Open Access
Add to BibSonomyAbstract
Noninvasive monitoring of cerebral blood oxygenation with an optoacoustic technique offers advantages over current invasive and noninvasive methods. We report the results of in vivo studies in the sheep superior sagittal sinus (SSS), a large central cerebral vein. We changed blood oxygenation by increasing and decreasing the inspired fraction of oxygen (FiO2). Optoacoustic measurements from the SSS were performed at wavelengths of 700, 800, and 1064 nm using an optical parametric oscillator as a source of pulsed near-infrared light. Actual oxygenation of SSS blood was measured with a CO-Oximeter in blood samples drawn from the SSS through a small craniotomy. The amplitude of the optoacoustic signal induced in the SSS blood at λ = 1064 nm closely followed the changes in blood oxygenation, at λ = 800 nm was almost constant, and at λ = 700 nm was changing in the opposite direction, all in accordance with the absorption spectra of oxy- and deoxyhemoglobin. The optoacoustically predicted oxygenation correlated well with actual blood oxygenation in sheep SSS (R 2 = 0.965 to 0.990). The accuracy was excellent, with a mean difference of 4.8% to 9.3% and a standard deviation of 2.8% to 4.2%. To the best of our knowledge, this paper reports for the first time accurate measurements of cerebral venous blood oxygenation validated against the “gold standard” CO-Oximetry method.
©2009 Optical Society of America
1. Introduction
Patients with severe traumatic brain injury or undergoing craniotomy or cardiac surgery are at risk for cerebral ischemia and could benefit from a continuous noninvasive monitor of cerebral venous blood oxygenation. There is strong clinical evidence that cerebral venous hemoglobin desaturation can detect cerebral ischemia and guide therapeutic interventions [1,2]. Currently, both invasive and noninvasive methods for assessing cerebral oxygenation are in clinical use [3,4]. One of the invasive methods measures regional brain tissue pO2 (partial oxygen tension) with an intracranial probe inserted into the brain parenchyma, and another one utilizes catheterization of the internal jugular vein and measurement of oxygenation in the jugular bulb. Both methods have technical limitations, can cause complications due to the high invasiveness of the procedures, and do not actually provide the necessary information on how well the brain tissues are supplied with oxygen. An intracranial probe can supply only local values of brain tissue pO2, while a catheter in the jugular bulb often provides false values when it is proximally displaced or touches the vessel wall. There have been promising attempts to quantify cerebral oxygenation with the use of near-infrared (NIR) spectroscopy [4,5], a noninvasive optical technique that detects light diffusively scattered from a target. However, this technique can provide only volume-averaged oxygenation of brain tissue and cannot distinguish between venous and arterial blood.
We proposed to use an optoacoustic technique for monitoring of blood oxygenation [6,7]. The technique detects ultrasound waves generated in tissue as pulsed NIR radiation is absorbed, resulting in thermo-elastic expansion of the irradiated volume. A series of in vivo experiments conducted previously [8-10] confirmed the clinical potential of this technique. Photoacoustic tomography was tested for quantitative imaging of oxyhemoglobin saturation in the cerebral vasculature of small animals [11]. Despite the encouraging results, the imaging depth in these experiments was not sufficient for human applications. Moreover, validation of cerebral blood vessel oxygenation in small animals with “gold standard” blood sampling is challenging due to the small size of the vessels.
Here we report the results of in vivo, multi-wavelength, optoacoustic measurements in the superior sagittal sinus (SSS) of sheep validated against the “gold standard” CO-Oximetry method. The SSS, a large central cerebral vein located on top of the brain, collects blood from both hemispheres before emptying into the internal jugular vein. Hemoglobin saturation in the cerebral venous blood below 50% (normal range is 55-75% [12]) is an indication of brain ischemia, a condition associated with poor clinical outcome and requiring immediate intervention. Compared to our previous proof-of-concept study in sheep, in which we used a single wavelength [8], the present study incorporates a novel multi-wavelength optoacoustic system with greater sensitivity and better specificity for measurements in the SSS. The measurements at three wavelengths in the NIR spectral range allowed not only for monitoring of relative changes in blood oxygenation (previously demonstrated with the single-wavelength technique [8]), but also for quantitative measurement of blood oxygenation. A similar experimental setup was successfully used for optoacoustic measurements in the sheep external jugular vein [10], a large neck vein covered with a thick layer of soft tissue, but no quantitative data on blood oxygenation were compared. In this paper, we performed quantitative, accurate measurements of SSS blood oxygenation, through the intact skull, which is more challenging because both the NIR input and the acoustic output is attenuated by the skull.
2. Materials and methods
For our optoacoustic measurements we used a compact optical parametric oscillator (OPO; Opolette 532 II, Opotek Inc., Carlsbad, CA) as a source of pulsed tunable NIR radiation (range of available wavelengths 680-2440 nm, pulse duration - 10 ns, repetition rate - 20 Hz). The custom-made optoacoustic probe (Fig. 1) incorporated a single-element broadband unfocused piezoceramic transducer (Keramos Inc., Indianapolis, IN) for detection of generated ultrasonic waves and a fiber-optic light-delivery system (4 fibers with a core diameter of 1 mm around the transducer). The transducer had a bandwidth of 3 MHz with a resonance frequency of 2 MHz and a sensitivity of 40 μV/Pa. It had an area of 10 mm2 and a thickness of 1 mm. The whole probe was 5 cm high and 1.5 cm in diameter. The optoacoustic signals were amplified with a low-noise 20-dB preamplifier (Onda Corp., Sunnyvale, CA) and a low-noise 40-dB amplifier (Analog Modules Inc., Longwood, FL) and then digitized with a 100-MHz 8-bit digitizer (NI-5112, National Instruments Corp., Austin, TX). The digitized signals were processed in real time by a laptop computer, which was also used to control the OPO with manufacturer-supplied software.
We calibrated our system using a tissue phantom with a cylindrical cavity simulating a blood vessel. We molded a tissue-like turbid slab from liquid plastic polyvinyl chloride-plastisol (M-F Manufacturing Inc., Fort Worth, TX) [13] with white plastic color from the same manufacturer added for turbidity (the resulting effective attenuation coefficient was μeff = 1.5 cm-1 at 1064 nm, which is close to the reported values of μeff for soft tissues in the NIR spectral range [14]). The diameter and the depth of the cavity were 24 and 9 mm, respectively. We positioned the slab vertically with the cavity opening facing up and filled the cavity with fresh heparinized arterial sheep blood. Then, we gradually deoxygenated blood by adding small amounts of sodium dithionate (about 30 mg at a time) and performed optoacoustic measurements at three different wavelengths (700, 805, and 1064 nm) after every addition. Simultaneously, we measured actual blood oxygenation with a standard CO-Oximeter (IL 682, Instrumentation Laboratories, Lexington, MA). We attached the optoacoustic probe to the side of the slab with a thin layer of ultrasound gel for acoustic coupling. The blood was stirred during the experiment to avoid sedimentation.
We chose the three wavelengths because they lay in characteristic regions of the blood absorption spectrum [15-17]. The wavelength of 805 nm is close to the isobestic point, where optical absorption of blood does not depend on oxygenation. At 1064 nm, melanin and water have low absorption [18,19], and light scattering in tissues is relatively low as well [20,21]. At 1064 nm the blood absorption increases with oxygenation, while at 700 nm it decreases.
We then evaluated the performance of the system during the in vivo experiments in 5 merino sheep. We aimed to detect signals generated in the SSS through the intact skull. Although the thickness of the skull in adult sheep (5-6 mm) is smaller than that in adult humans (8-10 mm), the bone over the SSS is thick enough to demonstrate the feasibility of measuring cerebral venous oxygenation in humans using the optoacoustic technique.
The study was approved by the Institutional Animal Care and Use Committee of the University of Texas Medical Branch (UTMB). The animals were housed in the Animal Resources Center of UTMB under the daily supervision of full-time veterinarians. During the experiments the sheep were anesthetized with a 1.5% to 2.0% isoflurane. The animals were kept in a prone position. We performed tracheal intubation for the delivery of both isoflurane and a gas mixture of oxygen and nitrogen to the animal. By varying the fraction of oxygen (FiO2) in the inhaled gas mixture in the range from 100% to 10%, we were able to lower venous blood oxygenation down to 10-15% and then increase it up to 100%. In each sheep we performed 2-3 cycles.
The scalp was cut along the midline and retracted to the sides. A small craniotomy was made close to the site of the optoacoustic measurements. We inserted a 22-gauge catheter into the SSS through this craniotomy to sample blood immediately after every optoacoustic measurement and to measure the actual value of cerebral venous hemoglobin saturation at the moment with a CO-Oximeter.
We continuously monitored the vital signs of the animals. Blood pressure was controlled using a catheter inserted into the femoral artery. A pulse oximeter attached to lip, tongue, or ear provided data on arterial blood oxygenation. Heartbeat rate and cardiac rhythm were monitored by electrocardiography. All experiments were terminal. At the end of the experiment the sheep were given saturated KCl solution intravenously (1 mL/kg) under deep isoflurane anesthesia.
The optoacoustic probe was placed in contact with the exposed skull over the SSS (a thin layer of ultrasound gel ensured acoustic coupling). To obtain SSS signals with the highest amplitude, we used a 3D translation stage for the probe alignment. The laser fluence at the site of probing was less than 4 mJ/cm2 (well below the maximum permissible exposure for skin in this spectral range [22], which is 20 to 100 mJ/cm2 for our irradiation conditions).
By adjusting FiO2 in the inhaled gas mixture, we stabilized arterial oxygenation at varying levels and performed optoacoustic measurements during stable conditions. At each level of arterial oxygenation, we consecutively irradiated the surface of the skull with light of three wavelengths: 700 nm, 800/805 nm, and 1064 nm. We averaged 400 signals for every record to minimize the influence of electronic noise on the results. The set of measurements at three wavelengths required about 1.5-2 min. After that, a blood sample was drawn from the SSS.
3. Results
The optoacoustic signals from blood with different oxygenation levels in our SSS phantom are shown in Fig. 2 for λ = 700 nm (a) and 1064 nm (b). The amplitude of the signal clearly is dependent on blood oxygenation and wavelength: at 1064 nm it increases with blood oxygenation, while it decreases at 700 nm. We divided the peak-to-peak amplitudes of the signals at 700 nm and 1064 nm by the amplitude of the signal at 805 nm for the same level of oxygenation. The purpose was to minimize interference of factors other than blood oxygenation (inhomogeneity of the blood and possible sedimentation). The resulting data are presented in Fig. 3(a), which demonstrates a linear increase of the normalized amplitude at 1064 nm (red circles) and a linear decrease of the amplitude measured at 700 nm (blue triangles). These results were in good agreement with the absorption spectra of oxy- and deoxygenated blood. Each straight line is a least-square fit to the corresponding set of data points. The correlations were high: R 2 = 0.984 and 0.980 for 700 and 1064 nm, respectively. We then divided the fitting line for 1064 nm by the fitting line for 700 nm (dividing the corresponding data point sets by each other and performing polynomial fit to new data points yields practically the same dependence). The resulting line shown in Fig. 3(b) was used as a calibration curve in our further experiments.
Fig. 2. Optoacoustic signals recorded from the SSS phantom with blood at different oxygenation at the wavelengths of 700 nm (a) and 1064 nm (b).
Fig. 3. (a) Optoacoustic signal amplitudes at 700 nm and 1064 nm normalized by the amplitude at 805 nm. (b) Calibration curve obtained by dividing the fitting line for 1064 nm by the fitting line for 700 nm in (a).
Figure 4 shows typical optoacoustic signals from sheep skull over the SSS at wavelengths 700 nm (a), 800 nm (b), and 1064 nm (c). For each wavelength, the signals for three oxygenation levels are shown: 91.5% (blue line), 21.5% (green line), and 60%, at the end of the cycle (red line). The leftmost peak in each signal is generated on the skull surface (major chromophores within the skull bone in the NIR spectral range are lipids, water, hemoglobin, and collagen [23]), while the second prominent peak is due to absorption by hemoglobin in the SSS blood. All signals in Fig. 4 are normalized for the amplitude of the skull peak in order to eliminate the uncertainty associated with energy instability of the OPO system. As the blood content of the upper layers of skull is extremely low, the amplitude of this peak is not dependent on blood oxygenation level and can be used for normalization.
Fig. 4. Optoacoustic signals from sheep skull over the SSS at wavelengths of 700 nm (a), 800 nm (b), and 1064 nm (c) for different levels of blood oxygenation.
One can see that the amplitude of the signal from the SSS decreases with blood oxygenation at the wavelength of 700 nm, stays almost constant at 800 nm, and increases with oxygenation at 1064 nm. These trends are in accordance with the dependence of the absorption coefficient of blood on oxygenation level at the specified wavelengths and with our in vitro results. The slight changes in the SSS signal at 800 nm could not be caused by the change in blood oxygenation, as this wavelength is close to the isobestic point. Rather, we speculate that those changes resulted from changes in total hemoglobin concentration of blood, motion artifacts, and variations in the SSS diameter accompanying changes in cerebral blood flow produced by hypoxemia. The same factors might also influence the measurements at two other wavelengths (700 nm and 1064 nm) performed at practically the same time. Thus, for every measurement, we divided the peak-to-peak amplitudes of the SSS peak at 700 nm and 1064 nm by the amplitude at 800 nm to eliminate the influence of these confounding factors.
The normalized peak-to-peak amplitudes measured during two cycles of blood oxygenation variation are presented in Fig. 5 for 700 nm (a) and 1064 nm (b). The amplitude at 1064 nm closely followed actual SSS blood oxygenation measured with the CO-Oximeter, whereas at 700 nm it changed in the direction opposite to the oxygenation change. Moreover, the amplitudes were linearly dependent on blood oxygenation with high correlation coefficients (R 2 = 0.88 and 0.95 for 700 nm and 1064 nm, respectively; Fig. 6).
Fig. 5. Directly measured SSS blood oxygenation (black, both a and b) and normalized optoacoustic signal amplitude at 700 nm (blue, a) and 1064 nm (red, b) during 2 cycles of changes in blood oxygenation.
Fig. 6. Optoacoustic signal amplitudes measured from the sheep SSS in vivo at 700 nm (blue triangles) and 1064 nm (red circles) corrected by the change in the amplitude at 805 nm. The linear fit to the data is presented as well.
We predicted the SSS blood oxygenation by using the calibration curve (Fig. 3(b)) and the ratio of the amplitudes at 1064 nm and 700 nm measured from the sheep SSS in vivo. Figure 7(a) shows the optoacoustically predicted SSS blood oxygenation vs. actual oxygenation measured with the CO-Oximeter for the data in Figs. 5 and 6. The correlation between the optoacoustically predicted and actual oxygenation was high (R 2 = 0.965). To estimate the bias (<Δ>) and standard deviation (SD) of the optoacoustic measurements, we calculated the difference Δ between the optoacoustically predicted and actual oxygenation measured with the CO-Oximeter (Fig. 7(b)). These data (<Δ> = -9.3%, SD = 4.2%) demonstrate that the accuracy of the optoacoustically measured blood oxygenation approaches that of invasive measurements.
Figures 8(a) and 8(b) show SSS blood oxygenation and optoacoustic signal amplitude at 700 nm and 1064 nm, respectively, that were measured from a sheep that had minimal incidental motion (more deeply anesthetized). The signal amplitude correlated better with oxygenation (R 2 = 0.99, Fig. 9(a)) than that measured in a sheep with some incidental motion (Fig. 7(a)). This resulted in a lower bias and standard deviation (4.8% and 2.8%, respectively, Fig. 9(b)) compared to that for the sheep in which incidental motion occurred (Fig. 7(b)).
Fig. 7. (a) Correlation between optoacoustically predicted and actual SSS blood oxygenation during two cycles of changes in blood oxygenation. (b) Standard deviation and bias of the difference between predicted and actual oxygenation.
Fig. 8. SSS blood oxygenation (black, both a and b) and normalized optoacoustic signal amplitude at 700 nm (blue, a) and 1064 nm (red, b) measured in a sheep with low motion artifacts.
Fig. 9. (a) Correlation between optoacoustically predicted and actual SSS blood oxygenation for the measurement with low motion artifacts. (b) Standard deviation and bias of the difference between predicted and actual oxygenation.
By analyzing our in vivo data, we concluded that the confounding variables in optoacoustic measurements in sheep were due to motion artifacts, total hemoglobin concentration variations, and changes in the SSS diameter (dilation, etc.). Since the probe has high lateral resolution (1.5 mm) and the diameter of the sheep SSS (2-3 mm) is similar to the lateral resolution, variation in the SSS diameter or slight displacement of the probe (up to 1 mm) can change the signal detected from the SSS. Because the adult human SSS is much wider (8-11 mm) than that of the sheep, we expect that measurements in humans will be less prone to these artifacts and thus will be more accurate.
Several technical modifications can be made to further improve the performance of our optoacoustic system. The use of high-power pulsed laser diodes that can operate at a very high repetition rate (up to tens kHz) will substantially reduce measurement time and thus increase the accuracy of measurements. In addition, laser diodes are inexpensive, light in weight, and compact, thereby facilitating the design of systems for clinical use.
4. Conclusions
We demonstrated that the optoacoustic signal from the sheep SSS is detectable through the skull. The amplitude of the SSS peak correlated well with wavelength and blood oxygenation changes caused by variation of FiO2: the SSS signal at 700 nm decreased, at 800 nm was almost constant, and at 1064 nm increased with oxygenation. All these changes are in accordance with the absorption spectra of oxy- and deoxygenated blood. We were able to monitor changes in optoacoustic signal amplitude for extended periods of time (up to 5 hours). The normalization of the signal amplitudes at 700 nm and 1064 nm by the amplitude at 800 nm improves their correlation with oxygenation, as it minimizes the influence of confounding factors. The calibration curve obtained from in vitro experiments in the SSS phantom was used to predict blood oxygenation in sheep measurements. The predicted values strongly correlated with actual SSS blood oxygenation that resulted in high accuracy of optoacoustic measurements (R 2 = 0.99, <Δ> = 4.8%, SD = 2.8% in sheep with low motion artifacts).
Acknowledgments
This work is supported in part by the Moody Center for Traumatic Brain & Spinal Cord Injury Research/Mission Connect of the University of Texas Medical Branch, the National Institutes of Health (Research Grant # R01 EB00763 from the National Institute of Biomedical Imaging and Bioengineering and Research Grant # R01 NS044345 from the National Institute of Neurological Disorders and Stroke), and John Sealy Memorial Endowment Fund for Biomedical Research. Both Drs. Prough and Esenaliev are co-owners of Noninvasix, Inc., a UTMB-based startup that has licensed the rights to optoacoustic monitoring technology.
References and links
1. M. Soehle, M. Jaeger, and J. Meixensberger, “Online assessment of brain tissue oxygen autoregulation in traumatic brain injury and subarachnoid hemorrhage,” Neurol. Res. 25, 411–417 (2003). [CrossRef] [PubMed]
2. J. M. Murkin, “Perioperative detection of brain oxygenation and clinical outcomes in cardiac surgery,” Semin. Cardiothorac. Vasc. Anesth. 8, 13–14 (2004). [CrossRef] [PubMed]
3. W. J. Stevens, “Multimodal monitoring: head injury management using SjvO2 and LICOX,” J. Neurosci. Nurs. 36, 332–339 (2004). [CrossRef]
4. F. F. Jobsis, “Noninvasive, infrared monitoring of cerebral and myocardial oxygen sufficiency and circulatory parameters,” Science 198, 1264–1267 (1977). [CrossRef] [PubMed]
5. J. H. Choi, M. Wolf, V. Toronov, U. Wolf, C. Polzonetti, D. Hueber, L. P. Safonova, R. Gupta, A. Michalos, W. Mantulin, and E. Gratton, “Noninvasive determination of the optical properties of adult brain: near-infrared spectroscopy approach,” J. Biomed. Opt. 9, 221–229 (2004). [CrossRef] [PubMed]
6. R. O. Esenaliev, K. V. Larin, I. V. Larina, M. Motamedi, and D. S. Prough, “Optoacoustic technique for non-invasive continuous monitoring of blood oxygenation,” In Biomedical Topical Meetings (Optical Society of America, Washington DC, 2000), pp. 272–274.
7. R. O. Esenaliev, I. V. Larina, K. V. Larin, D. E. Deyo, M. Motamedi, and D. S. Prough, “Optoacoustic technique for noninvasive monitoring of blood oxygenation: a feasibility study,” Appl. Opt. 41, 4722–4731 (2002). [CrossRef] [PubMed]
8. Y. Y. Petrov, D. S. Prough, D. E. Deyo, M. Klasing, M. Motamedi, and R. O. Esenaliev, “Optoacoustic, noninvasive, real-time, Continuous monitoring of cerebral blood oxygenation: an in vivo study in sheep,” Anesthesiology 102, 69–75 (2005). [CrossRef]
9. Y. Y. Petrov, I. Y. Petrova, I. Patrikeev, R. O. Esenaliev, and D. S. Prough, “Multiwavelength optoacoustic system for noninvasive monitoring of cerebral venous oxygenation: a pilot clinical test in the internal jugular vein,” Opt. Lett. 31, 1827–1829 (2006). [CrossRef] [PubMed]
10. H. P. Brecht, D. S. Prough, Y. Y. Petrov, I. Patrikeev, I. Y. Petrova, D. J. Deyo, I. Cicenaite, and R. O. Esenaliev, “In vivo monitoring of blood oxygenation in large veins with a triple-wavelength optoacoustic system,” Opt. Express 15, 16261–16269 (2007), http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-15-24-16261. [CrossRef] [PubMed]
11. X. Wang, X. Xie, G. Ku, L. V. Wang, and G. Stoica, “Noninvasive imaging of hemoglobin concentration and oxygenation in the rat brain using high-resolution photoacoustic tomography,” J. Biomed. Opt. 11, 024015 (9 pages). [PubMed]
12. C. Metz, M. Holzschuh, T. Bien, C. Woertgen, R. Rothoerl, B. Kallenbach, K. Taeger, and A. Brawanski, “Monitoring of cerebral oxygen metabolism in the jugular bulb: reliability of unilateral measurements in severe head injury,” J. Cereb. Blood Flow Metab. 18, 332–343 (1998). [CrossRef] [PubMed]
13. G. M. Spirou, A. A. Oraevsky, I. A. Vitkin, and W. M. Whelan, “Optical and acoustical properties at 1064 nm of polyvinyl chloride-plastisol for use as a tissue phantom in biomedical optoacoustics,” Phys. Med. Biol. 50, N141–N153. [PubMed]
14. W. - F. Cheong, S. A. Prahl, and A. J. Welch, “A review of the optical properties of biological tissues,” IEEE J. Quantum Electron. 26, 2166–2185 (1990). [CrossRef]
15. B. L. Horecker, “The absorption spectra of hemoglobin and its derivatives in the visible and near infra-red regions,” J. Biol. Chem. 148, 173–183 (1943).
16. Y. Y. Petrov, D. S. Prough, D. J. Deyo, I. Y. Petrova, M. Motamedi, and R. O. Esenaliev, “In vivo noninvasive monitoring of cerebral blood oxygenation with optoacoustic technique,” in Proceedings of the 26th Intern. Conf. of IEEE EMBS, (Institute of Electrical and Electronics Engineers, New York, 2004), pp. 2052–2054.
17. J. Laufer, C. Elwell, D. Delpy, and P. Beard, “Pulsed near-infrared photoacoustic spectroscopy of blood,” in Photons Plus Ultrasound: Imaging and Sensing, A. A. Oraevsky and L. V. Wang, eds., Proc. SPIE5320, 57–68 (2004).
18. S. Jacques, “Optical absorption of melanin,” Oregon Medical Laser Center, http://omlc.ogi.edu/spectra/melanin/mua.html.
19. G. M. Hale and M. R. Querry, “Optical constants of water in the 200nm to 200μm wavelength region,” Appl. Opt. 12, 555–563 (1973). [CrossRef] [PubMed]
20. C. R. Simpson, M. Kohl, M. Essenpreis, and M. Cope, “Near-infrared optical properties of ex vivo human skin and subcutaneous tissues measured using the Monte Carlo inversion technique,” Phys. Med. Biol. 43, 2465–2478 (1998). [CrossRef] [PubMed]
21. V. Tuchin, Tissue optics: light scattering methods and instruments for medical diagnosis (SPIE Press, Bellingham, WA, 2000).
22. “ANSI Z136.1 - 2000” in American national standard for safe use of lasers (The Laser Institute of America, Orlando, FL, 2000).
23. P. Taroni, D. Comelli, A. Farina, A. Pifferi, and A. Kienle, “Time-resolved diffuse optical spectroscopy of small tissue samples,” Opt. Express 15, 3301–3311 (2007), http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-15-6-3301. [CrossRef] [PubMed]
