Otitis Media (OM) is related to a group of inflammatory diseases of the middle ear (ME) commonly encountered, worldwide. A method based on a simple device, which can be used by medical staff and non-experts to detect OM is presented. The method is based on detection of tympanic membrane (TM) vibrations. A laser beam is pointed on an infra-sonic stimulated TM with fast camera capturing the back scattered secondary speckle patterns. A camera enables inspection of the frequency and amplitude of the changes in TM characteristics obtained by analysis of the spatial-temporal statistics of the speckle patterns. The results may provide information that express ME effusion.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
The tympanic membrane (TM) is a semi-translucent structure dividing between two aerated spaces: the external ear and the middle ear. It transfers sound waves into mechanical vibration from the ear canal into the middle ear. Therefore, any changes in middle ear aeration due to infectious or inflammatory disease, impair sound transmission and causes hearing loss .
Middle ear infections are commonly encountered in primary care medicine, worldwide. It is presented by either acute or chronic forms that may involve all ages. The prevalence in China, for example, was 14.0% in 2-year-olds, tapering down to nearly 2% in older children [2–11].
The hallmark of middle ear infection is the presentation of fluids, instead of air, in the middle ear cavity. No less important, however, is the differentiation between acute and non-acute infections expressed by different types of effusion in terms of color and viscosity . Hence, it is reasonable to assume differences of TM impedance dependent upon the physical characteristics of middle ear effusion,
From an acoustic point of view, MEE decreases TM mobility and impedes the transfer of energy via the middle ear, resulting in hearing losses, commonly between 15 and 40 dB HL .
Physical inspection of the TM (otoscopy) is the standard evaluation of diseases of the middle ear. Changes in the normal appearance of the TM (i.e. discoloration, position or translucency) reflect pathological conditions. A more 'sophisticated' otoscopic examination, namely pneumatic otoscopy, is aimed to observe TM mobility. It is based on an otoscope that includes a rubber bag attached to the ear speculum that is fitted snugly to the external ear canal. Inflation and deflation of the rubber bag changes the air pressure in the ear canal and result in TM outward and inward mobility.
The “gold” standard of verifying middle ear effusion is myringotomy (a small incision of the membrane enabling suctioning of the contained fluids). However, in daily clinical practice diagnosis is established on non-invasive techniques, mainly otoscopy.
Current clinical guidelines for treating MEE are based primarily on pneumatic otoscopy. Separate studies report sensitivity values from 85% to 91% and specificity from 58% to 89%. Still, interpretation of test results was characterized by highly variable outcomes either across otoscopists, or across repeated tests by a single otoscopist [14, 15].
Acoustic reflectometry  and spectral gradient  belongs to an acoustic a technique which measures the sound amplitude transmitted and reflected from the middle ear to a microphone located in a probe tip placed against the ear canal and directed toward the tympanic membrane. It involves sending a harmless, inaudible sonar-like sound wave from the emitter that goes through the tympanic membrane, hits the posterior wall of the middle ear space, and bounces back to the sound detector in the device. These methods did not reach clinical popularity and use. Therefore, a need for improving diagnosis of middle ear infections still exists.
Tympanometry is an objective measurement of the acoustic immittance that depends on the pressure of the middle ear and the compliance of the tympanic membrane. The guidelines of the American Academy of Otolaryngology—Head and Neck Surgery Foundation strongly recommend that clinicians should obtain tympanometry only in children with suspected SOM for whom the diagnosis is uncertain . The diagnosis of serous otitis media in adult patients (as in our study) is performed using otoscopy, taking advantage of the transparency of the tympanic membrane that enables us to see whether the middle ear cavity is aeriated or with fluid and even assess the type of fluid (serous, mucoid or purulent). Weber examination, using a tuning fork, is another bed-side examination that helps the otolaryngologist to establish the diagnosis and support the findings of the otoscopy. All our patients underwent otoscopy, performed by a senior otolaryngologist, and the diagnosis of fluid in the middle ear cavity was supported by either tympanometry or Weber exam.
In this paper we propose a novel photonic technique for early diagnosis of OM. It is meant to be simple and inexpensive and to be used by medical staff and non-experts for the detection of early stages of otitis media. The described configuration includes observation of the secondary speckle patterns that are created by illuminating the ear drum with a class 1 (i.e. eye and tissue safe) laser beam. Nano vibrations of the ear drum cause the self-interference random patterns (i.e. speckle patterns)  to change as the interference affects the light waves. By using this approach, the ear drum’s temporal movement due to external remote sound waves can be tracked [20–22]. The tissue vibrations characteristics will be affected due to different ear drum elasticity conditions such as middle ear effusion. These nano-vibrations’ changes can be detected using the presented method [23–28], which is based on a very simple and inexpensive device.
There are several advances of the presented method with respect to reflectometry. (1) The presented method is demonstrated by illuminating with a laser beam to sense nanometric vibrations. The vibrations sensing is directional, i.e. the vibrations are extracted only from the illuminated area. Due to this fact, vibrations from only a single point (i.e. the size of the laser beam) of the ear drum are extracted. However, reflectometry is demonstrated using a microphone which collect the surrounding sound from all the directions, hence, in reflectometry, the noise is higher. (2) To sense the vibrations using the presented method, the camera was strongly defocused close to the far field region (Fig. 1 of the optical configuration which demonstrates the optical design of the system). At the far field, tilting movement of the ear drum is translated to movement at X and Y plane. Therefore, tracking the peak location (rather than the peak value) of simple correlation between the time varied speckle images will extract the vibrations. To sense this movement with sub pixel resolution, sub pixel approximation was calculated. Using this calculation, a movement of 50nm of the speckle field is extracted (1/100 the size of the camera pixel which is 5µm). Therefore, the presented method is much sensitive than the reflectometry method. (3) The configuration consists of a camera and a laser diode which will be added to the otoscope device. A chip that contains these components with a size of 5x5mm has been developed. The market price of this chip will be less than $10. This low price will allow integration of this chip for clinical use.
2. Theoretical explanation
A prototype device was constructed to detect middle ear effusion based on a simple otoscope with a CCD camera, a green laser (at 532 nm) and an ear lamp as shown in Fig. 1. To vibrate the ear drum, sound waves were generated by an external loud speaker. Via the analysis of the generated secondary speckle patterns we extract the tilting movement of the ear drum. Mathematically the light distribution can be expressed as follows :
where the far field is located as follows:
Recently the ear drum width roughness variability was measured using optical coherence tomography method . The results show significant variation of the roughness (>20µm) with respect to the illuminated wavelength in our case (i.e. 532nm). Therefore, a fully developed speckle pattern is generated due to the eardrum roughness.
Please note that the setup consists of two optical configurations. One is for imaging to visualization of the ear drum and the second was defocused to sense the ear drum vibration with high sensitivity. The camera frame for visualization should be more than twice of the acoustic stimulation frequencies. Therefore, the small ROI of the camera 128x128 pixels was selected. Furthermore, maximum power in the eye safe region was selected to minimize the exposure time of the camera. Using a simple correlation-based algorithm, the 2-D ear drum movement can be extracted. Temporal movement of the reflecting surface causes changes in the random speckle pattern over time due to the temporal change in its tilting angle. In the first step, a set of images as a function of time was captured. In the second step, the sequential 2-D row data is correlated. The relative movement of patterns can be extracted using a 2-D correlation. The position of the correlation peak over time indicates this relative tilting movement. The temporal movement of the eardrum is caused by external acoustic waves generated by the remote loud speaker. Summary of the presented process is presented in Fig. 2.
Please note that the presented configuration consists of a camera and illimitation in angle with respect to the eardrum. Due to this fact, the eardrum’s motion  will cause a change of the peak correlation location as explained in Ref [30, 31]. The steps of the signal processing are as follows: (1) extracting the time varied speckle patterns. (2) Calculating the correlation image between two sequential images. (3) Finding the peak location of the correlation image. (4) Calculating sub pixel resolution of the peak location using interpolation. (5) Calculating the DTFT of the time varied peak location.
To detect ME effusion, we calculated the frequency response of the ear drum at the excitation vibration frequencies (main peak) when vibrated due to the acoustic excitation. The remote acoustic generator was located at 1 meter from the subject. The amplitude was 60db. The frequency response is expressed as a discrete Fourier transform (DFT). In this section 5 different normal ear drums created unique optical signatures that are significantly different than 5 optical signatures of EM effusion. Applying a series of different vibration frequencies at the examined eardrum and analyzing the 2-D time varying speckle patterns in response to the applied periodic pressure creates a different frequency response for otitis in contrast to a healthy eardrum. Different normal ear drums created unique optical signatures that are significantly different than the optical signatures of EM effusion [24, 27]. Analyzing these spectral responses is the first step toward a simple detection of otitis. The presented device was built as a simple otoscope with a laser and a camera as it shown in Fig. 1. The eardrum backscattered light was split by a beam splitter into speckle sensing and visual sensing (to locate the beam at the ear drum) simultaneously. Three difference infra-sonic frequencies were applied by the remote acoustic stimulator: 290Hz, later 390Hz and finally 490Hz. Each time, the power spectral density of the temporal movement was calculated. Figure 3 shows an example of ear drum vibrations due to the acoustic waves stimulation.
The speckle temporal movement was captured by the camera, later this movement was analyzed using MATLAB. To detect middle ear effusion, the ear drum’s optical frequency response was compared with the results of a healthy ear. One can see in Fig. 4 an example of a healthy ear’s frequency response with respect to an OM frequency response. During this test, to show the presented method with different set of frequencies as well, the acoustic excitation frequencies were 155Hz, 255Hz and 355Hz. As one can see, there are no peaks at the excitation frequencies of an examined OM. However, when a healthy ear was examined, one can clearly see the peaks at the excitation frequency.
3. In vitro test
3.1. External stimulation results
To examine the ability to detect middle ear effusion, first an in vitro test was performed. During this test, the frequency response of a synthetic skin membrane (SSM) was evaluated due to remote acoustic wave stimulation. To compare between the in vitro and in vivo tests, the same acoustic stimulation as the one in the in vitro test was used. The in vitro configuration is presented in Fig. 5. To avoid mechanical vibrations, the remote acoustic stimulation device was located on external optical board which was separated from the configuration table.
During the test, acoustic stimulation was generated at 440Hz and the measurement was repeated 10 times. Figure 6 compares the frequency response of the SSM with only air surrounding the membranes, in contrast to SSM with fluid under the SSM.
3.2. Decay time results
To distinguish between different types of effusion, additional in vitro experiments were conducted. During this experiment, 5 different samples with different agarose concentrations were evaluated using the same configuration of Fig. 5. The agarose role is to solidify the samples to show the ability to separate between different effusions according to the sample adhesion. 2% - 6% agarose concentrations were prepared. While 2% adhesion present fluid sample and 6% present solid sample. Different concentrations were used to assess, in vitro, the feasibility of our device to detect different middle ear effusions (i.e. serous, mucoid and purulent otitis media). The next step of our research will be assessing these findings in-vivo. To measure the sample adhesion, the sample was illuminated with a 532nm laser, without external sound waves, and a camera captured the images at 800 fps. Since fluid samples will generate faster random movement by the particles scattering, the decay time of the correlation value between a reference image to a set of the next images will be faster than a solid sample. In this way due to the decay time the emission type can be evaluated. One can see in Fig. 7 that samples with higher adhesion (i.e. higher concentration of agarose) will decay to a lower correlation.
To evaluate the decay rate, the correlation value with respect to the first image as a function of time was fitted to an exponential function (R – square > 0.988 for all the concentrations). The fit functions are shown in Fig. 8.
One can see in Fig. 9. that the decay rate is higher while high adhesion (high agarose concentration) is under investigation. For higher agarose gel concentrations, the decay time is lower due to the fact there are fewer scattering particles in the sample, hence, the time for correlation loss is higher. This method together with the movement of the ear drum due to external acoustic waves can be an indicator for ME effusion type.
4. In vivo test
4.1. External stimulation results
Five adult patients with middle ear effusion were examined as it is shown in Table 1.
The study was reviewed and approved by the Research Ethics Committee of the Chaim Sheba Medical Center, Tel-Hashomer, Israel (application 1624-14-SMC). During this test, three optical signatures of the response of the ear drum were recorded for three different remote acoustic stimulation frequencies. First, the time domain of each excitation frequency was extracted. Later, the frequency response of each excitation was calculated and plotted as it shown in Fig. 10 where one can see healthy ear drum frequency response (a-c) and frequency response of middle ear effusion at the same excitations frequency (d-e). These excitation frequencies generated a unique signature. The camera sample rate was 1100 Hz to sample the acoustic vibrations above Nyquist criterion.
As it will be shown, using this signature, a middle ear effusion can be distinguished from a healthy ear drum. Figure 10 and 11 represent measurements taken from Patient No. 2.
Figure 12 shows the difference between the optical signature of the healthy ear drum and the infected ear drum in the other ear of five different individuals. All the patients were diagnosed by a physician at the hospital. During our next step, samples of the effusion will be taken from large population before ear drum surgery. The presented methods will be evaluated with respect to the effusion sample.
One can see in Fig. 13, the different between the normalized energy of the middle ear effusion and the healthy ear drum. This parameter can be an indicator for OM monitoring. The normalized energy was calculated by first summing the frequency response at the excitation frequencies (i.e. 155 Hz, 255 Hz and 355 Hz) and later dividing these results by the sum of the frequency response vector (i.e. sum of all the frequencies).
4.2. Decay time results
During this test, the decay correlation time was calculated as it was explained in section 3.2. One can see in the following Fig. 14 an example of a healthy ear drum and ME decay time graphs.
During this test, again, 5 different subjects were examined. The decay time of ME with respect to healthy ear is shown in the following Fig. 15. As one can see, ME effects the decay time with respect to healthy ear drum. The next step will be to examine large population with different infections using the vibrations sensing and the decorrelation time.
5. Discussion and Conclusions
In this paper, the usage of optical remote configuration for detection of middle ear effusion was presented. The proposed novel approach was experimentally demonstrated in vitro as well as in vivo with 5 different subjects (healthy vs. having middle ear infusion). The proposed configuration consists of only a camera, an eye safe laser (class 1) and a remote acoustic excitation source which stimulates vibrations in the ear drum. The future device will be independent of family doctor visual examination. Furthermore, future developments will include a simple device which will be suitable for personal home use. The presented approach will be demonstrated using a real time algorithm as well. Please note that the algorithm calculates the correlation between two sequential images. The time to calculate the correlation frame to frame is less than 1ms, hence, this approach satisfies the real time criterion. Regarding the decay rate constant method, each sequence requires 200 frames to calculate the decay constant. In this case, the correlation time is less than 0.2 s. This process could be applied in real time systems as well. Furthermore, in future, fast algorithms for finding the peak location between two sequential images and for decay constant calculation will be developed.
References and links
1. M. Tos, “Epidemiology and natural history of secretory otitis,” Am. J. Otol. 5(6), 459–462 (1984). [PubMed]
2. Q. Zhang, J. Wei, and M. Xu “Prevalence of otitis media with effusion among children in Xi'an, China: a randomized survey in China's mainland,”, 120, 617–621 (2011).
4. M. Iversen, L. Birch, G. R. Lundqvist, and O. Elbrønd, “Middle ear effusion in children and the indoor environment: an epidemiological study,” Arch. Environ. Health 40(2), 74–79 (1985). [CrossRef] [PubMed]
5. M. L. Casselbrant, L. M. Brostoff, E. I. Cantekin, M. R. Flaherty, W. J. Doyle, C. D. Bluestone, and T. J. Fria, “Otitis media with effusion in preschool children,” Laryngoscope 95(4), 428–436 (1985). [CrossRef] [PubMed]
6. J. Lous and M. Fiellau-Nikolajsen, “Epidemiology and middle ear effusion and tubal dysfunction. a one-year prospective study comprising monthly tympanometry in 387 non-selected 7-year-old children,” Int. J. Pediatr. Otorhinolaryngol. 3(4), 303–317 (1981). [CrossRef] [PubMed]
7. M. L. Casselbrant, L. M. Brostoff, E. I. Cantekin, M. R. Flaherty, W. J. Doyle, C. D. Bluestone, and T. J. Fria, “Otitis media with effusion in preschool children,” Laryngoscope 95(4), 428–436 (1985). [CrossRef] [PubMed]
8. D. W. Teele, J. O. Klein, and B. A. Rosner, “Otitis media with effusion during the first three years of life and development of speech and language,” Pediatrics 74(2), 282–287 (1984). [PubMed]
10. T. Muderris, A. Yazıcı, S. Bercin, G. Yalçıner, E. Sevil, and M. Kırıs, “Consumer acoustic reflectometry: accuracy in diagnosis of otitis media with effusion in children,” Int. J. Pediatr. Otorhinolaryngol. 77(10), 1771–1774 (2013). [CrossRef] [PubMed]
11. J. E. MacClements, M. Parchman, and C. Passmore, “Otitis media in children: use of diagnostic tools by family practice residents,” Fam. Med. 34(8), 598–603 (2002). [PubMed]
12. G. L. Monroy, P. Pande, R. L. Shelton, R. M. Nolan, D. R. Spillman Jr, R. G. Porter, M. A. Novak, and S. A. Boppart, “Non-invasive optical assessment of viscosity of middle ear effusions in otitis media,” J. Biophotonics 10(3), 394–403 (2017). [CrossRef] [PubMed]
13. I. Williamson, “Otitis media with effusion,” Clin. Evid. 7(7), 469–476 (2002). [PubMed]
14. M. E. Pichichero, “Acute otitis media: Part I. Improving diagnostic accuracy,” Am. Fam. Physician 61(7), 2051–2056 (2000). [PubMed]
15. G. S. Takata, L. S. Chan, T. Morphew, R. Mangione-Smith, S. C. Morton, and P. Shekelle, “Evidence Assessment of the Accuracy of Methods of Diagnosing Middle Ear Effusion in Children With Otitis Media With Effusion,” Pediatrics 112(6), 1379–1387 (2003). [CrossRef] [PubMed]
16. H. Lindén, H. Teppo, and M. Revonta, “Spectral gradient acoustic reflectometry in the diagnosis of middle-ear fluid in children,” Eur. Arch. Otorhinolaryngol. 264(5), 477–481 (2007). [CrossRef] [PubMed]
17. M. K. Laine, P. A. Tähtinen, K. K. Helenius, R. Luoto, and A. Ruohola, “Acoustic Reflectometry in Discrimination of Otoscopic Diagnoses in Young Ambulatory Children,” Pediatr. Infect. Dis. J. 31(10), 1007–1011 (2012). [PubMed]
18. R. M. Rosenfeld, J. J. Shin, S. R. Schwartz, R. Coggins, L. Gagnon, J. M. Hackell, D. Hoelting, L. L. Hunter, A. W. Kummer, S. C. Payne, D. S. Poe, M. Veling, P. M. Vila, S. A. Walsh, and M. D. Corrigan, “Clinical Practice Guideline: Otitis Media with Effusion Executive Summary (Update),” Otolaryngol. Head Neck Surg. 154(2), 201–214 (2016). [CrossRef] [PubMed]
19. J. W. Goodman, “Some fundamental properties of speckle,” J. Opt. Soc. Am. 66(11), 1145–1150 (1976). [CrossRef]
20. Z. Zalevsky, Y. Beiderman, I. Margalit, S. Gingold, M. Teicher, V. Mico, and J. Garcia, “Simultaneous remote extraction of multiple speech sources and heart beats from secondary speckles pattern,” Opt. Express 17(24), 21566–21580 (2009). [CrossRef] [PubMed]
21. N. Ozana, I. Margalith, Y. Beiderman, M. Kunin, G. A. Campino, R. Gerasi, J. Garcia, V. Mico, and Z. Zalevsky, “Demonstration of a Remote Optical Measurement Configuration That Correlates with Breathing, Heart Rate, Pulse Pressure, Blood Coagulation, and Blood Oxygenation,” Proc. IEEE 103(2), 248–262 (2015). [CrossRef]
22. N. Ozana, N. Arbel, Y. Beiderman, V. Mico, M. Sanz, J. Garcia, A. Anand, B. Javidi, Y. Epstein, and Z. Zalevsky, “Improved noncontact optical sensor for detection of glucose concentration and indication of dehydration level,” Biomed. Opt. Express 5(6), 1926–1940 (2014). [CrossRef] [PubMed]
23. Y. Beiderman, I. Horovitz, N. Burshtein, M. Teicher, J. Garcia, V. Mico, and Z. Zalevsky, “Remote estimation of blood pulse pressure via temporal tracking of reflected secondary speckles pattern,” J. Biomed. Opt. 15(6), 061707 (2010). [CrossRef] [PubMed]
24. A. Shenhav, Z. Brodie, Y. Beiderman, J. Garcia, V. Mico, and Z. Zalevsky, “Optical sensor for remote estimation of alcohol concentration in blood stream,” Opt. Commun. 289, 149–157 (2013). [CrossRef]
25. Z. Zalevsky, I. Margalit, Y. Beiderman, A. Skaat, M. Belkin, R.-P. Tornow, V. Mico, and J. Garcia, “Remote and Continuous Monitoring of Intraocular Pressure Using Novel Photonic Principle,” Invest. Ophthalmol. Vis. Sci. 53, 1972 (2012).
26. N. Ozana, S. Buchsbaum, Y. Bishitz, Y. Beiderman, Z. Schmilovitch, A. Schwarz, A. Shemer, J. Keshet, and Z. Zalevsky, “Optical remote sensor for peanut kernel abortion classification,” Appl. Opt. 55(15), 4005–4010 (2016). [CrossRef] [PubMed]
27. N. Ozana, Y. Beiderman, A. Anand, B. Javidi, S. Polani, A. Schwarz, A. Shemer, J. Garcia, and Z. Zalevsky, “Noncontact speckle-based optical sensor for detection of glucose concentration using magneto-optic effect,” J. Biomed. Opt. 21(6), 65001 (2016). [CrossRef] [PubMed]
28. N. Ozana, Y. Bishitz, Y. Beiderman, J. Garcia, and Z. Zalevsky, “Remote optical configuration of pigmented lesion detection and diagnosis of bone fractures,” Proc. SPIE 9689, 968916 (2016).
29. S. Van der Jeught, J. J. J. Dirckx, J. R. M. Aerts, A. Bradu, A. G. Podoleanu, and J. A. N. Buytaert, “Full-Field Thickness Distribution of Human Tympanic Membrane Obtained with Optical Coherence Tomography,” J. Assoc. Res. Otolaryngol. 14(4), 483–494 (2013). [CrossRef] [PubMed]
30. P. Razavi, M. E. Ravicz, I. Dobrev, J. T. Cheng, C. Furlong, and J. J. Rosowski, “Response of the human tympanic membrane to transient acoustic and mechanical stimuli: Preliminary results,” Hear. Res. 340, 15–24 (2016). [CrossRef] [PubMed]
31. R. Z. Gan, C. Dai, and M. W. Wood, “Laser interferometry measurements of middle ear fluid and pressure effects on sound transmission,” J. Acoust. Soc. Am. 120(6), 3799–3810 (2006). [CrossRef] [PubMed]