Correlation Between the Basic Video Laryngostroboscopic Parameters and Multidimensional Voice Measurements  _ and †Viktoras Saferis, *Virgilijus Uloza, *Aurelija Vegiene, *yKaunas, Lithuania Summary: Objective. The aim of this study is to evaluate the correlations among the basic video laryngostroboscopic (VLS) parameters and vocal function assessed using a multidimensional set of perceptive, acoustic, aerodynamic, and subjective measurements. Methods. Digital VLS recordings and multidimensional voice assessment were performed for 108 individuals, namely 26 healthy and 82 patients with mass lesions of vocal folds and paralysis. The VLS variables (glottal closure, regularity, mucosal wave on the affected/healthy side, symmetry of vibration, and symmetry of image) were rated and quantified on a visual analog scale. Correlations among the VLS parameters and results of acoustic voice analysis and voice range profile (VRP), data of subjective (voice handicap index [VHI] and glottal function index [GFI]), perceptual (G-grade, R-roughness, B-breathiness, A-asthenic scale), and dysphonia severity index (DSI) measurements were tested using Pearson’s correlation coefficient. Results. The intercorrelations of VLS parameters in vocal performance were moderate to strong and statistically significant for the entire functional measurements obtained through different modalities. The definitive correlations between VLS and VRP parameters were as follows: r ¼ 0.69–0.79 for normal profile coverage, r ¼ 0.67–0.76 for dynamic intensity, and r ¼ 0.67–0.74 for maximum intensity. All VLS parameters correlated moderately with VHI, GFI, and DSI (r ¼ 0.5–0.65, r ¼ 0.4–0.57, and r ¼ 0.61–0.69, respectively). The strongest correlations were found between VLS parameters and G factor of the GRBA scale (r ¼ 0.68–0.88). Conclusions. Correlation analysis of the vibratory pattern of the vocal folds obtained via VLS provides more comprehensive insight into the pathophysiology of phonation and suggests the documented and measurable evidence of complex mechanisms of vocal outcome. Key Words: Laryngostroboscopy–Acoustic voice assessment–Voice range profile–VHI–DSI.

INTRODUCTION Qualified and complex evaluation of patients with dysphonia and diagnostics of laryngeal diseases typically include patient’s complaints, history, perceptual assessment of voice quality and severity of dysphonia, measurement of acoustic and aerodynamic voice parameters, and visualization of larynx.1 This reflects voice quality as the outcome of a perceptual analysis of an acoustic signal generated mainly by vibrations of the vocal folds during phonation.2 Video laryngostroboscopy (VLS) currently represents the most important and the most commonly used well-established method to visualize larynx and vocal fold vibrations.3–6 Laryngeal visualization via VLS has been shown to be critical to diagnose the underlying cause of hoarseness, thereby increasing the accuracy of diagnostics up to 68.3%.7 This unalterable and clinically feasible imaging tool is also implemented into practice to assess the outcomes of therapy of laryngeal diseases or functional results of phonosurgical interventions.1,2,4 However, a real value of VLS for diagnosis still needs more comprehensive scientific evidence. Accepted for publication June 12, 2013. From the *Department of Otolaryngology, Academy of Medicine, Lithuanian University of Health Sciences, Eiveniu 2, Kaunas, Lithuania; and the yDepartment of Physics, Mathematics & Biophysics, Lithuanian University of Health Sciences, Eiveniu 4, Kaunas, Lithuania. Address correspondence and reprint requests to Aurelija Vegien_e, Department Otolaryngology, Academy of Medicine, Lithuanian University of Health Sciences, Eiveniu 2, LT50009, Kaunas, Lithuania. E-mail: [email protected] Journal of Voice, Vol. 27, No. 6, pp. 744-752 0892-1997/$36.00 Ó 2013 The Voice Foundation http://dx.doi.org/10.1016/j.jvoice.2013.06.008

Moreover, the subjective nature of the interpretation of the VLS examination results significantly reduces the reproducibility and the use of VLS as a research tool or as a quantitative instrument for assessment of laryngeal phonatory function. To avoid this limitation, several methods of assessment and quantification of VLS findings have been suggested elaborating different VLS rating forms. A number of VLS variables that have been evaluated include a rather wide diversity of parameters, that is, periodicity and amplitude of vocal fold vibration, mucosal wave, vertical level, glottal closure, phase closure, phase symmetry, and presence of nonvibrating portions of the vocal fold.1,3,8–10 Unfortunately, a number of VLS variables and peculiarities of the VLS parameters presented in literature differed from study to study; therefore, data and results of different studies sometimes were hardly compatible. The most recent studies show the tendency to optimize the VLS evaluation for clinical settings reducing the number of VLS parameters and exhibiting the most reliable judgments, thus supporting the concept that even small set of stroboscopic ratings is an adequate representation for most of the variance of all laryngostroboscopic characteristics.11–13 In earlier research on quantification of the basic VLS parameters, we found a high interrater and intrarater reliability for most basic VLS parameters that revealed high sensitivity and specificity distinguishing healthy and pathological voice patient groups.13 However, there is a paucity of information in literature on correlations between data of VLS examination and the results of assessment of laryngeal phonatory function using other measurements.

Virgilijus Uloza, et al

745

Correlation Analysis of the Vibratory Pattern of the Vocal Folds

The aim of this study is to evaluate correlations among the basic VLS parameters and vocal function by using a multidimensional set of perceptive, acoustic, aerodynamic, and subjective measurements. MATERIALS AND METHODS A study group consisting of 108 individuals was examined at the Department of Otolaryngology of Lithuanian University of Health Sciences, Kaunas, Lithuania. The normal voice subgroup was composed of 26 selected healthy volunteer individuals from a random group of 42 subjects who considered their voice as normal. They had no complaints concerning their voice and no history of chronic laryngeal diseases or other long-lasting voice disorders. The voices of this group of individuals were also evaluated as healthy voices by clinical voice specialists. Furthermore, no pathological alterations in the larynx of the subjects of the normal voice subgroup group were found during VLS. Acoustic voice signal parameters of these normal voice subgroup subjects that were obtained using Voice Diagnostic Center lingWaves software, Version 2.5 (WEVOSYS, Forchheim, Germany) were within the normal range. The pathological voice subgroup consisted of 82 patients who represented a rather common, clinically discriminative group of laryngeal diseases, that is, mass lesions of vocal folds and paralysis. Mass lesions of vocal folds included in the study consisted of nodules, polyps, cysts, papillomata, keratosis, and carcinoma. Pathological voice group patients were recruited from consecutive patients who were diagnosed with the laryngeal diseases mentioned previously. The clinical diagnosis was based on typical clinical signs revealed during VLS and direct microlaryngoscopy. The final diagnosis was proven by the results of histological examination of removed tissue in all cases of mass lesions of vocal folds. The required sample size for achieving 80% power was fulfilled. Demographic data of the total study group and diagnoses of pathological voice group are presented in Table 1. These patients were serially enrolled and, therefore, likely represented the real incidence of pathologies in our series.

Digital high-quality VLS recordings were performed with the XION EndoSTROB DX device (XION GmbH, Berlin, Germany) using a 90 rigid endoscope. The subjects were seated for the VLS examination. The VLS examination and recordings were performed during modal phonation, that is, each subject was asked to sustain the vowel ‘‘ee’’ at a steady, comfortable pitch and loudness. Phonation time was kept long enough to allow for registration of a sustained phonation and at least one complete cycle of vibration. The following basic VLS parameters were evaluated and quantified using a 100-mm long visual analog scale (VAS): glottal closure—longitudinal, oval, hourglass-shaped gap; regularity of vibrations—defined as the degree to which one phonatory cycle suits the next; mucosal wave on affected side; mucosal wave on healthy side; symmetry of glottal image; and symmetry of vibration.13 A score of ‘‘zero’’ (extreme left) meant normal perception of the parameter (no deviance), whereas ‘‘100’’ (extreme right) meant extreme deviance of the parameter evaluated.1,14 Digital VLS recordings were rated two times with the time interval of 1 year by three experienced laryngologists/phoniatricians. To evaluate interrater and test-retest reliability, the intraclass correlation coefficients (ICCs) were calculated, and moderate-to-almost perfect levels (ICC 0.46–0.90) of interrater reliability were revealed for most of the basic VLS parameters. The ICC of the test-retest reliability was 0.71–0.95, P < 0.001, respectively. To verify the reliability of visual-perceptive measurements of distinct VLS parameter—glottal closure—objective analysis of frozen VLS images was performed and relative glottal area— RGA (RGA ¼ glottal area of the maximum opening [GAmax]/ glottal area of maximum closure [GAmin]) was calculated. Detailed description of the investigation of reliability of VLS parameters, sensitivity, and specificity of VLS parameters separating normal voice and pathological voice groups was presented in our previous study elsewhere.13 Voice assessment Vocal function was evaluated using a multidimensional set of perceptive, acoustic, aerodynamic, and subjective measurements

TABLE 1. Demographic Data of the Study Group Gender Diagnosis

Age (y)

Total Number (n ¼ 108)

Female (n ¼ 58)

Male (n ¼ 50)

x

±SD

30 9 4 9 11 8 8 3 26

12 8 4 9 0 7 0 3 15

18 1 0 0 11 1 8 0 11

44.2 38.0 32.3 52.7 58.8 54.3 54.1 38.0 30.8

11.8 11.5 8.5 8.5 5.7 13.8 14.2 21.4 10.9

Polyp Nodules Cyst Reinke’s hyperplasia Carcinoma Vocal fold paralysis Keratosis Papillomatosis Normal voice Abbreviations: x, mean; SD, standard deviation.

746 according to the protocol established by the Committee on Phoniatrics of the European Laryngological Society.1 Perceptual voice evaluation Digitized voice recordings of a standard phonetically balanced passage read out on habitual pitch and loudness were subjected to perceptual evaluation of dysphonia by the same experts as for VLS rating. The GRBAS scale consisting of G (grade), R (rough), B (breathy), and A (asthenic) factors was adapted to assess dysphonia. A four-point grading system (0—normal, 1— slight, 2—moderate, and 3—extreme) was used in this study to quantify perceptual assessment of dysphonia.15 Acoustic analysis Digitized voice recordings of sustained phonation of the vowel sound /a/ (as in the English word ‘‘large’’) were obtained in a soundproof box by the lingWAVES sound pressure level (SPL) meter microphone placed at a 30.0 cm distance from the mouth and at about 90 microphone-to-mouth angle. The typical original utterance duration before processing was about 2–5 seconds. The voice recordings were made in the ‘‘wav’’ file format at the rate of 44.100 samples per second. Sixteen bits were allocated to one sample. The average length of each recording was 2.4 seconds. The very first parts of the phonation sample (0.25 seconds) were cut off, and the subsequent 2.0 seconds of the sample was used for the measurements, thus reducing the variability resulting from sampling errors. The remaining parts of the sustained vowel /a/ were discarded. This was done to ensure that the rather unstable beginning and the end of the sampling had no effect on the final result. Segments of at least 2second duration of the sustained vowel /a:/ of three separate voice samples from each recording session were analyzed using Voice Diagnostic Center lingWaves software, Version 2.5 software. Acoustic voice signal data were obtained for: (1) fundamental frequency (Mean F0, hertz), (2) standard deviation (SD) of F0 (SDF0, hertz), (3) maximum F0 (MaxF0, hertz), (4) minimum F0 (MinF0, hertz), (5) percent of jitter, (6) percent of shimmer, and (7) glottal noise energy (GNE). Voice range profile The voice range profiles (VRPs) were recorded according to the recommendations of the Union of European Phoniatricians.16 The lingWaves SPL meter microphone and Voice Diagnostic Center lingWaves software, Version 2.5 were used for the registration and assessment of the VRPs. The acoustic environment was a soundproof box of 2.0 3 3.0 m with interior ambient noise lower than 30 dB. The following instructions were given to the subjects: after the investigator generates the tone using lingWAVES system, the subject is asked to imitate the tone by phonating the vowel /a/. When the tone is correctly imitated, the individual is instructed to produce the tone as softly and as loudly as he/she can. The frequencies are obtained in descending order starting from 261.6 Hz (C4) to the lowest possible, then followed by ascending order to the highest possible and are recorded in semitones. The following parameters of VRP were evaluated: (1) maximum tone (MaxTon) in hertz, (2) minimum tone (MinTon) in

Journal of Voice, Vol. 27, No. 6, 2013

hertz, (3) tone diapason (hemitones), (4) maximum intensity (MaxInt) in decibels adjusted, (5) minimum intensity (MinInt) in decibels adjusted, (6) dynamic intensity in decibels adjusted, and (7) normal profile coverage in percent (Norm%). Aerodynamics Maximum phonation time (MPT) in seconds was used as the simplest aerodynamic parameter of voicing. The prolongation of /a:/ for as long as possible after maximal inspiration and at a spontaneous, comfortable pitch and loudness level was used. The longest time record from three consequent trials was selected. Dysphonia severity index (DSI) was calculated using lingWaves VDC Vospector analysis. According to Wuyts et al,16 DSI is based on the weighted combination of the following selected set of voice measurements: highest frequency in hertz, lowest intensity in decibels adjusted, MPT in seconds, and jitter in percentage. The DSI for perceptually normal voices equals +5 and for severely dysphonic voices 5. The more negative the patient’s DSI, the worse is his or her vocal quality.17 Subjective voice evaluation Subjective voice evaluation was performed by the investigators using validated Lithuanian versions of questionnaires of Voice Handicap Index (VHI_LT)18—to evaluate the level of handicap of a person resulting from a voice disorder—and Glottal Function Index (GFI_LT)19—to assess patients’ self-perception of voice disorder. Normative data of these questionnaires are presented elsewhere.18,19 Statistics Statistical analysis was performed using IBM SPSS Statistics for Windows, Version 20.0 (IBM Corporation Software, Armonk, NY). Data were presented as mean (x) ± SD. The Student’s t test was used for testing hypotheses about equality of the mean. The size of the differences among the mean values of the groups was evaluated by estimation of type I and type II errors (a and b) of the tests. The size of the difference was considered to be significant if a  .05 and b  .2. For testing hypotheses of independence, the c2 test was used. Discriminant analysis was performed to determine the limiting values of VLS parameters discriminating normal and pathological voice groups. The correlations among VLS parameters and multidimensional sets of acoustic, perceptive, and subjective voice measurements were tested using Pearson’s correlation coefficient (rho). The level of statistical significance by testing statistical hypothesis was 0.05. RESULTS The mean values and SDs of the VLS parameters in normal voice and pathological voice subgroups are presented in Table 2. A statistically significant difference (P < 0.001) between patients’ subgroup and the normal voice subgroup was found of all VLS parameters measured, with the patients having much worse results, thus demonstrating measurable evidence of the differences between a normal and a deteriorated phonation pattern. The limiting values of the VLS parameters were

Virgilijus Uloza, et al

747

Correlation Analysis of the Vibratory Pattern of the Vocal Folds

TABLE 2. Values of VLS Parameters in Normal and Pathological Voice Groups VAS Points Normal Voice Group, n ¼ 26

Difference

x

SD

x

SD

Absolute

%

Limiting Value

P

b*

8.36 6.88 4.18 4.21 1.33 1.38 137.16y

6.16 4.73 4.72 4.80 2.00 2.01 98.65

53.29 72.47 81.08 59.61 74.22 70.96 10.45y

17.30 18.80 19.30 34.90 24.30 26.10 6.32

44.93 65.59 76.90 55.40 72.89 69.58 126.71y

537.40 953.30 1839.70 1315.90 5480.50 5042.00 92.40y

30.80 39.70 42.60 31.90 37.40 36.20 73.80y

0.000 0.000 0.000 0.000 0.000 0.000 0.000

0.000 0.000 0.000 0.000 0.000 0.000 0.001

VLS Parameters Glottal closure Regularity Mucosal wave on affected side Mucosal wave on healthy side Symmetry of vibration Symmetry of glottal image Relative glottal area

Pathological Voice Group, n ¼ 82

Abbreviations: VLS, video laryngostroboscopy; VAS, visual analog scale; x, mean; SD, standard deviation. * Statistically significant difference between the groups, computed using a ¼ .05. y Relative measure.

determined as the optimum point for separating normal and pathological voice subgroups. Therefore, the parametric values of the VLS parameters that were found to be larger than the limiting values were considered as pathological. Furthermore, results of this study demonstrated that even in normal voice subgroup, the mean values of VLS parameters were found to be above the ‘‘zero,’’ or ‘‘ideal perception level,’’ reflecting some natural imperfectness of phonation pattern. Rather unexpectedly, an objective measure of RGA showed a statistically significant, however lesser, difference (92.4%) between normal and pathological voice subgroups. Table 3 presents correlations between VLS and acoustic voice parameters. In general, the main objective acoustic voice parameters correlated significantly with VLS parameters assessed on VAS. As shown in Table 3, statistically significant moderate correlations between acoustic voice parameters, re-

flecting perturbations of voice signal, that is, jitter, shimmer and SDF0, and VLS, were revealed. The GNE reflecting turbulent GNE and including pitch and amplitude perturbation showed statistically significant slight-to-moderate negative correlations with VAS-assessed VLS parameters. However, correlations among GNE and VLS glottal closure and objective RGA parameter were weak. Slight statistically significant correlations between MeanF0 and VLS parameters were detected, except for glottal closure and mucosal wave on healthy side. Acoustic MinF0 correlated slight to moderate with VLS parameters; however, MaxF0 did not. An objective measure of RGA showed a weak positive correlations to jitter, shimmer, SDF0, and GNE. In Table 4, correlations between VLS and VRP parameters are presented. From these results, the VLS parameters revealed statistically significant moderate-to-strong correlations to all

TABLE 3. Correlations Between VLS and Acoustic Voice Parameters Deviance of VLS Parameters

Acoustic Parameters MeanF0 SDF0 MaxF0 MinF0 Jitter Shimmer GNE

Glottal Closure

Regularity

Mucosal Wave on Affected Side

Mucosal Wave on Healthy Side

Symmetry of Vibration

Symmetry of Glottal Image

RGA

0.28* 0.34* 0.15 0.39* 0.42* 0.48* 0.31*

0.24* 0.34* 0.11 0.33* 0.39* 0.44* 0.26*

0.03 0.21* 0.14 0.09 0.24* 0.29* 0.23*

R 0.09 0.35* 0.05 0.24* 0.50* 0.53* 0.23*

0.23* 0.31* 0.13 0.36* 0.44* 0.48* 0.38*

0.21* 0.36* 0.09 0.35* 0.49* 0.53* 0.39*

0.14 0.27* 0.02 0.25* 0.43* 0.46* 0.37*

Abbreviations: VLS, video laryngostroboscopy; RGA, relative glottal area; R, Pearson’s correlation coefficient; MeanF0, fundamental frequency; SDF0, standard deviation of F0; MaxF0, maximum F0; MinF0, minimum F0; GNE, glottal noise energy. * P < 0.05.

748

Journal of Voice, Vol. 27, No. 6, 2013

TABLE 4. Correlations Between VLS and VRP Parameters Deviance of VLS Parameters Glottal Closure

Regularity

Mucosal Wave on Affected Side

Mucosal Wave on Healthy Side

Symmetry of Glottal Image

RGA

0.69* 0.02 0.67* 0.67* 0.48* 0.68* 0.74*

0.64* 0.09 0.66* 0.67* 0.47* 0.68* 0.71*

0.48* 0.18 0.57* 0.48* 0.38* 0.53* 0.67*

R

VRP Parameters MaxTon MinTon Tone diapason MaxInt MinInt DinamInt Norm%

Symmetry of Vibration

0.62* 0.21* 0.72* 0.74* 0.52* 0.73* 0.76*

0.69* 0.08 0.70* 0.72* 0.55* 0.76* 0.77*

0.67* 0.10 0.72* 0.72* 0.50* 0.74* 0.79*

0.54* 0.17 0.63* 0.63* 0.47* 0.67* 0.69*

Abbreviations: VLS, video laryngostroboscopy; VRP, voice range profile; RGA, relative glottal area; R, Pearson‘s correlation coefficient; MaxTon, maximum tone; MinTon, minimum tone; MaxInt, maximum intensity; MinInt, minimum intensity; DinamInt, dynamic intensity; Norm%, normal profile coverage in percent. * P < 0.05.

VRP parameters, except the MinTon in hertz. A resumptive VRP parameter—Norm%—demonstrated negative strong correlations to all VLS parameters. However, an objective measure of RGA indicated only slight correlations to four of seven VRP parameters. Thus, objective measures of RGA did not show stronger correlations between this measure and VRP parameters when compared with VLS/VRP correlations. Results in Table 5 show correlations between VLS and DSI parameters. From the data presented in Table 5, all VLS parameters correlate significantly and moderately with DSI as an aggregate measure, and with the separate components of the DS individually. However, the DSI as a compound measure those that includes voice perturbation measurements, and voice capabilities measurements demonstrate a tendency to statistically significant upper-limit moderate correlations to VLS parameters. In Table 6, correlations between VLS parameters and parameters of subjective voice evaluation using self-perception instru-

ments, that is, VHI_LT and GFI_LT are presented. The VHI_LT measuring the influence of voice problems on a patient’s quality of life and quantifying the biopsychosocial impact of a voice disorder correlated statistically significant and moderate to all VLS parameters. Analyzing each VHI_LT domain separately, moderate correlations with VLS parameters were also revealed. However, physical (P) VHI_LT domain showed tendency to more strong correlations with VLS parameters. The GFI_LT presenting a short four-item symptom-focused battery and assessing the degree of vocal dysfunction based on the patient’s self-perception revealed statistically significant moderate correlations with all VLS parameters. This could have been anticipated because a strong correlation between the VHI_LT and GFI_LT was determined in a previous study.19 Results of correlation analysis between VLS parameters and perceptual voice evaluation using GRBA factors scale are presented in Table 7. As shown in Table 7, perceptual evaluation of severity/grade of hoarseness (G factor) revealed statistically

TABLE 5. Correlations Between VLS and DSI Parameters Deviance of VLS Parameters

DSI Parameters Jitter MPT MinInt MaxTon DSI

Glottal Closure

Regularity

Mucosal Wave on Affected Side

Mucosal Wave on Healthy Side

Symmetry of Vibration

Symmetry of Glottal Image

RGA

0.42* 0.50* 0.48* 0.69* 0.64*

0.39* 0.52* 0.47* 0.64* 0.61*

0.24y 0.67* 0.38* 0.48* 0.51*

R 0.50* 0.63* 0.52* 0.62* 0.69*

0.44* 0.52* 0.56* 0.69* 0.67*

0.49* 0.53* 0.50* 0.67* 0.68*

0.43* 0.51* 0.47* 0.54* 0.62*

Abbreviations: VLS, video laryngostroboscopy; DSI, dysphonia severity index; RGA, relative glottal area; R, Pearson’s correlation coefficient; MPT, maximum phonation time; MinInt, minimum voice intensity; MaxTon, maximum tone. * P ¼ 0.00. y P ¼ 0.01.

Virgilijus Uloza, et al

749

Correlation Analysis of the Vibratory Pattern of the Vocal Folds

TABLE 6. Correlations Between VLS Parameters and Subjective Voice Evaluation (VHI_LT and GFI_LT) Deviance of VLS Parameters Glottal closure Regularity Mucosal wave on affected side Mucosal wave on healthy side Symmetry of vibration Symmetry of glottal image RGA

R

F

E

P

VHI_LT

GFI_LT

0.61* 0.55* 0.55* 0.47* 0.43* 0.43* 0.45*

0.51* 0.46* 0.49* 0.43* 0.38* 0.40* 0.45*

0.70* 0.67* 0.71* 0.56* 0.58* 0.59* 0.61*

0.65* 0.60* 0.63* 0.52* 0.50* 0.51* 0.54*

0.57* 0.54* 0.56* 0.48* 0.41* 0.40* 0.50*

Abbreviations: VLS, video laryngostroboscopy; VHI_LT, Lithuanian version of voice handicap index; GFI_LT, Lithuanian version of glottal function index; F, functional; E, emotional; P, physical; R, Pearson’s correlation coefficient, RGA, relative glottal area. * P ¼ 0.00.

significant strong correlation with VLS parameters. The R, B, and A factors showed moderate correlation to VLS parameters. However, an objective measure of glottal closure—RGA— indicates significant moderate negative correlations with GRBA factors. Thus, these results confirm a direct and strong impact of deteriorated phonation pattern measured by VLS on perception of voice quality.

Therefore, in clinical settings, the VLS is considered to be the gold standard in determining the pathological changes and the vibratory mode of the vocal fold. To validate the VLS quantification measures, objective or subjective, results have to be related to perceptual and acoustical voice quality parameters and aerodynamic analyses.2 Comparisons and correlations from VLS findings with other clinical assessments, such as acoustic voice analysis and perceptual voice quality judgments, provide objective evidence of dynamic voice events and are important to interpret the results of the techniques and to increase their clinical relevance.24 This study presents the first comprehensive attempt at defining relationships between measures derived from VLS assessment of vocal fold vibration and perceptual, subjective to instrumental assessments of the patients’ voices. In this study, it was found that the parameters obtained through different modalities (VLS, acoustic voice analysis, perceptual and subjective voice evaluation, and VRP) showed significant correlations with quantitative VLS measurements. This could have been anticipated because each of the modalities measure different aspects of voice production, and therefore together represent as complementary measurements. However, data in literature on this subject are rather sparse and fragmentary. Some studies have attempted to measure the glottal area from VLS images objectively and to determine the relationship between the relative glottal area and the degree of breathiness in

DISCUSSION Voice represents a complex and multidimensional phenomenon and therefore should be investigated by means of voice quality and vocal function analyses.20 The following most common approaches of clinical assessment of various voice function aspects include: (1) subjective and perceptual evaluation of voice quality, (2) acoustic voice signal analysis, (3) evaluation of functional voice capabilities—VRP, (4) aerodynamic assessment, and (5) endoscopic imaging of vocal fold vibration (VLS, laryngeal high-speed videoendoscopy [HSV], and kymography).21 The VLS is currently the most useful tool in demand for the examination of vocal fold vibration of normal and pathological voices. Although VLS does not show the fine details of each vibratory cycle, it allows the estimation of a vibratory pattern averaged over many successive cycles and can provide realtime information about abnormal vocal fold vibration.22,23

TABLE 7. Correlations Between VLS Parameters and Perceptual Voice Evaluation (GRBA Factors) Deviance of VLS Parameters Glottal closure Regularity Mucosal wave on affected side Mucosal wave on healthy side Symmetry of vibration Symmetry of glottal image RGA

R

G

R

B

A

0.77* 0.83* 0.88* 0.68* 0.75* 0.75* 0.64*

0.49* 0.65* 0.71* 0.55* 0.65* 0.68* 0.46*

0.74* 0.75* 0.75* 0.69* 0.59* 0.57* 0.55*

0.56* 0.61* 0.54* 0.49* 0.44* 0.38* 0.32*

Abbreviations: VLS, video laryngostroboscopy; G, grade; R, roughness; B, breathiness; A, asthenic; R, Pearson’s correlation coefficient; RGA, relative glottal area. * P ¼ 0.00.

750 dysphonic patients. These investigators found a positive correlation between glottal area and breathy voice in unilateral vocal fold paralysis and confirmed that even only measurement of VLS relative glottal area provides a direct assessment of glottal incompetence and objectively demonstrates the effect of surgical medialization or augmentation of vocal fold with autologous fascia.25–27 The measured maximum postoperative gap between vocal folds (gap index) correlated strongly and statistically significantly not only to perceived voice quality (grade of hoarseness and breathiness) but also to the acoustic voice parameters, namely jitter, shimmer, and noise-toharmonics ratio (NHR).26 There was also a statistically significant strong correlation between VLS mucosal wave phase synchrony and perceived breathiness, as well as between phase synchrony and measured jitter, shimmer, and NHR.27 In contrast, Verdonck-de Leeuw et al23 did not find significant correlations between acoustic voice quality measures on the sustained vowel /a/ and stroboscopic measures. In a recent study, Kelley et al11 identified the relationship between the stroboscopic ‘‘vibration factor’’ and ‘‘edge factor’’ and a mean perceived clinical rating for severity of dysphonia. Results of the present study determined statistically significant moderate correlations between acoustic voice parameters reflecting perturbations of voice signal, that is, jitter, shimmer and SDF0, and VLS parameters, reflecting asymmetrical vibratory pattern. Thus, our data confirm sensitivity of VLS measurements to vocal fold asymmetry assuming that severity and not just presence of asymmetry is the key aspect of the variable.28 Acoustic GNE parameter related to turbulent glottal noise showed statistically significant slight-to-moderate negative correlations with VAS-assessed VLS parameters. Objective measure of RGA presented weak positive correlations to jitter, shimmer, SDF0, and GNE. However, the DSI as complex measure, including voice perturbation and some voice capabilities measurements, demonstrated statistically significant upperlimit moderate correlations to VLS parameters. It can be assumed that deteriorated phonation pattern detected and quantified from VLS would be reflected in reduction of phonation capabilities, that is, measured by means of VRP. However, to our best knowledge, there are only a few data in literature on comparison of VLS to VRP results. To assess voice characteristics of patients following radiotherapy for early glottal cancer, Verdonck-de Leeuw et al23 used a multidimensional analysis protocol including vocal function and quality measures and VLS. Some moderate correlations between VRP and stroboscopic measures were found, that is, pitch range correlated with VLS ‘‘supraglottic edema’’ and intensity range correlated with stroboscopic ‘‘vascular injection’’ and ‘‘vocal fold edge’’ measures.20 In another study, subjects with an insufficient glottal closure on VLS showed a higher percentage of phonationassociated vocal fold alterations (ie, vocal nodules) and reached lower maximum SPLs defined by VRP in female students.29 Results of the present study show statistically significant moderate-to-strong correlations between VLS and VRP parameters, confirming a significant impact of deteriorated VLS pattern reducing functional vocal capabilities, that is, maximum tone, tone diapason, maximum intensity, and dynamic intensity.

Journal of Voice, Vol. 27, No. 6, 2013

Moreover, the resumptive VRP parameter—Norm%—demonstrated negative strong correlations to all VLS parameters. Generally, correlations between VLS and VRP parameters were stronger when compared with correlations between VLS and acoustic parameters. This is because both VLS and VRP reflect more dynamic aspects of the phonation process, rather than objective ‘‘momentum picture’’ of phonation as presented by acoustic analysis. Strong impact of deteriorated phonation pattern measured by VLS on perception of voice quality was confirmed by results of correlation analysis in our series. Perceptual evaluation of grade of dysphonia (G factor) correlated significantly and strongly with all VLS parameters. Strong correlation between B factor and VLS ‘‘glottal closure’’ parameter demonstrates that incomplete glottal closure and air leakage are perceived mainly as breathiness. Perceived roughness (R) correlated moderate to strongly with VLS parameters with respect to asymmetrical vibration pattern. Lau et al30 determined the positive correlation between VHI and quantitative VLS for patients undergoing injection laryngoplasty for unilateral vocal paralysis. Correlation coefficients between VHI and VLS parameter ‘‘glottic closed phase’’ showed moderate-to-strong correlation, whereas VLS ‘‘glottic open’’ area and ‘‘wave duration’’ parameters showed weak-tomoderate correlation, thus suggesting that duration of vocal fold closure during the glottic cycle best represents patients’ subjective postprocedure outcome.30 Results of the present study are in concordance with the data presented above and extend the knowledge on influence of deteriorated phonation on the perceived handicap by the patient self, as measured with VHI_LT. Data of our study suggest that the strongest impact of the voice on the perception of laryngeal discomfort or negative voice output characteristics (P domain) is in direct correlation with the disturbances of phonation pattern quantified on VAS from VLS. It should be noted that the short four-item GFI_LT, presenting symptom-focused battery to assess the degree of vocal dysfunction based on patient’s self-perception, revealed statistically significant moderate correlations with all VLS parameters, being consistent to the VHI_LT data. This confirms validity of GFI_LT as sensitive and suitable instrument to assess the presence and degree of self-percepted vocal dysfunction. On the other hand, it suggests that in clinical settings VHI_LT may be substituted with shorter GFI_LT questionnaire. It can be assumed that quantification of VLS parameters on VAS is not a true quantification; rather, it is a qualitative measure of subjective perception. Therefore, comparison of VLS ratings with objective data obtained from other laryngeal visualization methods is of great interest. In the study performed by Verdonck-de Leeuw et al,2 video kymographic images of deviant or irregular vocal fold vibration were compared with the synchronously recorded acoustic speech signals. A clear relation was shown between laryngeal video kymographic image sequences and acoustic speech signals. The effect of irregular or incomplete vocal fold vibration patterns was recognized in the amount of perceived breathiness and roughness and by the harmonics-to-noise ratio in the speech

Virgilijus Uloza, et al

Correlation Analysis of the Vibratory Pattern of the Vocal Folds

signal. Various other studies have determined a high intrarater and interrater reliability in assessment of left-right phase asymmetry using stroboscopy, both in vocally healthy individuals and subjects with voice disorders. Mild correlations between visual judgments and objective measurements were found revealing stroboscopy sensitive, but most likely not specific, to phase asymmetries.28 However, the validity of stroboscopy-based judgments of phase asymmetry was called into question owing to lower correlations with an objective measure of phase asymmetry as compared with laryngeal HSV-based modalities.24,31 Laryngeal HSV is frequency independent and has the potential to overcome the limitations of stroboscopy in the analysis of aperiodic vocal fold motion to allow for the direct assessment and study of vocal fold vibratory irregularities. However, the scales for rating irregularities via HSV are likely different from that used for stroboscopy.24 Comparing the accuracy of the VLS and HSV for visualizing periodic vocal fold vibrations, no significant differences were found in a group of vocally healthy subjects, except for visual judgments of symmetry.32 However, in the group of voice disorders, including disorders with aperiodic voices, HSV proved to be significantly more accurate and interpretable than VLS, and HSV allowed for observation of phase asymmetry when stroboscopy did not.24,33 Objective measures from HSV-derived playbacks provided higher precision in quantifying both glottal width and period irregularity in comparison with visual ratings of VLS.24 Mehta et al34 in their study demonstrated that laryngeal HSV recordings of human subjects revealed no direct correlation between vocal fold vibratory asymmetries and acoustic spectral tilt measures. However, the more recent study of Mehta et al34 defined significant correlations between physiological measures derived from HSV assessment of vocal fold vibration and acoustic cepstral-based measures before and after phonomicrosurgery for vocal fold lesions.31 Acoustic cepstral peak magnitude correlated significantly with the HSV-based measures, that is, average speed quotient and F0 deviation. The acoustic jitter correlated significantly with the SD of left-right phase asymmetry and the SD of left-right amplitude asymmetry, respectively. In the group of patients with glottic cancer, significant, however, moderate correlations between acoustic jitter and the SD of HSV measures in the symmetry of phase and amplitude across the vibratory cycles were found.35 Thus, data in literature obtained from these two different laryngeal visualization modalities are rather contradictory and therefore may be compatible only to some extent. Nevertheless, results of the present study are in concordance with the data of the literature mentioned previously. The general tendency determined is the same: more the deteriorated vibratory pattern was detected by different methods of laryngeal visualization, more the prominent deviances in acoustic voice signal were revealed. Results of the present study indicate that despite the wellrecognized limitations, VLS really works and remains the main tool for visualization of vocal fold vibratory phenomenon in clinical settings. Perhaps, it could be partially explained by referring to two basic visual perception phenomena recently showed by Mehta et al,36 which are critical in laryngeal strobo-

751

scopy, that is, the perception of a flicker-free uniformly illuminated image (satisfied at strobe rates above 50 Hz) and the perception of apparent motion from sampled images when no real motion exists (satisfied at display rates above 17 Hz). These two requirements are satisfied in modern VLS systems, which integrate stroboscopic principles with video-based technologies. Correlations of VLS parameters in vocal performance were substantial and statistically significant for the entire functional measurements used in the present study, including instrumental acoustic analysis, VRP, perceptual judgments on voice quality, and calculations of VHI_LT and GFI_LT. These results supports the current consensus that voice is a multidimensional phenomenon and cannot be described by one parameter, but should be investigated by means of voice quality and voice function analyses. Moreover, the different types of information about phonation obtained and accumulated from different sources (visual, voice acoustics, and questionnaires) should be considered as complementing each other. However, multiple significant correlations between VLS parameters and other voice functional measurements may show some overlapping and excess of information. Therefore, further research using evolutionary computing (genetic algorithms) for reduction and selection of parameters and integrated search of clinically most relevant features is advocated. This would allow elaborating reduced and clinically feasible battery of tests to contemplate the pressure to perform a fast and secure evaluation. Future elaboration of automated methods of aggregation and analysis of these miscellaneous types of information would create the necessary background for the development of automated decision support systems for diagnosis of voice and laryngeal disorders.37 CONCLUSION Correlation analysis of the vibratory pattern of the vocal folds obtained via VLS provides more comprehensive insight into the pathophysiology of phonation and suggests the documented and measurable evidence of complex mechanisms of vocal outcome. REFERENCES 1. Dejonckere PH, Bradley P, Clemente P, et al. 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. 2001;258:77–82. 2. Verdonck-de Leeuw IM, Hilgers FJ, Keus RB, et al. Multidimensional assessment of voice characteristics after radiotherapy for early glottic cancer. Laryngoscope. 1999;109(2 pt 1):241–248. 3. Hirano M, Bless DM. Videostroboscopic Examination of the Larynx. San Diego, CA: Singular Publishing Group; 1993:249. 4. Sataloff RT. Professional Voice: The Science and Art of Clinical Care. 3rd ed. San Diego, CA: Plural Publishing, Inc.; 2005:425–446. 5. Woo P. Stroboscopy. San Diego, CA: Plural Publishing, Inc.; 2010:350. 6. Mehta DD, Hillman RE. Current role of stroboscopy in laryngeal imaging. Curr Opin Otolaryngol Head Neck Surg. 2012;20:429–436. 7. Paul BC, Chen S, Sridharan S, Fang Y, Amin MR, Branski RC. Diagnostic accuracy of history, laryngoscopy, and stroboscopy. Laryngoscope. 2013; 123:215–219.

752 8. Dejonckere PH, Crevier L, Elbaz E, Marraco M, Millet B, Remacle M, Woisard V. Quantitative rating of video-laryngostroboscopy: a reliability study. Rev Laryngol Otol Rhinol (Bord). 1998;119:259–260. 9. Poburka BJ. A new stroboscopy rating form. J Voice. 1999;13:403–413. 10. Rosen CA. Stroboscopy as a research instrument: development of a perceptual evaluation tool. Laryngoscope. 2005;115:423–428. 11. Kelley RT, Colton RH, Casper J, Paseman A, Brewer D. Evaluation of stroboscopic signs. J Voice. 2011;25:490–495. 12. Nawka T, Konerding U. The interrater reliability of stroboscopy evaluations. J Voice. 2012;26:812.e1–812.e10. 13. Uloza V, Vegiene A, Pribuisiene R, Saferis V. Quantitative evaluation of video laryngostroboscopy: reliability of the basic parameters. J Voice. 2013;27:361–368. 14. Shiffman S, Reinolds MI, Young FW. Introduction to Multidimensional Scaling: Theory, Methods and Applications. New York, NY: Academic Press; 1981:227–236. 15. Schutte HK, Seidner W. Recommendation by the Union of European Phoniatricians (UEP): standardizing voice area measurement/phonetography. Folia Phoniatr (Basel). 1983;35:286–288. 16. Wuyts FL, De Bodt MS, Molenberghs G, et al. The dysphonia severity index: an objective measure of vocal quality based on a multiparameter approach. J Speech Lang Hear Res. 2000;43:796–809. 17. Dejonckere PH, Obbens C, de Moor GM, Wieneke GH. Perceptual evaluation of dysphonia: reliability and relevance. Folia Phoniatr (Basel). 1993; 45:76–83. 18. Uloza V, Pribuisien_e R. Voice Handicap Index (VHI) in clinical practice [Article in Lithuanian]. Lietuvos Bendrosios Praktikos Gydytojas. 2003; 7:91–95. 19. Pribuisien_e R, Baceviciene M, Uloza V, Vegiene A, Antuseva J. Validation of the Lithuanian version of the Glottal Function Index. J Voice. 2012;26: e73–e78. 20. Mehta DD, Hillman RE. Voice assessment: updates on perceptual, acoustic, aerodynamic, and endoscopic imaging methods. Curr Opin Otolaryngol Head Neck Surg. 2008;16:211–215. 21. Yumoto E. Aerodynamics, voice quality, and laryngeal image analysis of normal and pathologic voices. Curr Opin Otolaryngol Head Neck Surg. 2004;12:166–173. 22. Krausert CR, Olszewski AE, Taylor LN, McMurray JS, Dailey SH, Jiang JJ. Mucosal wave measurement and visualization techniques. J Voice. 2011; 25:395–405. 23. Verdonck-de Leeuw IM, Festen JM, Mahieu HF. Deviant vocal fold vibration as observed during videokymography: the effect on voice quality. J Voice. 2001;15:313–322. 24. Bonilha HS, Deliyski DD. Period and glottal width irregularities in vocally normal speakers. J Voice. 2008;22:699–708.

Journal of Voice, Vol. 27, No. 6, 2013 25. Jeannon JP, Carding PN, Wilson JA. Vocim analysis of laryngeal images: is breathiness related to the glottic area? Clin Otolaryngol Allied Sci. 1998;23: 351–353. 26. Omori K, Kacher A, Slavit D, Blaugrund S. Quantitative videostroboscopic measurement of glottal gap and vocal function: an analysis of thyroplasty type I. Ann Otol Rhinol Laryngol. 1996;105:280–285. 27. Rihkanen H, Reijonen P, Lehikoinen-S€oderlund S, Lauri ER. Videostroboscopic assessment of unilateral vocal fold paralysis after augmentation with autologous fascia. Eur Arch Otorhinolaryngol. 2004;261: 177–183. 28. Bonilha HS, Deliyski DD, Whiteside JP, Gerlach TT. Vocal fold phase asymmetries in patients with voice disorders: a study across visualization techniques. Am J Speech Lang Pathol. 2012;21:3–15. 29. Schneider B, Bigenzahn W. Vocal risk factors for occupational voice disorders in female teaching students. Eur Arch Otorhinolaryngol. 2005;262: 272–276. 30. Lau DP, Zhang EZ, Wong SM, Lee G, Chan YH. Correlating voice handicap index and quantitative videostroboscopy following injection laryngoplasty for unilateral vocal paralysis. Otolaryngol Head Neck Surg. 2010; 143:190–197. 31. Mehta DD, Zeitels SM, Burns JA, Friedman AD, Deliyski DD, Hillman RE. High-speed videoendoscopic analysis of relationships between cepstralbased acoustic measures and voice production mechanisms in patients undergoing phonomicrosurgery. Ann Otol Rhinol Laryngol. 2012;121: 341–347. 32. Kendall KA. High-speed laryngeal imaging compared with videostroboscopy in healthy subjects. Arch Otolaryngol Head Neck Surg. 2009;135: 274–281. 33. Patel R, Dailey S, Bless D. Comparison of high-speed digital imaging with stroboscopy for laryngeal imaging of glottal disorders. Ann Otol Rhinol Laryngol. 2008;117:413–424. 34. Mehta DD, Zaeartu M, Quatieri TF, Deliyski DD, Hillman RE. Investigating acoustic correlates of human vocal fold vibratory phase asymmetry through modeling and laryngeal high-speed videoendoscopy. J Acoust Soc Am. 2011;130:3999–4009. 35. Mehta DD, Deliyski DD, Zeitels SM, Quatieri TF, Hillman RE. Voice production mechanisms following phonosurgical treatment of early glottic cancer. Ann Otol Rhinol Laryngol. 2010;119:1–9. 36. Mehta DD, Deliyski DD, Hillman RE. Commentary on why laryngeal stroboscopy really works: clarifying misconceptions surrounding Talbot’s law and the persistence of vision. J Speech Lang Hear Res. 2010;53: 1263–1267. 37. Verikas A, Gelzinis A, Bacauskiene M, Hallanderb M, Uloza V, Kaseta M. Combining image, voice, and the patient’s questionnaire data to categorize laryngeal disorders. Artif Intell Med. 2010;49:43–50.