The human paranasal sinuses are air-filled cavities embedded in the cranial skeleton. The frontal sinuses are located behind the frontal bone, while the maxillary sinuses are located below the orbital floor in the cheek bones. The sphenoidal sinus, split by a partial wall, is located centrally in the cranium. The ethmodial cells are multifocal small cavities located on each side of the upper part of the nose cavity. The sinuses are lined with epithelium-bearing glands and cilia, which transport mucosa toward the channel, or ostium, connecting the sinus to the nasal cavity. The frontal and maxillary sinus ostia connect to the middle nasal meatus. The function of sinuses is still under discussion. They seem to be locations of nitric oxide production important for ventilation functions.1, 2 The anatomy of the head is shown in Fig. 1a ,3 with the sinuses schematically indicated; the frontal and maxillary sinuses are clearly discerned in the magnetic resonance image (MRI) shown in Fig. 1b.
Good ventilation seems to be a requisite for healthy sinuses. Infection in the cavities, sinusitis, is affecting a very large number of patients, in the U.S. alone about 37 million people are annually diagnosed.4 Narrow ostia or partial or full blockage can cause severe problems and chronic sinusitis. Nasal polyps can be one cause. Excessive mucosal and pus production are secondary signs of sinusitis. Infection is frequently treated with antibiotics, and decongestants, such as oxymetazoline and zylometazoline, are prescribed to improve ventilation. Surgical intervention with intended ostial widening is necessary in severe cases for achieving long-term improvement. Clearly, a precise diagnosis of sinusitis is of importance, particularly for deciding whether antibiotics should be administered or not. Common present diagnostics may include the use of ultrasound, capable of seeing from the echo structure if the cavity is air filled or not, and in severe cases the use of x-ray radiography. However, none of these techniques can assess the ventilation of the cavities or the composition of enclosed gas.
Ventilation studies for the paranasal sinuses have been pursued with intrusive and nonintrusive techniques. Among the former, cannulae have been used to measure pressure in the sinuses and in the nasal cavity and their interrelation. Nitrogen gas has also been injected as well as gaseous radioactive isotopes, in particular , with a half-life of 5.2 days. Scintillation cameras or single photon emission computerized tomography (SPECT) have been used to assess the transport of the gas. has also been administered noninvasively through breathing, followed by repeatedly performing the Valsalvas maneuver.5 Stable Xe gas, having a certain radio opacity, has also been used noninvasively, followed by repeated computerized tomographic (CT) recordings.6 Clearly, these methods expose the patients to a considerable radiological dose. Further, radioactive as well as stable Xe gas is very costly, as is the the use of SPECT and CT installations. Hyperpolarized has been used in MRI studies,7 also costly investigations. Thus, there is a need for a complementary technique avoiding these hazards and costs while still allowing a detailed assessment of ventilation and gas composition.
We demonstrate the possibility to statically and dynamically nonintrusively study human sinuses by observing the normal physiological gases of oxygen and water vapor. This is done by using tunable diode laser spectroscopy on the gases exhibiting narrow-band absorptive imprints in the near infrared spectral region. The present study extends and establishes the methodology, which was first described in two recent exploratory papers.8, 9
The new development became possible by taking advantage of optical spectroscopic techniques refined in the environmental monitoring of free gas, and an understanding of tissue optics, developed for purposes such as optical mammography and photodynamic therapy dosimetry (e.g., Ref. 10). We could thus demonstrate the possibility to investigate the frontal sinuses using tunable diode laser spectroscopy employing phase sensitive detection of wavelength-modulated absorption signals.8 The presence of molecular oxygen was measured close to in a backscattering detection geometry. To understand the behavior of the expected signals, a phantom was first investigated. It consisted of two Delrin plates separated by a variable air gap, where the first plate represented the frontal bone, the air space the frontal sinus, and the second plate the underlying tissue. Some preliminary results from a human frontal sinus were obtained. These measurements have been modeled in Monte Carlo simulations, and the possibility of imaging using a scanned probe was discussed.11 Since then, the experimental techniques have been improved by using pigtailed lasers and balanced detection to suppress noise.12 This allowed us to study gas exchange between the maxillary sinuses and the nasal cavity by using a transmission detection geometry. The fiber was placed inside the mouth on the palate and the detector externally on the cheek bone. For a short time, the nasal cavity was flushed through the nostril with nitrogen gas and the decay of the oxygen signal was recorded on a healthy volunteer, resulting in the conclusion that the channels between the cavities were open, allowing free gas exchange. No such behavior could be observed on a volunteer with constantly recurring sinus problems.9
We demonstrate molecular oxygen measurements on maxillary sinuses for 11 healthy volunteers using the transmission detection geometry discussed. The measured values have been compared with values obtained using Monte Carlo simulations of a model representing the maxillary sinuses. To further investigate the clinical potential of the technique, other different possible detection geometries have been explored, allowing monitoring of both the maxillary and the frontal sinuses. A further gas—water vapor interrogated at —has also been studied, providing new information. Water vapor lines around were studied in a previous paper on drying processes.13 Since liquid water, always abundantly present in human tissue, also has an absorption peak at , we now choose a shorter wavelength where most of the liquid water absorption is eliminated to allow depth penetration. Investigations of the gas exchange of the sinuses and the nasal cavity in the different geometries have as well been performed using pure nitrogen gas flushed through the nostril.
In the next section, the experimental setup and the data analysis are presented. The gas monitoring procedure, including the different detection geometries used, are described, followed by an explanation of the model used in the Monte Carlo simulations. Results of the measurements of the maxillary and frontal sinuses using the different detection geometries are then presented and discussed. Finally, conclusions are drawn and suggestions for further work are put forward.
Materials and Methods
A schematic drawing of the experimental setup used in this study is shown in Fig. 2 . In the case of molecular oxygen detection, a single-mode pigtailed distributed feedback (DFB) diode laser (Nanoplus, Germany) operating at around with an output of about is used as a spectroscopic light source. When performing measurements on water vapor, the laser is changed to another unit operating around (Nanoplus, Germany) with similar output. The laser is scanned across an absorption line of the gas under study by supplying a saw-tooth ramp (Hz) to the laser driver current. To achieve wavelength modulation spectroscopy (WMS) with lock-in detection, the operating current is also modulated by a sinusoidal wave at , with a modulation index of about 2.2 in both gas cases. In the case of molecular oxygen, the R11Q12 absorption line at (vacuum wavelength) is used, and in the case of water vapor, the absorption line at (vacuum wavelength) is used [vibration; , rotation; , , ].
The light is fiber optically split 90/10(%) into two parts with a single-mode fiber-coupled beamsplitter (Laser2000, Sweden). The low-intensity part is directly guided to a silicon detector (UDT 10DP/SB). This part is later used in the analysis as the reference beam to obtain balanced detection for efficient noise subtraction.12 The other part is guided to different positions on the human head, and the multiply scattered light is detected externally on the human head according to different detection geometries further discussed in Sec. 2.3 on the gas monitoring procedure. In the case of molecular oxygen detection, a photomultiplier tube (PMT) (Hamamatsu 5070A) with a circular detection area of diam is used, and in the case of water vapor detection, a photodiode (Hamamatsu S3584-06) with similar detection area is used. Both detectors are mounted as handheld probes.
The two detected signals are each divided into two parts. They are sent to a computer-controlled digital oscilloscope, one directly, referred to as the direct signal, and one via a lock-in amplifier where the signal at twice the modulation frequency is recorded, referred to as the signal. WMS is often referred to as derivative spectroscopy, since the different harmonics are proportional to the corresponding derivative of the signal. Thus, in the case of detection at twice the modulation frequency, the signal looks like the second derivative of the direct absorption.
The signals are normalized with the direct signals to make them independent of the amount of light reaching the detector. The normalized sample signal is then subtracted with a processed version of the reference signal to achieve a balanced detection signal.12 An ideal absorption signal is then fitted to the balanced detection signal, and from this the peak-to-peak distance is estimated, referred to as the WMS value. For small absorptions, the WMS value is proportional to the absorbance, and thus to the product of the gas concentration and the path length traveled by the light. To transform the WMS value to a more meaningful quantity, the standard addition method is used.14 The standard addition procedure is to add known path lengths of ambient air in-between the light source and the detector. At each distance, the WMS value is measured. A so-called equivalent mean path length can then be estimated, corresponding to the air distance that the light has to travel in ambient air to obtain the same oxygen or water vapor absorption imprint. The calibration procedures for both gases were performed at room temperature of and a relative humidity of about 20%. It should be noted that one would expect 100% humidity in the sinus cavities due to liquid water in the surrounding tissue, with a temperature close to . At such a temperature, the water vapor pressure is about a factor of 2 higher than at .15 Thus, a particular water vapor signal from a sinus corresponds to about 1/10 of the surrounding air path length yielding the same signal. This factor is further discussed in Sec. 4.
Gas Monitoring Procedure
The maxillary and frontal sinuses of human volunteers were investigated using the nonintrusive technique presented. In the case of the maxillary sinus, two different detection geometries were used (PC and CC), and in the case of frontal sinus studies, a single detection geometry was used (OF) as follows.
(PC) Palate - cheek bone geometry: fiber placed inside the mouth against the palate close to the sinus under study and the detector positioned on and around the cheek bone.
(CC) Cheek - cheek bone geometry: fiber placed outside on the cheek below the cheek bone and the detector positioned on the cheek bone.
(OF) Orbital edge - frontal bone geometry: fiber placed on the horizontal part of the frontal bone under the eyebrow pointing up toward the frontal bone and the detector placed at different positions on the frontal bone.
Before any measurements were performed on any human sinuses, two types of reference measurements were carried out to investigate any possible offset in the current setup. The oxygen and water vapor signal were measured through the cheek below the cheekbone and through the hand of the volunteers where no free gas is expected. These measurements resulted in a consistently nondetectable gas imprint signal, showing that no gas offset is present due to gas between the setup components.
Molecular oxygen measurements were performed with geometry PC on 11 volunteers, 10 of which have no history of sinus problems (volunteers 1 to 10) and one with constantly recurring sinus problems (volunteer 11). The detector was placed at three different positions; on the cheek bone, slightly toward the nose, and slightly under the cheek bone. At each detector position, three consecutive measurements were performed by removing and repositioning the fiber and the detector between each recording. The reproducibility of the measured values was also investigated by measuring ten consecutive times at each detector position for one volunteer.
Measurements of oxygen and water vapor were performed with detection geometry PC on the maxillary sinuses on volunteers 1, 2, and 11, and on volunteer 1 also with geometry CC. In addition, frontal sinus measurements with geometry OF were performed on volunteer 1. In the case of geometry PC and CC, the detector was positioned on the cheek bone. In the case of frontal sinus study, detection geometry OF, the detector was positioned at three different locations on the fontal bone toward the particular sinus; about and about straight up above the light source, and about straight up over the center of the eyebrow. At each detector position, ten consecutive measurements were performed by removing the fiber and the detector between each recording. Molecular oxygen was measured first, followed by water vapor directly after using the same procedure.
To study the communication and ventilation between the nasal cavity and the human sinuses, we investigated the transient response to pure nitrogen flushing of the nasal cavity for about through the nostril. These measurements were performed on volunteer 1 with all detection geometries and volunteer 11 with detection geometry PC. Measurements were continuously recorded for about . before flushing, during flushing, and after terminating the flush, all the time pursuing normal mouth breathing. Each recorded signal was averaged for about . Measurements of molecular oxygen and water vapor were performed directly after each other. To investigate the stability of the recorded signals for the gas exchange study time period, the same procedure was also carried out without flush.
Monte Carlo simulations were performed by using the Advanced Systems Analysis Program (ASAP) software.16 In this work, we present a model according to Fig. 3a to simulate measurements on the maxillary sinuses employing detection geometry PC when detecting the multiply scattered light on and around the cheek bone. The position of the light source is indicated by a cross in the model. The maxillary sinus was represented by an ellipsoid with typical human dimensions. It was placed embedded in a scattering medium with optical properties set to (anisotropy factor), (scattering coefficient), and (absorption coefficient) typical of bone.17
The flux and the pattern through the model for rays reaching the detector were recorded. The distance each ray traveled in the ellipsoid was calculated. By summarizing the product of the distance in the cavity and the flux for each ray and normalizing it with the total flux reaching the detector, the value could be estimated. Similar simulations have been presented in Ref. 11 for frontal sinus imaging.
Results and Discussion
Maxillary Sinus Study
Figure 3b shows obtained through simulation measurements on the maxillary sinuses according to detection geometry PC. It can be seen that values between 10 and can be expected when performing measurements on volunteers with typical maxillary sinus dimensions when the detector is placed on and around the cheek bone. These values are heavily influenced by the optical properties used in the simulations, but can still be used as an indication of the experimental results to be expected. If, for example, a higher absorption coefficient is used, as was experimentally obtained by Hoshi, 18 a decrease of would be obtained.
The results from the molecular oxygen measurements on the volunteers according to geometry PC are presented in Table 1 . The value shown is an average from the three different detector positions. It was observed that the lowest is always recorded when the detector is positioned below the cheek bone, in agreement with the simulation results, where the detection areas are also indicated [Fig. 3b]. All volunteers with no history of sinus problems presented a detectable oxygen absorption imprint. However, no oxygen imprint could be recorded on the left side of volunteer 11. The reproducibility of the measured values was also investigated at each detector position for one volunteer. A standard deviation of about 10% was obtained when the average measured value was of the order of . A clear left-right asymmetry was observed for volunteer 11 as discussed before. Smaller asymmetries were observed also for other volunteers. However, it should be noted that the value depends on concentration as well as path length, and thus detailed analysis regarding anatomy are hard to draw.
The results from the measurements of oxygen and water vapor with detection geometry PC on the maxillary sinuses on volunteers 1 and 2 are shown in Figs. 4a and 4b . The average values are presented together with two standard deviations. It should be noted that the recorded for water vapor is about 8 larger than the value for oxygen for volunteers 1 and 2. The same types of measurements were attempted on volunteer 11. An oxygen signal of was detected for the right sinus, whereas no signal could be observed for the left side. No water vapor signal could be detected on the right side due to heavy overall absorption in the tissue and the partially filled sinus cavity.
As discussed previously, the estimated depends on both the concentration of the gas and the scattering properties of the tissue. It can be assumed that the scattering properties are approximately the same for the two different wavelengths used.19, 20 However, the absorption coefficient may be different mostly depending on the degree of oxygenation of the blood and liquid water absorption.17, 19 If the same volume has been sampled for the two wavelengths, the ratio between the values for oxygen and water vapor could provide information on the oxygen concentration inside the sinuses, since the water vapor concentration is always the maximum possible concentration at the given temperature (thermostated to about and 100% humidity in the cavities). In Fig. 4c, the ratios for the left and right maxillary sinus are shown. On volunteers 1 and 2, it can be seen that this ratio is about the same for both sides. It can also be seen that the two volunteers maintain similar ratios. The average ratio is . Assuming that these volunteers have healthy and well ventilated sinuses with about 20% oxygen concentration, a reduced ratio may indicate a lower oxygen concentration in the cavity.
Gas exchange studies were performed on volunteers 1 and 11 with the same detection geometry PC. A reduction of the oxygen absorption imprint could be observed on volunteer 1 when flushing with , as well as the reperfusion of air when terminating the flush [Fig. 5a ]. In Fig. 5b, the stability of the oxygen signal during the same time scale without flush can be seen. In the case of water vapor detection, no reduction of the signal could be recorded during flush [Fig. 5c]. A series of water vapor signal measurements is also shown in Fig. 5d, illustrating the stability of the data. The results indicate that the flushing is not able to change the humidity in the cavities. Figure 6 shows signals recorded at three different time intervals during the gas exchange study (indicated by corresponding numbers in Fig. 5). The fractional absorption was and for the largest signals observed for oxygen and water vapor, respectively. The signals from reference measurements through the cheek directly after the study are also shown. The same molecular oxygen measurements with nitrogen flushing were performed on the right maxillary sinus of volunteer 11. These measurements did not show any decrease in oxygen signal during the flush period, indicating lack of communication with the nasal cavity.
Oxygen and water vapor were measured according to geometry CC on the maxillary sinuses of volunteer 1 [see Figs. 7a and 7b ]. We note a left-right asymmetry stronger than observed for the same volunteer with data shown in Fig. 4 (left) when using a different detection geometry. This indicates that tomographic information might be extracted regarding anatomy using multiple geometries. The ratio between the signals is shown for both sides in Fig. 7c. An average ratio of is obtained for this geometry.
The same gas exchange procedure was carried out with detection geometry CC on volunteer 1 (see Fig. 8 ). The same behavior was observed as for detection geometry PC.
Frontal Sinus Study
Measurements of the frontal sinuses were performed on volunteer 1 with detection geometry OF. In Figs. 9a and 9b , the average together with two standard deviations is shown for the two gases, and the three different detector positions discussed above and seen in the top of Fig. 9. The ratio between the for oxygen and water vapor can be seen in Fig. 9c. An average of is obtained.
The gas exchange between the nasal cavity and the frontal sinuses was also studied with geometry OF. In Fig. 10 , the invasion of and reinvasion of ambient air can be seen for volunteer 1 when detecting oxygen. Again, no such phenomena could be observed when detecting water vapor.
Measured Leq values for oxygen in the maxillary sinus for volunteers 1 through 11 according to detection geometry PC. N represents volunteers with no history of sinus problems and Y the volunteer with constantly recurring sinus problems. The presented values are averaged from three different detector positions; on the cheek bone, slightly toward the nose, and slightly under the cheek bone.
Conclusion and Outlook
We show that nonintrusive diode laser spectroscopy with monitoring of common physiological gases such as oxygen and water vapor is capable to assess important aspects of human sinuses and their potential diseases. It is demonstrated that oxygen as well as water vapor can be detected within the maxillary and frontal sinuses on volunteers by employing different detection geometries discussed in Sec. 2. Individual variations and left-right asymmetries are observed. The oxygen absorption imprint levels detected are in good agreement with the results from Monte Carlo simulations implemented with software. By forming the ratio between the two signals (normalizing the oxygen signal on the water vapor signal), it is shown that basically the same value is obtained for all well ventilated sinuses studied. This is to be expected if a constant concentration of (about 20%) oxygen is assumed in the healthy sinuses, and if 100% water vapor saturation develops in the sinus. Further, a prerequisite would also be that the scattering and absorption coefficients are the same at 760 and , resulting in the same effective path length through the gas-filled cavity. Consulting the literature, it can be found that this assumption is not fully warranted.17, 19, 20 In particular, the absorption coefficient depends on the degree of blood oxygenation and amount of liquid water, and would tend to increase the absorption at , as compared to . The scattering coefficient could be expected to be the same within about 10%. The fact that a very similar signal ratio is obtained for different volunteers, sinuses and geometries, suggests that the assumption of a similar effective path length is reasonable. We notice, however, that the experimental ratio was found to be about 0.12. Going from 20% relative humidity to 100% influences the equivalent mean path length by a factor of 5. Likewise, going from 25 to introduces a factor of 2 in water potential pressure.15 Thus an experimental ratio of 0.10 would be expected. The measured value of 0.12 could thus be interpreted in two ways. The air is not attaining in dynamic exchange but only about , which is not unlikely. A further explanation is that the effective sample pathway for the two wavelengths differ by a factor of 1.2 with a smaller path length at due to increased absorption. A combination of the two effects could be present. Given that the experimentally observed effects are rather small, it seems like it should be possible to perform an empirical correction after having gained clinical data. The results of the present work would mean that it would be possible to nonintrusively read off the oxygen concentration in a sealed-off sinus cavity, which might be of clinical value if a correlation to pathology could be established in a clinical trial. It should be noted that a water vapor signal is always expected as soon as there is a gas-filled cavity. A zero reading for oxygen would then by necessity mean that no oxygen is present in the gas, not that the cavity contains no gas at all.
We also demonstrate that the ventilation of the sinuses (the ostial function) could easily be measured observing the oxygen signal, which is reduced when nitrogen gas is given through the nostril as a displacement gas while the patient is breathing normally through the mouth. By instead administering high-concentration oxygen gas, a much stronger signal can favorably be achieved. The time constants for the gas exchange can be determined by fitting the curves. The convenience of a combination of a physiological gas and light for gas exchange studies is in strong contrast to the previous use of stable or radioactive xenon gas in combination with repeated CT scans or SPECT imaging, both leaving the patient with a considerable dose of ionizing radiation. It was found that the concentration of water vapor in the sinuses was not influenced by the flushing with dry nitrogen gas. The conclusion is that 100% water saturated gas is immediately obtained in the sinus. This fact also corrobates the earlier finding that water vapor reliably can be used as a reference gas of known concentration in the well thermostated sinus (at 33 to ). By recording the relative intensity of two suitable oxygen or water vapor lines, and their half-width with sufficient precision, the temperature and pressure can in principle be measured.
Sequential measurements are made on oxygen and water vapor in the present studies. In a clinically adapted system, the two types of measurements could in principle be made simultaneously using the same detector and different modulation frequencies. When exactly the same geometry pertains for both kinds of measurements an even better stability for the ratio signal can be expected. A clinical trial on sinus patients jointly with an ear-nose-throat (ENT) clinic is planned to assess the clinical practicality and potential benefits of the new technique.
The demonstrated possibility to noninvasively assess the oxygen concentration in a hollow organ by sending light through a substantial layer of human tissue suggests that the new technique might have important applications in assessing the lung function in newborn and small children. With higher power lasers, still without any noticeable thermal effect, corresponding measurements on the lung lobes in adults should be feasible in lung ventilation studies, where, however, the empirical path length difference correction might be larger.
The authors are grateful for fruitful discussions with N. G. Holmer, M. Jannert, K. Jonsson, L. Malm, and K. Prellner. We are grateful to L. Simonsson and E. Kristensson for collaboration on simulations. This research was supported by the Swedish Research Council, the Medical Faculty at Lund University, and the Knut and Alice Wallenberg Foundation.