochondrial level. However, this is unfortunately not routinely possible in vivo. First, absorption is not the only factor contributing to attenuation of the light intensity in human tissue because scatter also plays a major role.(4) Second, determination of the length of the ‘banana-shaped’ light pathway is intricate(4) and, up to now, not routinely implemented in bedside devices. Under the assumption that scatter similarly contributes to light attenuation at different wavelengths within the NIR range and is relatively constant over time, it is possible to determine relative changes in concentrations or the ratio of two concentrations because the light path length cancels out.(4)
In practice, the ratio of primary interest is haemoglobin oxygen saturation (oxygenated versus total haemoglobin). Different NIRS devices use different wavelengths, algorithms and nomenclature, but the common ground of all measurements is that relative attenuation of light at wavelengths characteristic for oxygenated and deoxygenated haemoglobin are compared to derive haemoglobin oxygen saturation. It seems appealing to also measure cellular oxygenation and, in fact, because the absorption peak of oxidised cytochrome c oxidase is very broad, there is substantial overlap with the absorption spectrum of haemoglobin.4 By using multiple wavelengths, it is principally possible to differentiate between the absorption caused by haemoglobin from that of cytochrome c oxidase but, because haemoglobin is present in much larger quantities in tissue, the total light attenuation caused by cytochrome c oxidase is only one tenth compared with that of haemoglobin.(3) Hence, currently available commercial NIRS monitors basically report regional tissue haemoglobin saturation (rSO2) but provide no direct information on cellular oxygenation. Whereas approximately 20–30% of blood in tissues is found in arterioles and approximately 5% in capillaries, most of the blood is located in venules.(3) The measured rSO2 is therefore a composite pre- and post-cellular haemoglobin saturation with a predominant post-cellular component and reflects the balance between cortical oxygen delivery and consumption.
NIRS aims to measure cerebral oxygenation but the beam of light also passes the scalp, skull, meninges and cerebrospinal fluid, and the question arises as to how far oxygenation of extracerebral tissue contaminates cerebral oxygenation measurements. Modern devices use two or more photosensors at different distances from the light emitter in an attempt to differentiate between oxygenation of cerebral and extracerebral tissue. Yet, a recent study demonstrates that extracranial oxygenation can still lead to significant contamination of the cerebral oxygenation reading.(5)
Haemoglobin and cytochrome c oxidase are the most relevant oxygen-dependent molecules that cause light absorption in the NIR range, but there are a number of molecules that absorb NIR light but which are unrelated to oxygenation (for example, melanin and bilirubin).(2,3) Under the assumption that the concentration of these molecules does not change over time, changes in NIRS readings should still reflect changes in cerebral oxygenation.
In clinical practice of cerebral NIRS monitoring, the pad(s) containing the light emitter and sensor(s) are usually placed on one or both sides of the patient’s forehead and the monitored frontal cortical region receives blood from the anterior and medial cerebral artery.(1) Obviously, NIRS is unable to detect subcortical hypoxia or selective cortical ischaemia in the posterior cerebral circulation. In this context, it should also be noted that saturation measurements over non-metabolising (for example, necrotic) tissue may be misleading, because they may be low (absent perfusion) or high (no oxygen consumption) but have no specific diagnostic value.(2) In fact, near-normal or even high cerebral oxygen saturations have been measured by NIRS in human cadavers, questioning the validity of the measurements.(2) However, autopsy studies have shown that cerebral venous oxygen saturation can actually be as high as 95% and might, in part, reflect the balance of oxygen delivery and consumption during the dying process.
Similarly to pulse oximetry, NIRS cannot account for the influence of dyshaemoglobinaemia unless additional characteristic wavelengths are monitored. Biased readings should be anticipated in the presence of carboxyhaemoglobin and methaemoglobin.
Validation of NIRS
Despite the limitations discussed above, numerous validation studies and clinical investigations suggest that NIRS may be a valuable instrument to monitor cerebral oxygenation in a variety of patient populations. Validation of NIRS technology is difficult because there is no well-defined ‘gold-standard’ of tissue saturation measurement to which the technology can be compared. Currently, the most direct measure of cerebral oxygenation is the invasive measurement of tissue oxygen tension, and a few studies suggest that both methods show significant correlation and similarly reflect changes in cerebral oxygenation. Other studies compared NIRS with jugular venous saturation. Some degree of correlation is reported in most of these studies but agreement between the measurements is not always convincing.(3) However, jugular venous saturation is a measure of global (or hemispheric) cerebral oxygenation, whereas NIRS determines regional tissue saturation, and therefore it is unsurprising that agreement is not always high.
Other studies have validated NIRS by measuring its ability to track controlled manipulations in cerebral perfusion and oxygenation in healthy subjects as well as patients. Such manipulations include occlusion of the carotid artery, pharmacologic interventions to manipulate cerebral or systemic haemodynamics, breathing of hypoxic gas, or manipulations of cerebral blood flow by altering carbon dioxide tensions.(3) A majority of these studies indeed does suggest that NIRS is able to adequately track changes in cerebral oxygenation.
NIRS-guided monitoring is especially used in patients at increased risk of developing cerebral ischaemia and hypoxia, such as patients undergoing carotid endarterectomy or cardiac surgery. It must be noted that large inter-individual differences in baseline rSO2 values have been observed, and normal values have been reported to range from approximately 60% to 75%. Moreover, elderly patients undergoing major surgery seem to have lower baseline values that could reflect advanced age and relevant comorbidities compared with healthy controls.(6) This suggests that individual values may be a poor predictor of cerebral hypoxia, and that changes relative to the individual baseline value should be considered. Decreases of 15–20% from baseline have been suggested as intervention threshold.(6) Likewise, absolute values below 50% are also considered pathological and have repeatedly been associated with poor outcome.(6) Nonetheless, further research is needed to validate reference values, intervention thresholds, treatment protocols, and therapeutic targets for different patient populations.
Carotid artery clamping during endarterectomy is associated with a high risk of cerebral ischaemia if collateral perfusion is insufficient. Monitoring of cerebral oxygenation is essential to determine whether a shunt needs to be placed. In patients undergoing general anaesthesia, surrogates such as somatosensory-evoked potentials (SSEP), stump pressure (SP) measurements or transcranial Doppler (TCD) monitoring are used for this purpose.(2) SSEP and TCD require special expertise and are not possible in all patients, and NIRS may be a simple and useful alternative. A number of studies have demonstrated the ability of NIRS to detect ischaemic episodes. Sensitivity and specificity vary depending on which cut-off values are used and are difficult to determine because of lack of a ‘gold-standard’.
Moritz et al compared NIRS with SSEP, SP and TCD in patients undergoing carotid endarterectomy under regional anesthesia and used neurologic deterioration as indicator of insufficient cerebral perfusion.(7) These authors found that a 20% reduction from baseline in rSO2 provided similar accuracy for detection of ischaemia as SP and TCD and was superior to SSEP monitoring. In a large cohort study involving almost 600 patients, Mille and colleagues identified a 12% decrease from baseline as ‘optimal’ statistical cutoff point, but report that the traditional 20% threshold may yet be important to identify patients who develop neurologic complications: of those patients with a >20% reduction in rSO2 from baseline, 37% developed neurological complications whereas complications were observed in only 2% of patients with a ≤20% decrease from baseline.(8)
Cardiac surgery is also frequently associated with postoperative neurological complications due to emboli or hypoperfusion when extracorporeal circulation is used. Several studies observed an association between cerebral oxygenation, complications and length of hospital or ICU stay. Goldman et al compared outcome of cardiac surgical patients after introduction of rSO2 monitoring to historical controls, and report a significant reduction in stroke rate and length of hospital stay.(9) Favourable effects of NIRS monitoring were also confirmed in a randomised controlled trial: in patients undergoing coronary artery bypass grafting, Murkin et al assigned 200 patients to receive either blinded rSO2 monitoring without further intervention, or an intervention algorithm to maintain rSO2 at ≥75% of the baseline reading. In the treatment group, a significant reduction in the incidence of major complications as well as shorter length of ICU stay was observed.10 Among cardiac surgical patients, those undergoing aortic arch repair in deep hypothermic circulatory arrest are particularly at risk for postoperative cerebral injury and death. Here, NIRS has been shown to detect cerebral hypoxia and to predict neurological complications,(2) but intervention thresholds and treatment algorithms still need to be established.
Patients with brain injuries or intracranial haemorrhage are at risk of cerebral hypoxia, and the relationship between impaired cerebral oxygenation and poor outcome has repeatedly been reported. In patients with subarachnoidal haemorrhage, a strong association between episodes of cerebral vasospasm and reduction in ipsilateral rSO2 reading has been observed. In traumatic brain injury, high intracranial pressures have been associated with lower rSO2 values, and NIRS has been proposed as diagnostic tool for the detection of intracranial hematomas. However, until now, few data on NIRS-guided management in head trauma are available and studies addressing its impact on outcome are lacking.
While this article mainly deals with measurements of cerebral oxygenation, NIRS is also used to assess perfusion and oxygenation of somatic tissues. Measurements of muscle oxygenation (typically the thenar or forearm) are useful for early detection and management of shock (for example, volume resuscitation in trauma patients), because cerebral perfusion is well preserved in early shock whereas muscle rSO2 deteriorates early during centralisation.(1) Other somatic measurement sites include the splanchnic, renal and hepatic circulation.
NIRS is a non-invasive continuous technique to monitor regional tissue haemoglobin saturation, and numerous studies have confirmed its ability to adequately track changes in cerebral oxygenation. This monitoring may particularly be useful for patients at risk for developing cerebral ischaemia, such as patients undergoing cardiac surgery, carotid endarterectomy or patients with brain injuries. A limited number of studies suggest improved outcome when NIRS-directed treatment protocols are used to maintain cerebral oxygenation. However, further research is needed to validate reference values, intervention thresholds and treatment protocols to allow widespread use of this promising technology in various patient populations.
- Scheeren TW, Schober P, Schwarte LA. Monitoring tissue oxygenation by near infrared spectroscopy (NIRS): background and current applications. J Clin Monit Comput 2012;26:279–87.
- Murkin JM, Arango M. Near-infrared spectroscopy as an index of brain and tissue oxygenation. Br J Anaesth 2009;103 Suppl 1:i3–13.
- Madsen PL, Secher NH. Near-infrared oximetry of the brain. Prog Neurobiol 1999;58:541–60.
- Wahr JA et al. Near-infrared spectroscopy: theory and applications. J Cardiothorac Vasc Anesth 1996;10:406–18.
- Davie SN, Grocott HP. Impact of extracranial contamination on regional cerebral oxygen saturation: a comparison of three cerebral oximetry technologies. Anesthesiology 2012;116:834–40.
- Casati A et al. New technology for noninvasive brain monitoring: continuous cerebral oximetry. Minerva Anestesiologica 2006;72:605–25.
- Moritz S et al. Accuracy of cerebral monitoring in detecting cerebral ischemia during carotid endarterectomy: a comparison of transcranial Doppler sonography, near-infrared spectroscopy, stump pressure, and somatosensory evoked potentials. Anesthesiology 2007;107:563–9.
- Mille T et al. Near infrared spectroscopy monitoring during carotid endarterectomy: which threshold value is critical? Eur J Vasc Endovasc Surg 2004;27:646–50.
- Goldman S et al. Optimizing intraoperative cerebral oxygen delivery using noninvasive cerebral oximetry decreases the incidence of stroke for cardiac surgical patients. Heart Surg Forum 2004;7:E376–81.
- Murkin JM et al. Monitoring brain oxygen saturation during coronary bypass surgery: a randomized, prospective study. Anesth Analgesia 2007;104:51–8.