Open Access
Add to BibSonomy
Abstract
Vocal folds lesions are commonly diagnosed using an endoscopic-stroboscope. However, the stroboscopic picture of the vocal folds vibrations is subjectively and qualitatively evaluated by the clinician and, due to technical limitations, is unable to accurately distinguish between healthy and pathologic regions. In this paper, we propose two optical approaches for objectively sensing the vocal folds vibrations, using either external or internal laser illumination, based on temporal tracking of the reflected spatial distribution of secondary speckle patterns. The external configuration (the neck) is noninvasive and the internal configuration (the larynx) allows simultaneous extraction of data from multiple sites on the vocal folds. In this paper, we present measurements of healthy human subjects. Quantitative and precise measurements of vibration parameters of the vocal folds will enable a better understanding of hidden pathologies and optimize the diagnosis and treatment.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Benign and malignant lesions on the vocal folds interfere with the normal vibration of the vocal fold’s mucosa. This vibration is necessary for producing sound. In normal phonation, the vocal folds are adducted to the midline and the air pressure built underneath the closed folds creates a dynamic movement of the covering mucosa. This rapid movement, vibration of the vocal folds, extends until the air pressure diminishes and the vocal folds stop vibrating. Vocal folds vibrate at a frequency of 80 to 1,000 cycles per second [1]. As the human eye is capable of perceiving no more than twenty five images per second, it is impossible to visually evaluate the vibration of the vocal folds during phonation [2] and our retina will see the movement as stationary.
Many attempts have been made to identify and characterize the patterns of the vocal folds’ vibration and the effect that vocal folds lesions may produce on those patterns. There are several techniques for vocal folds imaging. A laser Doppler vibrometer can measure vibrations through interferometry, but this is an expensive and complex technique [3,4]. Another technique is video-kymography, a digital technique for high-speed visualization of vocal folds vibrations [5]. The most common technique, video-laryngostroboscopy, has proved to be an essential tool for vocal fold vibration analysis [6–8].
The stroboscope detects the fundamental frequency of the patient’s voice via a microphone and generates high speed flashes of light timed to the sensed frequency. This enables the visualization of virtual cycles of vibrations at lower frequencies during voice production. The strobe effect is feasible only when the vocal folds motion is periodic and if the voice production is long enough for the stroboscope to detect. In addition, as it represents a subsampling of several vibrational cycles, it is not possible to access the variations between and within cycles, neither recording the onset and offset of phonation. Furthermore, video-laryngostroboscopy examination is subjectively interpreted by the clinician.
Several basic parameters of the vibration patterns, such as amplitude, phase and symmetry between the two folds should be addressed but the overall interpretation is related to the experience of the observer [9]. It has been indicated in previous studies that the evaluation of the static structures on the vocal folds such as lesion size rating and the description of the closure of the vocal folds are reliable and consistent parameters. Yet the evaluation of the dynamic patterns, such as the vocal fold vibratory amplitude measurements are of low reliability [10].
Moreover, several vocal fold lesions such as scars, sulci, mucosal bridge and cysts can affect the voice quality dramatically but, at times, are difficult to detect visually. Yet it is impossible to determine, following video-laryngotroboscopic examinations, specific vibration patters at different areas along the vocal fold and thus to distinguish between healthy and pathogenic regions.
The introduction of the high-speed camera and video-kymography provided viewing of the real cycle of vocal folds vibration through high sampling rate of successive frames and adequate spatial resolution. Their major disadvantage is that the images are recorded during the examination and observed later, in playback, frame by frame. This time-consuming procedure is the main reason why high-speed camera and video-kymography are not routinely used in the clinical setting [11,12].
The ability to classify different vibration patterns on the same vocal fold at different spatial locations along the mucosa can be of great value in the detection of areas with disturbed vibration. This will help the clinician to focus on this area and determine the reason of the abnormal vibration in order to decide on the possible way of intervention; conservative or surgical.
In this paper, we propose a novel and applicable technique of optical non-contact measuring of vocal folds vibrations. The non-contact optical sensing method is based on temporal tracking of the spatial distribution of secondary speckle patterns and the extraction and analysis of their changes. The non-contact optical configuration was previously used for biomedical measurements, such as monitoring heart rate, blood pressure, blood oximetry, blood coagulation [13–16], bone fractures [17,18], melanoma [19] and more. This method is inexpensive and relatively simple optical technique.
The optical configuration includes projection of a laser beam and a fast imaging camera and enables us to analyze multiple regions of the vocal folds simultaneously. The imaging setup allows observation of the movement of the secondary speckle patterns that are created by the back scattered light from the surface roughness of the inspected object [20]. Each individual speckle pattern is a self-interfered pattern that serves as a reference point that tracks the changes in the light phases while being scattered from the object’s surface. When defocused, the constant vibrating image of speckles turns into an image in which one can see the same speckle pattern moving exclusively in the transversal plane proportionally to a tilt movement of the inspected surface. The tilting movement is expressed by the Fourier transform (far field approximation obtained when defocusing) as a lateral shift of the speckles pattern. It allows us to measure the object's displacement, using correlation to identify the vibration frequencies, by following in time the space-varying location of the correlation peak [21,22].
We present two non-contact optical sensing systems for measuring vocal folds vibrations, both based on secondary speckle patterns, either using an external configuration (of the neck) or an internal configuration (of the vocal folds). Each system has a laser illuminating several different regions of the vocal folds in healthy subjects.
All research procedures were performed with approval by the Helsinki committee of the Sheba Medical Center and an informed consent was obtained from all subjects who participated in the research experiment.
2. Methods
2.1 Experimental setup for external configuration
The system (Fig. 1) includes projection of a laser beam and a grating that splits and diffracts the light into several beams in different directions on the subject’s neck with respect to the vocal folds location. The images were obtained via a fast imaging camera that observes the temporal intensity changes in the position of the correlation peak. The correlation was performed on the spatial speckle patterns created by the vocal folds. The camera (Basler acA800-510um) captured the speckle images at 500 frames per second (fps) and was positioned approximately 90cm from the subject’s neck.
Fig. 1. The optical setup of the external configuration consisting from: an eye safe 532 nm laser and a camera.
2.2 Experimental setup for internal configuration
The system (Fig. 2) includes projection of a laser beam via an endoscope (therapeutic PENTAX FNL-15RP3) introduced to observe the subject’s vocal folds (a standard procedure in clinical practice). The endoscope has insertion tube with diameter of 4.9mm, instrument channel with diameter of 2.1mm and angle of view of $75^\circ $ (NA=0.6). The laser wavelength was 650nm, with a power of 5mW, collimated to an optical fiber. In the first part, a GRIN lens was attached to the end of the fiber in order to help focusing the laser beam, so that a single point of illumination is obtained on the vocal folds. The fiber was inserted through the endoscope's working channel and directed toward the vocal folds. The sensing system consisted of an imaging module having a lens with focal length of 16mm that was coupled to the endoscope output and then to the camera (Basler acA800-510uc). The camera delivers 511 frames per second, with a pixel size of 4.8µm. Our field of view (FOV) is 512 × 512 pixels, which means that under an endoscope magnification with factor of X10, the FOV is 24.5 × 24.5 mm. The size of the analyzed area is 1.2 × 1.2 mm. The temporal resolution depends on the equation:
where N is the number of frames recorded and the frame rate is a camera dependent parameter. The back reflected scattered light was collected by the endoscope to the imaging system, allowing get a picture of the location of the vocal folds. A lens with a focal length of 70mm, coupled to the endoscope output and then to a second camera (Basler acA800-510um) was used in order to observe the speckles. The lens generated a defocused image on the camera’s detection array, and the camera captures the speckle images at 500fps.Fig. 2. The optical setups of the internal configuration consisting from: a 650 nm fiber laser, an endoscope, cameras and lens for imaging and for capturing defocused speckle patterns.
In the second part, the sensing system remained the same, but the frame rate of the camera changed to 246fps and the fiber was without a GRIN lens. The fiber coupled laser creates several spatial modes enabling illumination of several different spatial regions simultaneously.
2.3 Analysis of temporal-spectral information extracted from speckle patterns
The detection of sound frequency is done by analyzing speckle patterns while the light back reflecting surface is defocused. Due to the vibration of the vocal folds the speckles of each frame were shifting, and the processing of the captured images was done by correlation between the frames (Fig. 3).
After performing the correlation on the time changes images of the speckles, the position of the correlation peak was analyzed (red arrow in Fig. 3). Its Fourier transform allows comprehensive analysis and provides medical insight. The extraction and separation of vibration temporal frequencies from several spatial regions, allows analyzing the sound frequency in each region simultaneously. It enables exploring vibration characteristics at different regions along the vocal folds [23,24].
Please note that as a reference the vibration frequency was also measured with a calibrated microphone simultaneously during the experiment.
3. Results and discussion
At first, the external configuration was used to examine healthy subjects. The laser illumination was pointed on various points of the neck area in coordination with the vocal folds’ location [Fig. 4(a)]. During the illumination, the vocal folds formed a unified sound. Analysis was addressed for each speckle pattern created from the illumination points.
Fig. 4. External configuration analysis for multi spot illumination. (a) Image of the different illumination points. (b) The vibration amplitude of measured illumination points from different locations on the neck. (c) At the location of 2 cm, as well as (d) at 6 cm, the frequency is correctly identified as 225 Hz, with a decrease in the amplitude at this frequency.
The amplitude of the sound frequency depends on the distance of the illumination point from the vocal folds [Fig. 4(b)]. In our experiment, the maximal amplitude was obtained at distance of 2cm, meaning that this was the location closest to the vocal folds (the amplitude decreases as the distance increases). However, the sound frequency both at 2cm [Fig. 4(c)] and farther away from the vocal folds [Fig. 4(d)], was correctly detected as 225Hz, because the sound spreads throughout the neck.
Although the external approach provides a general location of the vocal folds, it does not provide the exact locations. It is beneficial as a non-invasive approach that does not require a physician, which can still be used to identify the fundamental frequencies of the vocal folds. On the other hand, the internal system while being invasive, provides the exact location of the vocal folds.
The internal configuration was used for different healthy subjects undergoing an endoscopy. All the subjects were instructed to produce a single “eee” phonation, commonly used to distend the airway during endoscopic imaging. This phonation, when the larynx is higher, is making the vocal folds clearest and fully seen, which enabling us to measure in different locations across the entire fold [25].
Inserting the laser fiber through the endoscope into the throat provided a clear picture of the two vocal folds as well as the ability to distinguish between them.
In the first part, the illumination point was on the right fold [Fig. 5(a) a breathing mode with laser, and Fig. 5(b) a vibrating mode with laser]. At each spectral analysis, an area of 25 × 25 pixels was selected. Note, that the illumination spot size depends on the distance between the vocal folds and the endoscope and the NA of the fiber.
Fig. 5. Detection of vocal folds vibration using internal configuration. Images taken through endoscope (a) while breathing and (b) vibrating, with laser illumination. While breathing there is space between right (white) and left (green) folds. In vibrating mode three areas were tested; the left fold (green) and two on the right anterior fold (white), the right posterior fold (magenta).
The measurement analysis was performed based on speckle patterns analysis and the temporal spectrum of the specific region with laser illumination during both breathing and vibrating.
While breathing, the vocal folds where both open and not vibrating, so that they did not produce sound and allowed passage of air in and out the trachea. Thereby, in breathing mode we observed that the voice frequency was very negligible [Fig. 6(a)] in the illuminated area on the right fold (blue) and on the left fold (green); which is not illuminated.
Fig. 6. The correlation spectrum extracted from the analysis of the speckle patterns (in Fig. 5) with and without laser illumination, both while breathing and while vibrating. (a) While breathing the right fold (blue) and the left fold (green) do not present a peak. (b) While vibrating, the right anterior fold (blue) presents a peak in the spectrum around 196Hz, while the left fold (green) and the right posterior fold (magenta) do not present a peak. Inset: grayscale image of the analyzed state from Fig. 5.
While vibrating, we observed in the illuminated area on the right anterior fold (blue) that the voice frequency was 196 ± 1 Hz [Fig. 6(b)]. However, on the left fold (green) and on the right posterior fold (magenta), which was not illuminated areas, the voice frequency was very negligible, even though they vibrated, and the right posterior area was on the same fold as the illuminated area. Note, that without illumination the sound frequency of the vibration of the vocal folds cannot be measured, because no speckle pattern is created. Also, please note that although they are not illuminated by the red laser, the locations are illuminated by the light from the stroboscope which is how they were measured. The red laser is in a specific location where there is a speckle pattern.
In the second part, the use of several illumination points on different spatial regions of the vocal folds enabled focusing on different regions of the same vocal fold and comparing between the two vocal folds. The illumination was made at various points- one on the right fold (red), another on the anterior third of the right fold (blue), and two on the left posterior fold (magenta), the left anterior fold (green).
The measurement analysis was performed (according to section 2.3) simultaneously for all regions [the colored boxes in Fig. 7(a) and Fig. 8(a)]. Note that the different spatial regions do not overlap. The spectrum of the different regions was analyzed while breathing and without laser illumination and being illuminated only from the endoscope light [Fig. 7(a)]. No apparent peak exists [Fig. 7(b)], since the vocal folds do not vibrate while breathing.
Fig. 7. Detection of vocal folds vibration simultaneously for several areas with no laser while breathing. (a) Four areas are selected on the vocal folds; The right fold (red), another on the anterior third of the right fold (blue), and two on the left posterior fold (magenta), the left anterior fold (green). (b) The correlation spectrum extracted from the speckle’s analysis of these areas. No apparent peak exists.
Fig. 8. Detection of vocal folds vibration simultaneously for several areas, using laser illumination while vibrating. (a) Four areas are selected on the vocal folds; The right fold (red), another on the anterior third of the right fold (blue), and two on the left posterior fold (magenta), the left anterior fold (green). The correlation spectrum extracted from speckle‘s analysis of the four areas while vibrating; (b)The right fold, (c) on the anterior third of the right fold, (d) on the left posterior fold, and (e) the left anterior fold. While vibrating under laser illumination a peak appeared at frequency of around 114 Hz in all areas.
Next, the vocal folds were analyzed while vibrating with laser illumination [Fig. 8(a)]. While vibrating with illumination [Figs. 8(b)–8(e)], we observed that the voice frequency in all the regions was 114 ± 1 Hz. This was expected since the patient did not present any pathologies. As the point of illumination became weak, the amplitude of the frequency became smaller.
Please note that indeed we use a defocused lens to analyze the speckle pattern. This defocusing causes to loss spatial resolution or allocation (between the measured vibration signal and the spatial domain). However, the defocusing is not very strong and thus we can indeed have proper spatial allocation of the measured signal with sufficient spatial resolution. The different spatial regions, where the measurement was performed, do not overlap. As shown in Fig. 8(a), the different regions are the colored boxes, which show where the correlation spectrum of the speckle patterns were analyzed.
Another comment is related to the fact that in this article we use the symmetry to distinguish between the right and left vocal folds. We do not need to distinguish between the phases because we are differentiating between vocal folds and different locations on vocal folds which have pathologies and those that do not. We use the arbitrary units because we are comparing only between the left and right vocal folds. The spectra we get in Fig. 8 is sufficient, because we show that we can analyze frequencies in different regions. A healthy subject will have the same frequencies in all regions as we can see in Fig. 8, but on the other hand a patient with pathologies is expected to have different frequencies at different regions.
4. Conclusions
We proposed an innovative technique that will enable the characterization of differences between vibrations of consecutive segments of the vocal folds in different pathologies and to typify various pathologies by these features. The measurements were made using both an external and internal measurement configuration. The external configuration, while nonintrusive, does not provide an accurate view of the location of the vocal folds (but it is a non-invasive approach that does not require a physician and which can still be used to identify the fundamental frequencies of the vocal folds). Thus, the sound frequency can be detected but it is not possible to separate and identify different regions on the vocal folds. The internal configuration requires the intervention of an ENT (ear, nose and throat) physician for the endoscopic examination.
The significant advantage of this approach is the ability toenable real-time analysis of temporal-frequencies of different spatial regions of the vocal folds simultaneously. This contribution allows the extraction of additional parameters of vocal folds physiology and pathology for better evaluation of the voice pathology.
Identification of disturbed vibration characteristic parameters in voice pathologies will enable a thorough understanding of hidden pathologies in stroboscopic examination. Focusing on the area of the disorder will allow the physician to make a more accurate decision about the cause of the problem as well as to select the preferred treatment.
In the future, we aim to examine (based on the internal configuration), characterize, and profile various pathologies of the vocal folds by their temporal vibration characteristics.
References
1. P. Woo, Stroboscopy (Plural Publishing Group Inc., 2010, pp. 11–17).
2. S. Hecht and C. D. Verrijp, “Intermittent stimulation by light: iv.A theoretical interpretation of the quantitative data of flicker,” J. Gen. Physiol. 17(2), 269–282 (1933). [CrossRef]
3. A. B. Stanbridge and D. J. Ewins, “Modal testing using a scanning laser Doppler vibrometer,” Mech. Syst. Signal Pr. 13(2), 255–270 (1999). [CrossRef]
4. R. L. Goode, G. Ball, S. Nishihara, and K. Nakamura, “Laser Doppler vibrometer (LDV)- a new clinical tool for the otologist,” Am. J. Otol. 17(6), 813–822 (1996).
5. J. G. Švec and H. K. Schutte, “Videokymography: high-speed line scanning of vocal fold vibration,” J. Voice 10(2), 201–205 (1996). [CrossRef]
6. A. Ricci-Maccarini, G. Bergamini, and R. Fustos, “Proposal of a form for the collection of videolaryngostroboscopy basic findings,” Eur. Arch. Otorhinolaryngol. 275(7), 1927–1933 (2018). [CrossRef]
7. P. H. Dejonckere, P. Bradley, P. Clemente, G. Cornut, L. Crevier-Buchman, G. Friedrich, P. Van De Heyning, M. Remacle, V. Woisard, and Committee on Phoniatrics of the European Laryngological Society (ELS), “A basic protocol for functional assessment of voice pathology, especially for investigating the efficacy of (phonosurgical) treatments and evaluating new assessment techniques. Guideline elaborated by the Committee on Phoniatrics of the European Laryngological Society (ELS),” Eur. Arch. Otorhinolaryngol. 258(2), 77–82 (2001). [CrossRef]
8. G. Friedrich and P. H. Dejonckere, “The voice evaluation protocol of the European Laryngological Society (ELS) - first results of a multicenter study,” Laryngo Rhino Otol. 84(10), 744–752 (2005). [CrossRef]
9. J. W. Brunings, S. Vanbelle, A. Akkermans, N. M. M. Heemskerk, B. Kremer, R. J. Stokroos, and L. W. J. Baijens, “Observer Agreement for Measurements in Videolaryngostroboscopy,” J. Voice 32(6), 756–762 (2018). [CrossRef]
10. E. M. L. Yiu, V. Lau, E. P.-M. Ma, K. M. K. Chan, and E. A. M. Barrett, “Reliability of laryngostroboscopic evaluation on lesion size and glottal configuration: a revisit,” The Laryngoscope 124(7), 1638–1644 (2014). [CrossRef]
11. J. G. Švec, F. Šram, and H. K. Schutte, “Videokymography in voice disorders: what to look for?” Ann. Otol., Rhinol., Laryngol. 116(3), 172–180 (2007). [CrossRef]
12. Q. Qiu, H. K. Schutte, L. Gu, and Q. Yu, “An automatic method to quantify the vibration properties of human vocal folds via videokymography,” Folia Phoniatr. Logo. 55(3), 128–136 (2003). [CrossRef]
13. Y. Beiderman, I. Horovitz, N. Burshtein, M. Teicher, J. Garcia, V. Micó, 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]
14. M. Golberg, J. Ruiz-Rivas, S. Polani, Y. Beiderman, and Z. Zalevsky, “Large-scale clinical validation of noncontact and continuous extraction of blood pressure via multipoint defocused photonic imaging,” Appl. Opt. 57(7), B45–B51 (2018). [CrossRef]
15. 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]
16. 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]
17. Y. Bishitz, N. Ozana, Y. Beiderman, F. Tenner, M. Schmidt, V. Mico, J. Garcia, and Z. Zalevsky, “Noncontact optical sensor for bone fracture diagnostics,” Biomed. Opt. Express 6(3), 651–657 (2015). [CrossRef]
18. N. Ozana, Y. Bishitz, Y. Beiderman, J. Garcia, Z. Zalevsky, and A. Schwarz, “Remote optical configuration of pigmented lesion detection and diagnosis of bone fractures,” in Photonic Therapeutics and Diagnostics XII (International Society for Optics and Photonics, 2016), 9689, p. 968916.
19. Y. Bishitz, N. Ozana, A. Schwarz, Y. Beiderman, J. Garcia, and Z. Zalevsky, “Optical configuration of pigmented lesion detection by frequency analysis of skin speckle patterns,” Biomed. Opt. Express 7(3), 1003–1014 (2016). [CrossRef]
20. J. W. Goodman, Speckle Phenomena in Optics: Theory and Applications (Roberts & Co, 2007, pp. 413–414).
21. J. W. Goodman, Laser Speckle and Related Phenomena (Springer, 1975, pp. 9–75).
22. J. W. Goodman, “Some fundamental properties of speckle,” J. Opt. Soc. Am. 66(11), 1145–1150 (1976). [CrossRef]
23. Z. Zalevsky and J. Garcia, “Motion detection system and method,” United States patent US8638991B2 (2014).
24. J. A. Leendertz, “Interferometric displacement measurement on scattering surfaces utilizing speckle effect,” J. Phys. E: Sci. Instrum. 3(3), 214 (1970). [CrossRef]
25. S. E. Rubesin, B. Jones, and M. W. Donner, “Contrast pharyngography: the importance of phonation,” AJR, Am. J. Roentgenol. 148(2), 269–272 (1987). [CrossRef]
