Cerebral near infrared spectroscopy: emitter detector separation must be increased

Transcription

1 British Journal of Anaesthesia 82 (6): (1999) Cerebral near infrared spectroscopy: emitter detector separation must be increased T. J. Germon 1 *, P. D. Evans 3, N. J. Barnett 3,P.Wall 3, A. R. Manara 2 and R. J. Nelson 1 1 Department of Neurosurgery and 2 Department of Anaesthetics, Frenchay Hospital, Bristol BS16 1LE, UK. 3 Johnson & Johnson Medical Ltd, European Development Centre, Newport, UK *Corresponding author We have compared the effect of increasing optode separation (range cm) on the sensitivity of near infrared spectroscopy (NIRS) to discrete reductions in scalp and cerebral oxygenation in 10 healthy men (mean age 32, range yr) using multichannel NIRS. During cerebral oligaemia (a mean reduction in middle cerebral artery flow velocity of 47%) induced by a mean reduction in end-tidal PCO 2 of 2.4 kpa, the decrease in oxyhaemoglobin detected by NIRS became significantly greater with increasing optode separation (P ). In response to scalp hyperaemia induced by inflation and release of a pneumatic scalp tourniquet, increases in oxyhaemoglobin became significantly smaller with increasing optode separation (P ). These results are consistent with theoretical models of the behaviour of NIR light in the adult head and support the concept of using multi-detector NIRS to separate intra- and extracranial NIR signal changes. However, the emitter detector separation used by currently available cerebral oximeters is not large enough to provide optimal spatial resolution. Br J Anaesth 1999; 82: Keywords: measurement techniques, near infrared spectroscopy; brain, oxygen consumption; equipment, oximeters Accepted for publication: January 26, 1999 Cerebral near-infrared spectroscopy (NIRS) is based on the principle that NIR light is randomly scattered, absorbed and transmitted by tissues and can penetrate the scalp and skull before undergoing absorption in the cerebrum. As a result of random scattering, some incident light is reflected back to the surface where changes in NIR attenuation can be measured. The absorption of NIR light is governed by the Beer Lambert law and depends on the concentration, absorption spectra and coefficients of absorption of chromophores and the path length of NIR light. The wavelengths of the NIR light used in clinical NIRS are chosen to be sensitive to important biological chromophores, including oxyhaemoglobin (O 2 Hb), deoxyhaemoglobin (HHb) and oxidized cytochrome C oxidase (Cyt). The behaviour of NIR light as it passes through a complex medium such as the human head is poorly understood. A variable proportion of incident light may penetrate the cerebrum and there may be significant differences within the NIR part of the spectrum. Theoretical modelling suggests that the average depth of penetration of NIR light detected by transcutaneous, reflectance-mode NIRS is directly proportional to the separation of the emitting and receiving optodes. 1 5 When the optodes are placed close together, most of the transmitted light is thought to pass through the superficial tissues of the scalp and skull. If the distance between the optodes is increased, a greater proportion of the light should pass through the cerebrum. As a result, the influence of changes in extracranial tissue oxygenation on NIRS determination of cerebral O 2 Hb and HHb concentrations becomes less if optode separation is increased. While we have demonstrated that these principles hold true in practice, 6 some reports have questioned the validity of cerebral NIRS We have considered the proposition that the emitter detector separation used by current oximeters may be sub-optimal. We have determined optimal emitter detector separation by comparing changes in O 2 Hb, HHb and total haemoglobin (thb) detected by multichannel NIRS during induction of extracranial ischaemia using a pneumatic scalp tourniquet and cerebral hyperaemia as a standard test of cerebrovascular reactivity. Subjects and methods A prototype NIR spectrometer was used, the technical details of which have been described previously. 14 NIR light is conducted along 2.8 m of fibreoptic cable to a silicon rubber sensor where the light emitter is mounted together with an array of high sensitivity photodiodes forming eight detection channels. The photodiode detectors are mounted in a line at 0.7, 1.3, 2.0, 2.7, 3.4, 4.1, 4.8 and British Journal of Anaesthesia

2 Germon et al. Fig 1 Schematic diagram illustrating the experimental arrangement. 5.5 cm (channels 1 8, respectively) from the emitter, and arbitrary numeric baseline values for O 2 Hb, HHb, thb and Cyt are produced. Changes in the intensity of NIR light detected by the photodiodes are directly proportional to changes in chromophore concentration multiplied by the photon path length through the tissue. The units used are therefore arbitrary, but directly proportional to changes in the tissue concentration of the chromophores in mol litre 1. Cerebral oligaemia The NIRS sensor was placed on the left side of the forehead of 10 healthy, supine volunteers (all men, age range 26 39, mean 32.5 yr) above the temporal crest, thus avoiding the temporalis muscle and overlying the frontal projection of the middle cerebral artery (MCA) (Fig. 1). The left MCA blood flow velocity (MCAV) was determined using transcranial Doppler (TCD, PCDop 842, Scimed, Bristol, UK). End-tidal carbon dioxide concentration (E CO2 ), heart rate and Sp O2 were monitored (Hewlett Packard M1025B gas analyser, Palo Alto, CA, USA) and intermittent noninvasive measurements of mean arterial pressure were made (Dinamap 8100, Critikon, Tampa, FL, USA). Subjects breathed 5% carbon dioxide in oxygen through a mouth piece with a noseclip applied. When E CO2 was stable for 2 min, the study variables were measured five times at 5-s intervals. After reverting to breathing air, subjects were asked to hyperventilate in time to a metronome until a stable hypocapnic state had been achieved and the five measurements of each study variable were repeated. Scalp hyperaemia Using an identical sensor position in the same subjects breathing room air, the NIRS variables were monitored for 3 min before inflation of a pneumatic scalp tourniquet applied just above the supraorbital ridge to a pressure of 200 mm Hg, also for 3 min (Fig. 1). The tourniquet was then released and recording continued during a further 3 min of reactive hyperaemia. Systolic arterial pressure, recorded non-invasively from the right arm, was at least 50 mm Hg below the tourniquet pressure in all subjects throughout the study. Data analysis and statistical methods The absolute change in each chromophore detected in each channel between baseline and test conditions was illustrated using box and whisker plots (180 data points for each channel under each experimental condition, Systat version 5, Systat Inc., Evanston, IL, USA). However, the Beer Lambert law states that attenuation of light is proportional to the distance travelled by the incident light through the absorbing medium whereas the algorithm used by the spectrometer in this study assumes a photon path length of 1 at each emitter detector separation. Thus in a homogenous medium, a given change in chromophore concentration appears to increase as emitter detector separation increases. To compare the changes detected at increasing emitter detector separation, a correction for increasing path length must be made. We assumed that photon path length increases linearly with increasing emitter detector separation 5 and divided the absolute change in each channel by the emitter detector spacing for that channel to give the attenuation change per unit photon path length, which in turn is proportional to the chromophore concentration change per unit photon path length. The signal change between 1 min of baseline and 1 min of maximal scalp hyperaemia and between hyper- and hypocapnia recorded in each channel was divided by the respective interoptode distance so that results are proportional to the change in chromophore concentration per unit path length. The data recorded from each individual was then standardized using the formula: Z (X ((A B)/d))/SD where Z standardized change per unit photon path length; X mean absolute change across all channels; A initial condition; B test condition; d emitter detector separation; 832

3 Cerebral near infrared spectroscopy Table 1 Mean (SD) physiological variables during hypercapnia and hypocapnia Hypercapnia Hypocapnia P Heart rate (beat min 1 ) 63.2 (9) 70.3 (10) MAP (mm Hg) 97.4 (11) 92.2 (10) E CO2 (kpa) 6.6 (0.4) 4.2 (0.3) Sp O2 (%) (0) 99.6 (1) MCAV (cm s 1 ) 75.3 (10) 39.6 (9) and SD standard deviation of absolute change across all channels The calculations were made such that the more positive the value the greater the change whether the absolute concentration of the chromophore was increasing or decreasing. Repeated measures analysis of variance with two within-subject factors (emitter detector separation and intra- or extracranial change) was then used to establish the relationship between the size of the standardized concentration change in each channel and emitter detector separation. P 0.05 was considered significant for an isolated probability test. However, to allow for multiple comparisons across eight channels, a P value of 0.05/ was considered significant. A paired Student s t test was used to analyse significant differences between physiological variables in the two parts of the study. Results Cerebral oligaemia In response to a reduction in mean E CO2 from 6.7 to 4.3 kpa, mean MCAV decreased from 75 to 40 cm s 1. There was a small but significant (P 0.044) increase in heart rate (Table 1). Decreases in O 2 Hb and thb signals were observed and these were accompanied by reciprocal changes in the HHb signal (Fig. 2). Scalp hyperaemia There were no significant changes in systemic physiological variables in response to scalp ischaemia. There was an increase in O 2 Hb and thb and a reduction in HHb signal in all channels (Fig. 3). Standardized changes There was an increase in the magnitude of change per unit photon path length as emitter detector spacing increased for all three chromophores in response to cerebral oligaemia, although this was significant only for O 2 Hb and HHb (P ; P for thb) (Fig. 4). There was a significant decrease in the magnitude of change per unit photon path length for O 2 Hb and thb in response to scalp hyperaemia (P and P 0.003, respectively) but no significant change in HHb (P ) (Fig. 3). The relationship between standardized change and emitter detector separation was significantly different in response to scalp hyperaemia and cerebral oligaemia for Fig 2 Box and whisker plots illustrating mean and interquartile range (box) and 5th and 95th centiles (whiskers) of absolute change in oxyhaemoglobin (O 2 Hb), deoxyhaemoglobin (HHb) and total haemoglobin (thb) during cerebral oligaemia ( mol litre 1 total path length ). *Indicates outside values. all three chromophores (P for O 2 Hb and HHb, P for thb). Discussion Transmission cerebral NIRS in adults, as described by Jobsis, 15 is impractical because of the low intensity of transmitted NIR light and poor signal-to-noise ratio. Reflectance-mode cerebral NIRS, in which an NIR light source is placed ipsilateral to the receiver, has been developed to overcome these problems. This approach assumes that the scattering and absorption of NIR light 833

4 Germon et al. Fig 3 Box and whisker plots illustrating mean and interquartile range (box) and 5th and 95th centiles (whiskers) of absolute change in oxyhaemoglobin (O 2 Hb), deoxyhaemoglobin (HHb) and total haemoglobin (thb) during scalp hyperaemia ( mol litre 1 total path length ). *Indicates outside values. in the human head is consistent and approximates to mathematical and in vitro models of the behaviour of NIR photons in layered media. 1 5 If the assumptions are correct, mean light path of photons travelling from an emitting to a receiving optode, both of which are located on the surface of a sphere, should describe an ellipse whose mean depth is proportional to separation of the optodes. To use these theoretical principles for monitoring cerebral oxygenation, two fundamental questions must be answered. The first relates to the shape of the photon path and is the subject of this investigation. Does the mean path described by light detected on the surface of the scalp describe an ellipse, and if so what is the optimal emitter detector Fig 4 Mean (SEM) standardized change in oxyhaemoglobin (O 2 Hb), deoxyhaemoglobin (HHb) and total haemoglobin (thb) during scalp ischaemia and cerebral oligaemia for each channel. separation to achieve spatial resolution? Despite increasingly complex models, none has approached the complexity of the adult head. There are five histological layers in the scalp alone. in addition to the cortical and cancellous bone of the skull and dura. The heterogeneity of these tissues, with variable scattering properties and chromophore content, in addition to the multiple tissue interfaces at which variable reflection and refraction can occur, create the potential for the behaviour of light in vivo to differ from that predicted by simple models. 16 When the spectroscopic data have been spatially resolved, the next question is that of calibration and quantification. Calibration of an individual head would require measure- 834

5 Cerebral near infrared spectroscopy ment of light intensity on the surface of the scalp with the head devoid of any of the chromophores we wish to measure. 16 This is clearly impossible to achieve and therefore, absolute concentrations of chromophores cannot be measured. In order to measure changes in chromophore concentrations, the photon path length through the tissue of interest (the partial path length), the brain, must be known. Several methods of measuring NIR photon path length through tissue have been described but none can determine the fraction of the photon path which traverses cerebral tissue compared with extracerebral tissue Thus these methods may provide values for a photon path length which are inappropriate to the calculation of cerebral chromophore concentrations. We have assumed that the mean light path of photons travelling from the emitting optode to the closest receiving optode passes almost entirely through extracranial tissues and is thus mainly sensitive to changes in extracranial chromophore content. As the interoptode distance is increased, a progressively greater proportion of the light path traverses cerebral tissue and thus changes in cerebral chromophore content have a greater effect. The relative contributions of extracranial and intracranial chromophores to absorption of NIR light may be compared by using the signal change per unit path length (the absolute signal change divided by the emitter detector spacing). For an extracranial change in chromophore content, the signal change per unit path length is greatest in the proximal channels and decreases progressively in the distal channels. The converse is true for intracranial change. We sought to confirm or refute the theoretical model described above. The results of the cerebral oligaemia study followed those predicted by the model for a change in the concentration of an intracranial chromophore (increasing with emitter detector separation) (Fig. 4) while the results of the scalp hyperaemia study followed those predicted for an extracranial chromophore (decreasing with emitter detector separation) (Fig. 4). Thus these results strongly support the theoretical model. However, it is noteworthy that even in channel 1 (0.7 cm emitter receiver distance), significant changes occurred during cerebral oligaemia. Either this channel is interrogating cerebral tissue, which is unlikely, or the cutaneous and subcutaneous vascular beds of the scalp are more sensitive to changes in arterial PCO 2 than previously thought. In this case, cerebrovascular reactivity studies using NIRS which do not correct for blood flow changes in superficial tissues may be subject to substantial errors. The scalp tourniquet has no effect on MCAV in conscious subjects (unpublished data), strongly suggesting that no change occurs in cerebral blood flow and oxygenation. That the hyperaemia takes place in the scalp alone is shown by the changes in O 2 Hb and thb which are predicted for an extracranial change. The absence of a sustained reduction in the HHb change with increased emitter detector separation presumably reflects the relatively high scalp blood flow relative to its metabolic requirements. Under these circumstances the tissue concentration of HHb would be low. Reactive hyperaemia would produce an increase in thb accounted for by an increased volume of arteriolar and capillary blood (O 2 Hb) with little change in HHb. By the same token, the increase in the change in thb with emitter detector separation during cerebral oligaemia failed to reach significance (P ). Changes in cerebral blood volume in response to a reduction in arterial PCO 2 are the result of cortical arteriolar vasoconstriction 20 and are of a magnitude smaller than changes in CBF. 21 In this study, small reductions in thb only became significantly larger at the most distal detectors (Fig. 4) which supports the hypothesis that cerebral tissue is being interrogated. While these results support the general principle that an increase in optode separation increases the volume of cerebral tissue which is interrogated by NIRS, further information is required for absolute quantification of cerebral chromophore concentrations. Application of the Beer Lambert law requires, among other things, the true photon path length and the proportion of the path length in each tissue layer. Even at an optode separation of 5.5 cm, when the influence of intracerebral changes appears greatest and extracranial change least, most of the scattering of NIR photons may be taking place in superficial tissues because of high scattering coefficients and optical boundary effects. For example, up to 75% of incident NIR light is reflected from the dermis of skin. 22 Thus superficial tissue may account for the majority of NIR attenuation and path length changes. This interpretation is based on the assumption that there is a linear relationship between interoptode separation and NIR light path length whose linear constant has been termed the differential path length factor (DPF). 23 This assumption is true only if NIR light scattering in tissues becomes random within a few millimetres 24 of the source and differences in scattering characteristics of different tissues and their interfaces are negligible. Monte Carlo predictions are for DPF to increase as interoptode separation increases 25 whereas in vivo studies suggest that DPF decreases as optode separation increases, becoming constant above 2.5 cm. 26 The reasons for this are not clear: Monte Carlo predictions may be flawed; the different optode separations may interrogate tissues with different scattering characteristics or the acceptance angle of the detectors used in the in vivo studies may have been limited. 25 If the latter is true, it is likely that the real DPF remains constant below an emitter detector spacing of 2.5 cm. Multiple detector NIRS may overcome some of these problems and has been used in the design of cerebral oximeters in an attempt to provide spatial resolution (i.e. the ability to separate oxygenation changes in the scalp and skull from those occurring in cerebral tissue) These oximeters have a single light source and two or three detectors at different distances from the emitter. It is assumed that NIR light follows superficial and deep arcs 835

6 Germon et al. through the cranium whose mean depths are proportional to the separation between the emitting and receiving optodes. Since these arcs should have an almost identical path length through extracerebral tissues, subtraction of the absorption characteristics of the superficial arc from the deep arc should reflect changes taking place in cerebral tissue alone. 27 The first such device had receivers placed 1 cm and 2.7 cm from the emitter. This model was found to be insensitive to a doubling of cerebral blood flow in response to hypercapnia in anaesthetized patients 7 and did not reflect jugular venous bulb oxygen saturation measurements in patients undergoing cardiopulmonary bypass. 8 These observations suggest that the device was actually sensitive to extracerebral rather than cerebral haemoglobin saturation. Optode spacing was subsequently increased to 3 cm and 4 cm in an attempt to increase the volume of cerebral tissue interrogated. This updated version has been shown to be sensitive to hypoxia deep to the scalp 29 but it remains sensitive to oxygenation changes in the scalp and has not been shown to reliably separate changes in the two compartments. 30 More generally, while appropriate trends in group data can be demonstrated using cerebral NIRS, to date it has not been demonstrated that a NIR cerebral oximeter can reliably estimate changes in cerebral oxygenation in an individual subject This study has confirmed that increasing emitter detector separation increases sensitivity to cerebral oxygenation while decreasing sensitivity to extracerebral oxygenation. Our results suggest that emitter detector separations should be greater than those currently being used in commercially available oximeters; a separation of 5.5 cm may increase sensitivity to cerebral attenuation changes. At the same time, the effect of an extracranial change seems to plateau at cm (Fig. 4). If this is so, optimal sensor geometry for achieving spatial resolution would have a proximal detector at least 4.8 cm from the emitter and the second receiver as far away from the emitter as signal to noise ratios and practical constraints would allow. Acknowledgements T. J. G. was supported by the Stroke Association. We thank Ms C. Swinburn and Dr B. Reeves, Research and Development Unit, Bristol Royal Infirmary, for statistical advice. This work was presented in part to the XVII International Symposium on Cerebral Blood Flow and Metabolism, Cologne, Germany, July References 1 Bonner R, Nossal R, Havlin S, Weiss GH. Model for photon migration in turbid biological media. J Opt Soc Am A 1987; 4: Nossal R, Bonner RF, Weiss G. Influence of path lenght on remote optical sensing of properties of biological tissue. Appl Opt 1989; 28: Cui L, Ostrander L. Assessment of layered optical properties in biological tissue from dual wavelength optical reflectance measurement. Proceedings of the 12th Annual International Conference IEEE Engineering in Medicine and Biology Society 1990; 12: Nossal R, Kiefer J, Weiss GH, Bonner RF, Taitelbaum H, Havlin S. Photon migration in layered media. Appl Opt 1988; 27: Cui W, Kumar C, Chance B. Experimental study of migration depth for the photons measured at sample surface. SPIE 1991; 1431: Germon TJ, Evans PD, Manara AR, Barnett NJ, Wall P, Nelson RJ. Sensitivity of near infrared spectroscopy to cerebral and extracerebral oxygenation changes is determined by emitter detector separation. J Clin Monit Computing 1998; 14: Grubhofer G, Lassnigg A, Manlik F, Marx E, Trubel W, Hiesmayr M. The contribution of extracranial blood oxygenation on nearinfrared spectroscopy during carotid thrombendarterectomy. Anaesthesia 1997; 52: Harris DNF, Bailey SM. Near infrared spectroscopy in adults: Does the Invos 3100 really measure intracerebral oxygenation? Anaesthesia 1993; 48: Brown R, Wright G, Royston D. A comparison of two systems for assessing cerebral venous oxyhaemoglobin saturation during cardiopulmonary bypass in humans. Anaesthesia 1993; 48: Lewis SB, Myburgh JA, Thornton EL, Reilly PL. Cerebral oxygenation monitoring by near-infrared spectroscopy is not clinically useful in patients with severe closed head injury: a comparison with jugular bulb oximetry. Crit Care Med 1996; 24: Germon TJ, Kane NM, Manara AR, Nelson RJ. Possible limitations of cerebral oximetric monitoring in adults. Anaesthesia 1994; 49: Schwarz G, Litscher G, Kleinert R, Jobstmann R. Cerebral oximetry in dead subjects. J Neurosurg Anaesthesiol 1996; 8: Newton CR, Wilson DA, Gunnoe E, Wagner B, Cope M, Traystman RJ. Measurement of cerebral blood flow in dogs with near infrared spectroscpopy in the reflectance mode is invalid. J Cereb Blood Flow Metab 1997; 17: Germon TJ, Evans PD, Barnett N, Lewis TT, Wall P, Nelson RJ. Changes in tissue oxyhaemoglobin concentration measured using multichannel near infrared spectroscopy during internal carotid angiography. J Neurol Neurosurg Psychiatry 1997; 63: Jobsis FF. Noninvasive, infrared monitoring of cerebral and myocardial oxygen sufficiency and circulatory parameters. Science 1977; 198: Barnett N, Germon T. Theoretical principles and practical problems of cerebral near infrared spectroscopy. In: Litscher G, Schwarz G, eds. Transcranial Cerebral Oximetry. Lengerich: Pabst Scientific Publishers, 1997; Lakowicz JR, Berndt K. Frequency domain measurements of photon migration in tissues. Chem Phys Lett 1990; 166: Matcher SJ, Cope M, Delpy DT. Use of the water absorption spectrum to quantify tissue chromophore concentration changes in near infrared spectroscopy. Phys Med Biol 1994; 39: Cope M, Delpy DT, Wray S, Wyatt JS, Reynolds EO. A CCD spectrometer to quantitate the concentration of chromophores in living tissue utilising the absorption peak of water at 975 nm. Adv Exp Med Biol 1989; 248: Symon L. Pathological regulation in cerebral ischaemia. In: Wood JH, ed. Cerebral Blood Flow, Physiologic and Clinical Aspects. New York: McGraw-Hill, 1987; Kety SS, Schmidt CF. The effects of altered arterial tensions of carbon dioxide and oxygen on cerebral blood flow and cerebral oxygen consumption of normal young men. J Clin Invest 1948; 27: Parrish JA. Biochemistry and Physiology of Skin. Goldsmith LA, ed. Oxford: Oxford University Press, 1983; 2 23 Delpy DT, Cope M, van der Zee P, Arridge S, Wray S, Wyatt J. Estimation of optical pathlength through tissue from direct time of flight measurement. Phys Med Biol 1988; 33:

7 Cerebral near infrared spectroscopy 24 Svaasand LO, Ellingsen. Optical properties of human brain. Photochem Photobiol 1983; 38: van der Zee P, Arridge SR, Cope M, Delpy DT. The effect of optode positioning on the optical pathlength in near infrared spectroscopy of brain. Adv Exp Med Biol 1990; 277: van der Zee P, Cope M, Arridge SR, et al. Experimentally measured optical pathlengths for the adult head, calf and forearm and the head of the newborn infant as a function of interoptode spacing. Adv Exp Med Biol 1992; 316: McCormick PW, Stewart M, Goetting MG, Balakrishnan. Regional cerebrovascular oxygen saturation measured by optical spectroscopy in humans. Stroke 1991; 22: Kirkpatrick PJ, Smielewski P, Lam JMK, Al-Rawi P. Use of near infrared spectroscopy for the clinical monitoring of adult brain. J Biomed Opt 1996; 1: Germon TJ, Kane NM, Manara AR, Nelson RJ. Near-infrared spectroscopy in adults: the effects of extracranial ischaemia and intracranial hypoxia on the estimation of cerebral oxygenation. Br J Anaesth 1994; 73: Germon TJ, Young AER, Manara AR, Nelson RJ. Extracerebral absorption of near infrared light influences the detection of increased cerebral oxygenation monitored by near infrared spectroscopy. J Neurol Neurosurg Psychiatry 1995; 58: Kirkpatrick PJ, Smielewski P, Czosnyka M, Menon DK, Pickard JD. Near-infrared spectroscopy use in patients with head injury. J Neurosurg 1995; 83: Smielewski P, Kirkpatrick P, Minhas P, Pickard JD, Czosnyka M. Can cerebrovascular reactivity be measured with near infrared spectroscopy? Stroke 1995; 26: