Professional Documents
Culture Documents
Detection of Muscle Tension Dysphonia Using Eulerian Video Magnification
Detection of Muscle Tension Dysphonia Using Eulerian Video Magnification
Summary: Objective. To determine whether Eulerian Video Magnification software is useful in diagnosis of
muscle tension dysphonia (MTD).
Study Design. Prospective.
Methods. Adult patients scheduled in a tertiary care laryngology practice for evaluation of dysphonia were
recruited between November 2016 and March 2017. Demographic and clinical data were extracted from patient
charts. Diagnosis of MTD was confirmed with videostroboscopic and physical exam and by a speech-language
pathologist. Eighteen MTD patients were video recorded while at rest and with phonation. Five patients without
MTD also were analyzed as controls. Videos were analyzed using Eulerian Video Magnification software (Mas-
sachusetts Institute of Technology) to assess change in blood flow at the forehead, infrahyoid muscles, and ster-
nocleidomastoid muscles, while using the values of the background wall as a control value.
Results. Patients with MTD demonstrated little change in perfusion to the infrahyoid muscles of the neck while
phonating (+1% § 55%). Control subjects demonstrated an increase in perfusion to the infrahyoid muscles while
phonating (+102% § 164%), with this change being significant when comparing the two groups (P = 0.04,
t = 2.189, df = 21). A change in perfusion of 0% or less to infrahyoid muscles was 75% sensitive and 70% specific
for diagnosis of MTD. No differences in perfusion were found between other regions assessed. Patient age and
gender did not correlate with any change in perfusion between rest and phonation.
Conclusion. Our data suggest that Eulerian Video Magnification can be used in the diagnosis of MTD by
focusing on the difference in perfusion to the infrahyoid muscles between rest and phonation.
change in pixel redness over a given amount of time. This the statistical analysis. A total of 18 patients were thus
slight change in pixel redness, which may not be visible to included. Additionally, five patients with no history of or
the human eye, signifies a change in blood perfusion. This exam findings consistent with a voice disorder, such as
subtle change in blood perfusion can then extrapolated to MTD, were included in the control population. Demo-
measure muscle activation.26−29 graphic and clinical data were extracted from the charts on
In medicine, EVM has been used to monitor vital signs,30−32 each patient and control subject. Patients included in the
fasciculations in patients with Amyotrophic Lateral study gave written consent obtained by the authors. Chi-
Sclerosis,33 and microvascular blood perfusion.34−36 In one squared tests were used to test for the prevalence of different
study within the Otolaryngology-Head and Neck Surgery comorbidities in both groups.
literature, EVM was used for noninvasive monitoring of
perfusion in free flaps.37 Using EVM, the degree in color
Recording of Videos
change allowed for intraoperative assessment of perfusion,
Each patient underwent video recording with a Canon
with the hope of using this technology to monitor perfusion
Powershot SD1400IS camera, with a 28-mm wide lens and
in the postoperative setting following anastomosis and inset
14.1-megapixel resolution. In order to ensure stability
of free flaps. To the author’s knowledge, EVM has not been
throughout the recordings, a tripod was used. Autofocus
used otherwise within the Laryngology or Otolaryngology-
was not used to ensure consistent image quality. Videos of a
Head and Neck Surgery literature.
frontal view were recorded using standard studio lighting
Our study investigated whether EVM may have diagnos-
with a white backdrop in a quiet room. Videos were
tic utility for the assessment of MTD. To the knowledge of
recorded with the patient at rest and during phonation while
the authors, correlations between extralaryngeal muscle
saying /i/. Each video was recorded for 10 seconds, and a
blood perfusion and a diagnosis of MTD have not been
consistent, stable video segment of exactly 4 seconds was
studied. It is known that, with skeletal muscle contraction,
spliced out to ensure a homogenous number and quality of
both arterial and venous blood flow are increased due to a
frames in each video for analysis. All videos were de-identi-
phenomenon known as exercise hyperemia.26−29 Therefore,
fied and stored on a secure computer.
we hypothesized that in patients with baseline muscle con-
traction as noted in MTD, there would be less potential for
muscle contraction during phonation. In turn, there would Eulerian Video Magnification
not be a large change in blood flow to extralaryngeal Analysis of video data was conducted following a protocol
muscles in patients with MTD. Using EVM, we aimed to similar to the original publication describing the process of
examine differences in extralaryngeal muscle perfusion dur- EVM.24 Change in pixel redness over time was extracted
ing rest and phonation in patients with and without MTD. and amplified from the captured video recordings; these
With this knowledge, EVM has the potential to alter our changes resembled sine waves in each subject, correspond-
understanding of the complex body motions involved in ing to their heart rate (Figure 1). From each sine wave, a
phonation. To our knowledge, this is the first study to use Fourier Transform was used to compute the subject’s heart
EVM within the field of Laryngology. rate and the intensity of perfusion. Four specific regions of
interest were selected for each video to focus on change in
blood flow: (1) medial forehead, (2) the body of sternoclei-
MATERIALS AND METHODS domastoid, (3) infrahyoid muscles, and (4) the wall next to
Participants the patient (Figure 2). The medial forehead was included as
The study was approved by the Institutional Review Board at a control, as the forehead musculature is not involved in
Drexel University College of Medicine in Philadelphia, Penn- phonation. The body of the sternocleidomastoid muscle
sylvania. Adult patients in a tertiary care laryngology practice was included due to its proximity to the extralaryngeal mus-
were recruited for the study between November 2016 and culature in an effort to ensure no perfusion changes in other
March 2017. All patients were already scheduled for assess- muscles of the neck during phonation. The wall served as a
ment of dysphonia and underwent standard evaluation in the control to ensure that the software was not confounded by
laryngology office, including videostroboscopy and voice lighting or video recording. The infrahyoid muscles were
team evaluation. Patients with MTD were deemed candidates the most visible muscles involved with phonation that could
for inclusion if they demonstrated supraglottic compression, be recorded while the patient was at rest and phonating.
pressed phonation, excessive strap muscle tension, and other The suprahyoid region was not adequately captured on
signs, such as laryngeal elevation during phonation. The video with the patient in a seated position. Therefore, it was
selected patients were those only with findings of MTD and not included in our analysis.
presented with findings typical of MTD. This diagnosis was In order to extract perfusion data from a patient video, this
confirmed independently by a laryngologist and a speech-lan- technology requires the selection and calibration of several
guage pathologist. Patients were excluded from the study if parameters. The amplified frequency range was set from
they did not speak English or were under the age of 18 years. 0.8 Hz to 1.4 Hz, representing a resting heart rate range
Twenty-four patients were recruited for the study. Due to of 48−84 beats per minute. The alpha level, a magnification
excess movement in videos, six patients were excluded from parameter, was set to 25. The sampling rate was set to 30
ARTICLE IN PRESS
Jason Adleberg, et al MTD Detection Using Video Magnification 3
frames per second, chromatic attenuation was set to 4, and from each subject with MTD were recorded at one sitting,
the magnification type was set to color magnification instead due to uncertain patient follow-up. Volunteer control sub-
of motion magnification. These values were chosen from orig- jects were available for repeat testing to ensure accuracy
inal studies of the software by its authors and kept constant and reliability of the calibration methodology.
for all participants in the study. As the sampling rate far
exceeded the heart rate in each captured video, no concern
was warranted that the sampling rate could confound results. Analysis of Perfusion Intensity Data
Once EVM was applied to the four areas of interest in each
video, perfusion frequencies and intensities were extracted
Calibration for Control Subjects and recorded. For each patient, percent change in perfusion
To ensure the accuracy of our approach, videos from sub- was calculated between values at rest versus during phona-
jects without MTD were collected during three sessions tion. Unpaired t tests were used to measure differences in
over a 2-week period. Values for each patient were com- percent change in perfusion between control and MTD
pared over multiple dates to ensure consistency. Paired patients.
t tests were used to determine whether there was significant
variation with values recorded for control subjects. Videos Influence of Age and Gender on Perfusion Intensity
and Ratios
The influence of age and gender was investigated evaluating
percent change in perfusion for all patients. Pearson’s corre-
lation coefficients and unpaired t tests were calculated.
Microsoft Excel was used to make all calculations.
RESULTS
Characteristics of MTD Patients and Normal
Speakers
Eighteen patients with MTD were included (12 women, 6
men; mean age 44 years, age range 22−87 years). Five con-
trol subjects without a history of MTD were recruited (three
women, two men; mean age 32 years, age range 29−38
years). Patients from both groups exhibited various comor-
bidities, although there were no significant differences in
prevalence between the two groups (Table 1).
TABLE 1.
Prevalence of Comorbidities and Surgical History of Patients With Muscle Tension Dysphonia (MTD) and Control Subjects
MTD Prevalence Control Prevalence
Condition (n = 18) (n = 5) P Value* Chi-Squared
Pulmonary disease 6 (33%) 3 (60%) 0.28 1.148
Neck surgery 4 (22%) 0 (0%) 0.25 1.271
Chest/lung surgery 1 (6%) 0 (0%) 0.58 0.301
Phonosurgery 2 (11%) 0 (0%) 0.44 0.576
Neurologic disorders 2 (11%) 0 (0%) 0.44 0.576
Musculoskeletal disorders 3 (17%) 0 (0%) 0.33 0.938
Cervical spine surgery 1 (6%) 0 (0%) 0.58 0.301
* Significant P value <0.05.
DISCUSSION
Calibration for Control Subjects This study compared blood perfusion levels of four regions
Data from five control subjects were collected during three in 18 patients with MTD and five patients, each recorded
sessions over a 2-week period. Using one-sample t tests, val- three times, without MTD. In this pilot study, our data sug-
ues were not found to vary significantly over the 2 week gest that EVM can be used in the diagnosis of MTD by
time period (Table 4), except for the wall measurement (P = focusing on the difference in perfusion to the infrahyoid
0.04, t = 4.84, df = 2) for Patient #3. muscles between rest and phonation.
Patients with MTD experienced little change in infra-
hyoid muscle perfusion during phonation, whereas con-
EVM Values trol subjects experienced a two-fold increase in perfusion.
Percent change in perfusion values was calculated for con- We hypothesize that since patients with MTD have a
trol and MTD patients. Only the infrahyoid muscles higher baseline state of muscle tension, there is less
TABLE 3.
Difference in Percent Change in Perfusion (%) Versus Male and Female Gender
Percent Change in
Percent Change in Perfusion in
Site of Perfusion Perfusion in Men (%) Women (%)
Measurement (n = 11) (n = 21) P Value* t Value†
Forehead 2.2 § 55.3 36.6 § 103.0 0.31 1.032
Sternocleidomastoid muscle 106.6 § 177.7 66.4 § 110.1 0.43 0.800
Infrahyoid muscles 100.1 § 174.2 20.2 § 86.9 0.09 1.751
Wall 27.6 § 76.4 11.2 § 39.7 0.42 0.817
* Significant P value <0.05.
†
Degrees of freedom (df) for all cases was 30.
ARTICLE IN PRESS
Jason Adleberg, et al MTD Detection Using Video Magnification 5
P Value t Value
0.67
1.08
4.84
1.42
0.95
patients. Since there is less change in muscle contraction,
this results in less change in perfusion. This decreased
change of perfusion was demonstrated with EVM tech-
0.57
0.39
0.04
0.29
0.44
nology. A change in perfusion of 0% or less was 75% sen-
sitive and 70% specific for diagnosis of MTD. Percent
change in perfusion was smaller in MTD patients than in
75 § 108.9
11.9 § 33.7
¡32.4 § 57.8
38.1 § 22.9
44.4 § 91.7
control patients, with this difference being statistically
significant (P = 0.04). Due to the small number of
Wall
107.7 § 211.8
210.7 § 290.1
Sternocleido-
¡16.6 § 31.8
32.3 § 59.4
Infrahyoid
105.1 § 193
TABLE 5.
Percent Change in Perfusion (%) for Control and Muscle Tension Dysphonia (MTD) Patients
Percent Change in Percent Change in
Perfusion, Control (%) Perfusion, MTD (%)
Region of interest (n = 5) (n = 18) P Value* t Value†
Forehead 14 § 69 31 § 103 0.58 0.562
Sternocleidomastoid 119 § 168 59 § 103 0.26 1.157
Infrahyoid 102 § 164 1 § 55 0.04* 2.189
Wall 28 § 68 6 § 39 0.30 1.062
* Significant P value <0.05.
†
Degrees of freedom (df) for all cases was 21.
discomfort. This position may have also caused other neck REFERENCES
musculature to become contracted during phonation and 1. Roy N. Differential diagnosis of muscle tension dysphonia and spasmodic
at rest, which could have been a confounder to our analysis dysphonia. Curr Opin Otolaryngol Head Neck Surg. 2010;18:165–170.
if this was applied. Finally, our approach was unable to 2. Houtz DR, Roy N, Merrill RM, et al. Differential diagnosis of muscle
separate the individual contributions of the omohyoid, ster- tension dysphonia and adductor spasmodic dysphonia using spectral
nohyoid, sternothyroid, and thyrohyoid muscles. We moments of the long-term average spectrum. Laryngoscope.
2010;120:749–757.
believe that higher video resolution capture may facilitate 3. Van Houtte E, Van Lierde K, Claeys S. Pathophysiology and treat-
the ability to do so. ment of muscle tension dysphonia: a review of the current knowledge.
Further study of EVM is warranted for patients with J Voice. 2011;25:202–207.
MTD, not only preceding but also following voice therapy. 4. Sama A, Carding PN, Price S, et al. The clinical features of functional
Clinicians could potentially use this software to aid in nar- dysphonia. Laryngoscope. 2001;111:458–463.
5. Koufman JA, Blalock PD. Functional voice disorders. Otolaryngol
rowing their differential diagnosis as well as to track patient Clin North Am. 1991;24:1059–1073.
progress during voice therapy. Depending on patient progress 6. Garaycochea O, Navarrete JMA, Del Río B, et al. Muscle tension dys-
as detected by perfusion via EVM, clinicians could alter treat- phonia: which laryngoscopic features can we rely on for diagnosis? J
ment regimens in order to improve patient outcomes. Future Voice. 2018;17:30560.
studies may also include the use of this technology to help dif- 7. Belafsky PC, Postma GN, Reulbach TR, et al. Muscle tension dyspho-
nia as a sign of underlying glottal insufficiency. Otolaryngol Head
ferentiate patients with MTD due to muscle-based hyper- Neck Surg. 2002;127:448–451.
function from those patients with emotional-based 8. Behrman A, Dahl LD, Abramson AL, et al. Anterior-posterior and
hyperfunction, which could also help to direct therapy. medial compression of the supraglottis: signs of nonorganic dysphonia
or normal postures? J Voice. 2003;17:403–410.
9. Andreassen ML, Litts JK, Randall DR. Emerging techniques in assess-
CONCLUSION ment and treatment of muscle tension dysphonia. Curr Opin Otolar-
This study investigated the utility of EVM in detecting yngol Head Neck Surg. 2017;25:447–452.
differences in perfusion to the extralaryngeal and, specifi- 10. Morrison MD, Rammage LA, Belisle GM, et al. Muscular tension
dysphonia. J Otolaryngol. 1983;12:302–306.
cally, infrahyoid muscles. Patients with MTD demon- 11. Morrison MD, Nichol H, Rammage LA. Diagnostic criteria in func-
strated little change in perfusion to the infrahyoid muscles tional dysphonia. Laryngoscope. 1986;96:1–8.
while phonating, but there was an increase in perfusion to 12. Khoddami SM, Nakhostin Ansari N, Izadi F, et al. The assessment
the same location in control subjects. This difference in methods of laryngeal muscle activity in muscle tension dysphonia: a
percent change in perfusion between control and MTD review. Sci World J. 2013;2013 507397.
13. Angsuwarangsee T, Morrison M. Extrinsic laryngeal muscular tension
patients makes EVM a potential tool for diagnosing in patients with voice disorders. J Voice. 2002;16:333–343.
MTD and for monitoring efficacy of treatment, however, 14. Mathieson L, Hirani SP, Epstein R, et al. Laryngeal manual therapy: a
the utility of EVM for these purposes certainly warrants preliminary study to examine its treatment effects in the management
additional study. of muscle tension dysphonia. J Voice. 2009;23:353–366.
15. Stepp CE, Heaton JT, Braden MN, et al. Comparison of neck tension
palpation rating systems with surface electromyographic and acoustic
Acknowledgments measures in vocal hyperfunction. J Voice. 2011;25:67–75.
16. Lowell SY, Kelley RT, Colton RH, et al. Position of the hyoid and lar-
We would like to thank Drexel University for their support
ynx in people with muscle tension dysphonia. Laryngoscope.
of this research. 2012;122:370–377.
17. Stepp CE, Heaton JT, Jette ME, et al. Neck surface electromyography
as a measure of vocal hyperfunction before and after injection laryngo-
SUPPLEMENTARY MATERIALS plasty. Ann Otol Rhinol Laryngol. 2010;119:594–601.
Supplementary material associated with this article can be 18. Redenbaugh MA, Reich AR. Surface EMG and related measures in
found in the online version at https://doi.org/10.1016/j. normal and vocally hyperfunctional speakers. J Speech Hear Disord.
jvoice.2019.02.006. 1989;54:68–73.
ARTICLE IN PRESS
Jason Adleberg, et al MTD Detection Using Video Magnification 7
19. Hocevar-Boltezar I, Janko M, Zargi M. Role of surface EMG in diag- 31. Aubakir B, Nurimbetov B, Tursynbek I, Varol HA. Vital sign monitor-
nostics and treatment of muscle tension dysphonia. Acta Otolaryngol. ing utilizing Eulerian video magnification and thermography. Conf
1998;118:739–743. Proc IEEE Eng Med Biol Soc. 2016;2016:3527–3530.
20. Stepp CE, Heaton JT, Stadelman-Cohen TK, et al. Characteristics of 32. Bennett SL, Goubran R, Knoefel F. The detection of breathing behav-
phonatory function in singers and nonsingers with vocal fold nodules. ior using Eulerian-enhanced thermal video. Conf Proc IEEE Eng Med
J Voice. 2011;25:714–724. Biol Soc. 2015;2015:7474–7477.
21. Van Houtte E, Claeys S, D’Haeseleer E, et al. An examination of sur- 33. Van Hillegondsberg L, Carr J, Brey N, Henning F. Using Eulerian
face EMG for the assessment of muscle tension dysphonia. J Voice. video magnification to enhance detection of fasciculations in people
2013;27:177–186. with amyotrophic lateral sclerosis. Muscle Nerve. 2017;56:1063–1067.
22. Stepp CE, Hillman RE, Heaton JT. Use of neck strap muscle intermus- 34. Adams F, Schoelly R, Schlager D, et al. Algorithm-Based Motion
cular coherence as an indicator of vocal hyperfunction. IEEE Trans Magnification for Video Processing in Urological Laparoscopy. J
Neural Syst Rehabil Eng. 2010;18:329–335. Endourol. 2017;31:583–587.
23. Stepp CE, Hillman RE, Heaton JT. Modulation of neck intermuscular 35. Rezaeian M, Georgevsky D, Golzan SM, Graham SL. High speed in-
Beta coherence during voice and speech production. J Speech Lang vivo imaging of retinal hemodynamics in a rodent model of hyperten-
Hear Res. 2011;54:836–844. sion. Conf Proc IEEE Eng Med Biol Soc. 2016;2016:3243–3246.
24. Wu HY, Rubinstein M, Shih E, et al. Eulerian video magnification for 36. Bennett SL, Goubran RA, Bennett B, Bennett RA, Knoefel F. The use
revealing subtle changes in the world. ACM Trans Graph. of a thermal camera and Eulerian enhancement in the examination of
2012;54:836–844. pedal pulse and microvascular health. Conf Proc IEEE Eng Med Biol
25. Eulerian Video Magnification. Public Code. Massachusetts Institution Soc. 2016;2016:1385–1388.
of Technology. http://people.csail.mit.edu/mrub/vidmag/#code. 37. Liu YF, Vuong C, Walker PC, et al. Noninvasive Free Flap Monitor-
26. Joyner MJ, Casey DP. Regulation of increased blood flow (hyperemia) ing Using Eulerian Video Magnification. Case Rep Otolaryngol.
to muscles during exercise: a hierarchy of competing physiological 2016;2016 9471696.
needs. Physiol Rev. 2015;95:549–601. 38. Wang L, Zhang L, Tang Y, Qiu L. The value of high-frequency and
27. Hellsten Y, Nyberg M, Jensen LG, et al. Vasodilator interactions in color Doppler ultrasonography in diagnosing congenital muscular tor-
skeletal muscle blood flow regulation. J Physiol. 2012;590:6297–6305. ticollis. BMC Musculoskelet Disord. 2012;13:209.
28. Mortensen SP, Saltin B. Regulation of the skeletal muscle blood flow in 39. Gijsbertse K, Goselink R, Lassche S, et al. Ultrasound Imaging
humans. Exp Physiol. 2014;99:1552–1558. of Muscle Contraction of the Tibialis Anterior in Patients with
29. Calbet JA, Lundby C. Skeletal muscle vasodilatation during maximal Facioscapulohumeral Dystrophy. Ultrasound Med Biol. 2017;43:
exercise in health and disease. J Physiol. 2012;590:6285–6296. 2537–2545.
30. Gambi E, Agostinelli A, Belli A, et al. Heart Rate Detection Using 40. Bhansing KJ, Hoppenreijs EP, Janssen AJ, et al. Quantitative muscle
Microsoft Kinect: Validation and Comparison to Wearable Devices. ultrasound: a potential tool for assessment of disease activity in juve-
Sensors (Basel). 2017;17:1776. nile dermatomyositis. Scand J Rheumatol. 2014;43:339–341.