- Email: [email protected]
An Ultrasound Investigation of Tongue Shape in Stroke Patients with Lingual Hemiparalysis
An Ultrasound Investigation of Tongue Shape in Stroke Patients with Lingual Hemiparalysis Tim Bressmann, PhD,*† Sina Koch, MSc,‡ Amanda Ratner, MHSc,* Joanne Seigel, MSc,* and Ferdinand Binkofski, MD‡
Background: Stroke can cause hemilateral paresis of the tongue. The present study investigated the functional consequences of a lingual hemiparalysis on the symmetry and the grooving of the tongue in the coronal plane during the production of vowel-consonant-vowel sequences. The hypotheses were that, because of the lingual hemiparalysis, the stroke patients’ tongue shapes would be (1) more asymmetrical and (2) less grooved than the tongues of the control speakers. Methods: The participants in this prospective data collection were 9 stroke patients with lingual hemiparalysis and 6 control speakers. All participants produced vowelconsonant-vowel sequences with the vowels [a, i, and u] and the target consonants [k, t, !, s, and r]. The tongue shape in the coronal plane was traced and measured. The outcome measures were asymmetry and midlingual concavity. The participants and controls were compared using repeated measures analyses of variance with post hoc Scheffe tests. Results: There were no significant differences in asymmetry. There was significantly reduced midlingual concavity for the stroke patients (F[1, 13] 5 8.78; P , .05). There was also a within-subjects effect for consonant (F[4, 50] 5 14.26; P , .01). Post hoc testing with Scheffe tests indicated that the consonant [k] had significantly lower grooving than the other consonant sounds (P , .05). Conclusions: The hemilateral paresis affected not the symmetry but the midlingual grooving. Residual ipsilateral innervation in the hemiparalyzed tongue may help patients compensate. More research is needed to assess the impact of the intrinsic deformation of the tongue on speech acceptability and intelligibility in patients with a lingual hemiparalysis. Key Words: Stroke—tongue—ultrasound— speech—hemiparalysis. Ó 2015 by National Stroke Association
Introduction
From the *Department of Speech-Language Pathology, St. John’s Rehabilitation Program, Sunnybrook Health Sciences Centre; †Department of Speech-Language Pathology, University of Toronto, Toronto, Ontario, Canada; and ‡Department of Neurology, Rheinisch-Westf€ alische Technische Hochschule Aachen, Aachen, Germany. Received August 26, 2014; revision received September 29, 2014; accepted November 23, 2014. Address correspondence to Tim Bressmann, PhD, Department of Speech-Language Pathology, University of Toronto, 160-500 University Avenue, Toronto, ON M5G 1V7, Canada. E-mail: tim. [email protected]. 1052-3057/$ - see front matter Ó 2015 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.11.027
834
This study addressed the impact of a lingual hemiparalysis on coronal tongue shape in speech in stroke patients. The 3 most common neurologic deficits in the area of speech-language pathology described in the literature after first-ever ischemic stroke in the acute stage are dysphagia in more than 55% of patients,1 dysarthria in between 25%2 and 42%3 of patients, and aphasia in between 25%2 and 35%4 of patients. Umapathi et al5 examined 300 patients with acute unilateral ischemic stroke and found that 29% of the stroke victims and 5% of their control participants showed a deviation of the tongue to the weak side. The authors also described associations of tongue deviation with brachiofacial paresis, dysarthria, and dysphagia.
Journal of Stroke and Cerebrovascular Diseases, Vol. 24, No. 4 (April), 2015: pp 834-839
TONGUE SHAPE IN STROKE PATIENTS
Unilateral upper motor neuron damage caused by a stroke can cause paralysis, weakness, or incoordination of the speech musculature, which may result in dysarthria.6–10 Arguably, the most important speech articulator is the tongue, which receives its motor supply from the hypoglossus, the 12th cranial nerve.11 All extrinsic muscles (genioglossus, hyoglossus, and styloglossus muscles) and intrinsic muscles (transverse, vertical, and superior and inferior longitudinal muscles) of the tongue are innervated by the hypoglossal nerve. Speech movements of the tongue are controlled by the primary motor cortex in the precentral gyrus.11,12 Patients with a unilateral supranuclear lesion of the hypoglossal nerve may show weakness or paralysis of the contralateral side of the tongue and a deviation of the protruded tongue to the contralateral side of the lesion.13 The deviation of the tongue on protrusion is considered an important clinical sign.10,14 Duffy and Folger15 investigated speech characteristics in patients with single upper motor neuron lesion and found that the most common characteristic was an imprecise production of consonants. Urban et al9 described slow articulatory movements and reduced speech rate as well as a reduced modulation of pitch and intensity. The severity of the speech characteristics can vary from patient to patient. Dysarthria due to a monohemispheric stroke is often transient and mild to moderate.9,16,17 Muellbacher et al16,17 have argued that the decussation of the hypoglossal fibers in the pyramidal tract is incomplete and that uncrossed corticolingual pathways explain the often mild clinical course of the lingual hemiparalysis after a unilateral upper motor neuron lesion. Lingual hemiparalysis is usually evaluated with a tongue protrusion and lateralization task during the neurologic examination. However, it is not known whether and how the lingual hemiparalysis may affect the symmetry and intrinsic deformation of the tongue during speech. It is difficult to inspect these features of tongue movement during speech because the tongue is concealed in the oral cavity. Ultrasound imaging of the tongue allows a noninvasive assessment of tongue movement.18 A particular advantage is the ability to image the coronal plane of the tongue.19,20 In the present study, coronal tongue shapes of stroke patients with lingual hemiparalysis were compared with healthy control speakers. Because there were no previous reports on this topic, the hypotheses and outcome measures were informed by previous research about the effects of lateral lingual resections in tongue cancer patients.21,22 The hypotheses were that, because of the lingual hemiparalysis, the stroke patients’ tongue shapes would be (1) more asymmetrical and (2) less grooved than the tongues of the control speakers.
835
Method Participants Nine clinical participants (5 women and 4 men; mean age, 68 years; age range, 49-83 years) with a monohemispheric cerebrovascular accident as the primary disease were recruited from the Inpatient and Ambulatory Care Rehabilitation Department at St. John’s Rehabilitation Hospital of the Sunnybrook Health Sciences Centre in Toronto, Canada. All clinical participants were in an early chronic stage in terms of their disease progression. The clinical participants were identified as candidates for this study by a speech-language pathologist. The inclusion criterion for the clinical participants was a lingual hemiparalysis stemming from a stroke. Exclusion criteria for this study were difficulties in attention, vigilance, and cognitive and/or motor abilities. Data recordings took place after a regular appointment with a speechlanguage pathologist. Control speakers (3 females and 3 males; mean age, 25 years; age range, 23-35 years) were recruited through the Department of Speech-Language Pathology, University of Toronto, Faculty of Medicine. All control speakers had normal speech and hearing. They spoke fluent English with the standard Canadian English accent that is common to Southern Ontario. All research procedures were approved by the Research Ethics Review Boards of the St. John’s Rehabilitation Hospital of the Sunnybrook Health Sciences Centre and the University of Toronto.
Materials and Recording Procedure Recordings for the clinical participants took place at the St. John’s Rehabilitation Hospital of the Sunnybrook Health Sciences Centre, Toronto. Recordings for the control speakers took place at the University of Toronto. The procedures were the same in both locations. The participants were seated on a chair with their forehead against a headrest fashioned from a drum practice pad (RF-12G; HQ Percussion, Farmingdale, NY) mounted on a cymbal stand (Pearl BC 1030; Pearl Cooperation, Nashville, TN). Their chin was resting on an ultrasound transducer, which was stabilized with a lockable holder (Manfrotto Magic Arm, Manfrotto, Cassola, Italy). A portable B-mode ultrasound transducer with a center frequency of 5 MHz (Interson USB Probe; National Ultrasound, Pleasanton, CA) with a 90 microconvex array and a netbook (Acer Aspire One; Acer Canada, Mississauga, ON) running the Interson SeeMore imaging software (National Ultrasound, Pleasanton, CA) was used to record tongue movements of all participants in the coronal plane of the tongue. The frame rate of this system was 15 frames per second (fps). The transducer was placed in a central scanning position to get an image of the middle dorsum of the tongue in the coronal plane as displayed in Figure 1.
T. BRESSMANN ET AL.
836
Data Preparation
Figure 1. Ultrasound image of the middle dorsum of the tongue in the coronal plane for consonant /s/.
The video output from the ultrasound assessment was captured from the computer’s video graphics array output using a Sabrent converter TV-PC85 (Sabrent, Chatsworth, CA), which converted the signal to a National Television Systems Committee RS-170A signal. The video stream was recorded with a digital video disc recorder (Sony DVDirect VRD-MC6; Sony Canada, Toronto, ON). Although the ultrasound transducer produced 15 fps, the movie on the video disc had a frame rate of 30 fps, which means that the ultrasound images were doubled. The acoustic signal was recorded to the same digital video disc using a microphone (Apex, 435; APEX Electronics, Pickering, ON) phantom-powered by an ART Tube MP preamplifier (ART Proaudio, Pickering, ON). The participants in this study produced a series of vowel-consonant-vowel (VCV) sequences. All VCV sequences were repeated 3 times. Five target consonants in 3 different vowel contexts were chosen for this task, resulting in a total number of 45 individual VCV sequences that were produced by each participant. The target sounds for consonants were the velar stop [k], the alveolar stop [t], the sibilants [s] and [!] (the ‘‘sh’’ sound), and the approximant [ɹ] (North American ‘‘r’’). These consonants were chosen because they demonstrate different features of tongue motility, such as midlingual groove in [s] and [!], tongue dorsum elevation in [k], tongue tip elevation in [t], and retroflexion and midlingual groove in [ɹ]. Also, the consonant targets represented the alveolar, postalveolar, and velar articulation zones. The vowel contexts were [a], [i], and [u]. These vowels were chosen because they represent the extreme lingual positions for vowel production in the English language.22 The stimuli were presented visually on a paper printout held at eye level for the participant. They were also modeled by the examiner for the participant to repeat.
The ultrasound movies of the tongue movement were copied from the digital video discs to a computer and converted to Windows Media Video format. For the segmentation of the movies into single VCV sequences, the Sony Screenblast 3.0 software (Sony Canada, Toronto, ON) was used. Because the video stream lagged behind the audio, the duration of the lag was established based on the release of plosive sounds in the video and the audio channels. The event in the video image occurred 8 frames after the corresponding audio event (note that this corresponded to 4 unique ultrasound images because of the image doubling of the 15 fps ultrasound in the 30 fps movie). The delays for patients and control speakers were confirmed anew in every movie before further editing and analysis. The surface of the tongue in selected frames from the ultrasound movie was traced by the first author by hand using a graphics tablet (Bamboo; Wacom Technology Corporation, Vancouver, WA). The centers of the initial and final vowels in each VCV sequence were determined based on the sound oscillogram. The end point of the tongue excursion for the consonant sound was based on visual inspection and the acoustic landmarks in the sound oscillogram, such as plosive releases or noise segments of fricatives. The hand-drawn lines were measured using the Ultrasonographic Contour Analyzer for Tongue Surfaces software (University of Toronto, Toronto, ON).19 The resulting data described the distance between the anchor point and the tongue surface in millimeters along radiating gridlines in 5 intervals. The measurements were saved for descriptive and inferential statistical analyses.
Data Analysis All data for this study were processed with Microsoft Excel 2007 (Microsoft Canada, Mississauga, ON). In a first step, the intrarater reliability was calculated to verify the accuracy of the tracing of the tongue surface. A subset of 5% of the data was remeasured by the first author. Averages across 3 repetitions of each target consonant in each vowel context were calculated. In case of missing data in a measurement, the average was calculated from the remaining 2 repetitions. To facilitate comparison of cases, the data from the paralyzed side of the tongue were always assigned to the right side. Therefore, the data from the speakers with a paralysis on the left side of the tongue were flipped in the spreadsheet. The tongue shape in the coronal plane was evaluated using quantitative measures for asymmetry and concavity, based on Bressmann et al.21,22 To calculate these measures, the tongue height at the center of the tongue and at 2 grid angles 10 left and 10 right of the center were used. Asymmetry was calculated by subtracting the tongue height on the paralyzed side (right) at 10 from the nonparalyzed (left) side at 10 . For the control
TONGUE SHAPE IN STROKE PATIENTS
837
Table 1. Cumulative percentages for direct agreement of the ultrasound measurements between 10 left and 10 right and mean differences in mm Measurement differences
210 left
25 left
0 center
5 right
10 right
57.6 87.4 95 97 100 1.09 .89
60.7 89.1 96 97.7 100 1.04 1.11
64.7 90.1 97 98.7 100 .93 .89
58.1 84.5 93.7 97.4 100 1.10 .96
62.5 87 97 99 100 .99 1.03
,1 mm (%) ,2 mm (%) ,3 mm (%) ,4 mm (%) $4 mm (%) Mean difference in mm Standard deviation
speakers, the height of the right side of the tongue was subtracted from the height of the left side of the tongue. Concavity of the tongue was calculated by averaging tongue height at the left and right 10 grid angles and subtracting the tongue height at 0 .
Statistical Analysis For the analysis, the Number Cruncher Statistical System 8 (NCSS LLC, Kaysville, UT) was used. Partial missing data were noted for 3 patients because of recording errors. Since these missing data constituted only 3.6% of the overall data set, they were treated with case-wise deletion. Two repeated-measures analyses of variances (ANOVAs) were run to determine the possible differences in asymmetry and concavity between the 2 groups. For the repeated measures–ANOVAs, P values were considered significant if they were lower than .05 without further Bonferroni adjustment.23
Results Intrarater Reliability Table 1 represents the results for test–retest reliability. The results indicated that 80% of the repeated measurements were within a 2 mm range between measurements.
Asymmetry The results for the asymmetry measure are summarized in Table 2. A repeated-measures ANOVA with group (patients and control speakers) as the betweensubject factor and the vowels [/a/, /i/, and /u/] and the consonants [k, t, !, s, and ɹ] as the within-subject factors was not significant.
Concavity The results for the concavity measure are summarized in Table 2. A repeated-measures ANOVA with group (patients and control speakers) as the between-subject factor and the vowels [/a/, /i/, and /u/] and the consonants [k, t, !, s, and ɹ] as the within-subject factors was calculated. A main effect for group was found F(1, 13) 5 8.78, P less than .05. Figure 2 illustrates this finding. There was also a within-subjects effect for consonant F(4, 50) 5 14.26, P less than .01. Post hoc testing with Scheffe tests indicated that the consonant [k] had significantly lower grooving than the other consonant sounds (P , .05). Figure 3 illustrates this finding.
Discussion The present study compared stroke patients to control speakers with regard to their coronal tongue movement
Table 2. Results for asymmetry and concavity for patients and control speakers
Speech sound [a] [i] [u] [k] [t] [!] [s] [ɹ]
Patients mean (SD) asymmetry
Patients mean (SD) concavity
Control speakers mean (SD) asymmetry
Control speakers mean (SD) concavity
1.17 (2.03) .16 (2.67) .97 (2.31) 1.01 (3.07) 1.15 (2.97) 2.07 (2.58) .69 (3.64) 1.05 (2.02)
.85 (1.51) 2.04 (1.70) 2.22 (1.68) 22.54 (1.96) 1.35 (1.47) .41 (1.64) 1.67 (1.58) 1.53 (2.10)
.49 (1.98) 2.05 (2.15) .93 (1.34) 2.32 (1.86) .18 (1.56) 2.10 (1.99) .20 (1.75) 2.37 (2.09)
2.79 (2.00) 3.03 (3.43) 2.03 (2.55) 2.36 (3.32) 3.94 (2.19) 3.10 (2.01) 3.65 (1.67) 2.95 (3.43)
Abbreviation: SD, standard deviation.
T. BRESSMANN ET AL.
838
Figure 2.
Boxplot for concavity of patients and control speakers.
in speech. The results refuted the first hypothesis that the stroke patients’ tongues would be more asymmetrical but confirmed the second hypothesis that the stroke patients’ tongues would be less grooved than the tongues of the control speakers. For the purposes of the present study, the B-mode ultrasound imaging provided sufficient information. The assessment was well tolerated by participants and the setup and recording time was reasonably short. The intrarater reliability was acceptable, and 80% of the repeated measurements were in a 2-mm range difference. This result replicates previous findings in patients with lateral partial glossectomies.24
The results for asymmetry of the tongue were not significantly different between patients and control speakers. In general, the movement of the lateral free margins of the tongue in speech is limited.20 Because the tongue mass does not atrophy in a central paralysis, it is plausible that the paralyzed lateral free margin would occupy the same space in the oral cavity as before. It would be interesting to compare these results to participants with peripheral hemiparalysis of the tongue where one would expect to see increased asymmetry and atrophy. Previous research has demonstrated that a moderate amount of asymmetry of the tongue is normal in speech.25–27 Hamlet et al27 distinguished spatial and temporary asymmetry when using electropalatography to study the production of consonants /s/ and /l/. Because of the frame rate of the ultrasound system used in the present study, only the spatial aspect could be evaluated. A significant difference between patients and control speakers was found regarding concavity of the tongue. Results indicated a lower extent of midlingual grooving for patients compared with control speakers. The muscle thought to be responsible for midlingual grooving is the genioglossus. It can be speculated that either the loss of bilateral innervations of this muscle or instability in the paralyzed lateral free margin of the tongue led to the observed reduction in midlingual grooving. Bressmann et al21 found that patients with a partial resection of the lateral free margin of the tongue also showed a decrease in midlingual grooving. The within-subjects effect for the consonant sound showed a significantly reduced groove in the dorsum of the tongue during the production of [k]. Because the dorsum of the tongue is elevated during the production of this sound, a convex tongue shape is to be expected. The other 4 consonants [t, !, s, and ɹ] have more anterior constriction locations, so the more posterior aspects of the tongue dorsum and root will show grooving. A number of caveats should be kept in mind while considering the results. The control speakers of this study were considerably younger than the patients. It is not known whether younger speakers would show deeper midlingual grooves and if the grooving of the tongue decreases with age. The temporal resolution of the transducer was relatively low. However, the 15 fps scan rate did not affect the quality of the results for the vowels and consonants. In future research, it would be worthwhile to employ a 3D transducer that would allow data acquisition from multiple coronal planes. Finally, the differences between patients and control speakers were relatively small in the present study. This may indicate that there is indeed some residual ipsilateral innervation in the hemiparalyzed tongue that helps the speaker compensate.16,17
Conclusion Figure 3. Boxplot of concavity for different consonant sounds.
The present study provides first information about the impact of lingual hemiparalysis on tongue shape in the
TONGUE SHAPE IN STROKE PATIENTS
coronal plane. The results from the ultrasound data showed no differences between stroke patients and control speakers with regard to symmetry of the lateral free margins of the tongue. However, the stroke patients showed significantly less midlingual grooving. More research is needed to assess the impact of the intrinsic deformation of the tongue on speech acceptability and intelligibility in patients with a lingual hemiparalysis stemming from a stroke. Acknowledgments: This research was funded in part by an operating grant from the St. John’s Rehabilitation Hospital Research Foundation to the authors T.B., A.R., and J.S. We would like to thank the patients and control speakers for participating in our study. Also, we are grateful to Professor Dr. Klaus Willmes for statistical advice.
839
13.
14.
15.
16.
17.
18.
References 1. Martino R, Foley N, Bhogal S, et al. Dysphagia after stroke: incidence, diagnosis, and pulmonary complications. Stroke 2005;36:2756-2763. 2. Lubart E, Leibovitz A, Baumoehl Y, et al. Progressing stroke with neurological deterioration in a group of Israeli elderly. Arch Gerontol Geriatr 2005;41:95-100. 3. Lawrence ES, Coshall C, Dundas R, et al. Estimates of the prevalence of acute stroke impairments and disability in a multiethnic population. Stroke 2001;32:1279-1284. 4. Tsouli S, Kyritsis AP, Tsagalis G, et al. Significance of aphasia after first-ever acute stroke: impact on early and late outcomes. Neuroepidemiology 2009;33:96-102. 5. Umapathi T, Venketasubramanian N, Leck KJ, et al. Tongue deviation in acute ischemic stroke: a study of supranuclear twelfth cranial nerve palsy in 300 stroke patients. Cerebrovasc Dis 2000;10:462-465. 6. Darley FL, Aronson AE, Brown JR. Differential diagnostic patterns of dysarthria. J Speech Hear Res 1969;12:246-269. 7. Darley FL, Aronson AE, Brown JR. Cluster of deviant speech dimensions in the dysarthrias. J Speech Hear Res 1969;12:462-496. 8. Darley FL, Aronson AE, Brown JR. Motor speech disorders. Philadelphia: W.B. Saunders 1975. 9. Urban PP, Hopf HC, Fleischer S, et al. Impaired corticobulbar tract function in dysarthria due to hemispheric stroke. Functional testing using transcranial magnetic stimulation. Brain 1997;120:1077-1084. 10. Duffy JR. Motor speech disorder: substrates, differential diagnosis, and management. St. Louis: Mosby 2013. 11. Loh C, Menahem Maya M, Go JL. Cranial nerve XII: the hypoglossal nerve. Semin Ultrasound CT MR 2002;23: 256-265. 12. Walker HK. Cranial nerve XII: the hypoglossal nerve. In: Walker HK, Hall HD, Hurst JW, eds. Clinical methods:
19.
20.
21.
22.
23. 24.
25.
26.
27.
the history, physical, and laboratory examinations. Boston: Butterworths 1990:331. Schuenke M, Schulte E, Schumacher U. Atlas of anatomy: head and neuroanatomy. Leipzig, Germany: Thieme 2010. Murdoch BE. Acquired speech and language disorders: a neuroanatomical and functional neurological approach. Sussex, UK: John Wiley & Sons 2010. Duffy JR, Folger WN. Dysarthria associated with unilateral central nervous system lesion: a retrospective study. J Med Speech Lang Pathol 1996;4:57-70. Muellbacher W, Artner C, Mamoli B. Motor evoked potentials in unilateral lingual paralysis after monohemispheric ischaemia. J Neurol Neurosurg Psychiatry 1998; 65:755-761. Muellbacher W, Artner C, Mamoli B. The role of the intact hemisphere in recovery of midline muscles after recent monohemispheric stroke. J Neurol Neurosurg Psychiatry 1999;246:250-256. Menard L, Aubin J, Thibeault M, et al. Measuring tongue shapes and positions with ultrasound imaging: a validation experiment using an articulatory model. Folia Phoniatr Logop 2012;64:64-72. Bressmann T. 2D and 3D ultrasound imaging of the tongue in normal and disordered speech. In: Maassen B, Van Lieshout P, eds. Speech motor control: new developments in basic and applied research. New York: Oxford University Press 2010:351-369. Bressmann T, Flowers H, Wong W, et al. Coronal view ultrasound imaging of movement in different segments of the tongue during paced recital: findings from four normal speakers and a speaker with partial glossectomy. Clin Linguist Phon 2010;29: 589-601. Bressmann T, Ackloo E, Heng CL, et al. Quantitative 3D ultrasound imaging of partially resected tongues. Otolaryngol Head Neck Surg 2007;136:799-805. Bressmann T, Thind P, Uy C, et al. Quantitative threedimensional ultrasound analysis of tongue protrusion, grooving and symmetry: data from 12 normal speakers and a partial glossectomee. Clin Linguist Phon 2005;19: 573-588. Perneger TV. What’s wrong with Bonferroni adjustments. BMJ 1998;316:1236-1238. Rastadmehr O, Bressmann T, Smyth R, et al. Increased midsagittal tongue velocity as indication of articulatory compensation in patients with lateral partial glossectomies. Head Neck 2008;30:718-726. Stone M, Shawker T, Talbot L, et al. Cross-sectional tongue shape during the production of vowels. J Acoust Soc Am 1988;83:1586-1596. Stone MA. Three-dimensional model of tongue movement based on ultrasound and x-ray microbeam data. J Acoust Soc Am 1990;87:2207-2217. Hamlet SL, Bunnell HT, Struntz B. Articulatory asymmetries. J Acoust Soc Am 1986;79:1164-1169.
Background: Stroke can cause hemilateral paresis of the tongue. The present study investigated the functional consequences of a lingual hemiparalysis on the symmetry and the grooving of the tongue in the coronal plane during the production of vowel-consonant-vowel sequences. The hypotheses were that, because of the lingual hemiparalysis, the stroke patients’ tongue shapes would be (1) more asymmetrical and (2) less grooved than the tongues of the control speakers. Methods: The participants in this prospective data collection were 9 stroke patients with lingual hemiparalysis and 6 control speakers. All participants produced vowelconsonant-vowel sequences with the vowels [a, i, and u] and the target consonants [k, t, !, s, and r]. The tongue shape in the coronal plane was traced and measured. The outcome measures were asymmetry and midlingual concavity. The participants and controls were compared using repeated measures analyses of variance with post hoc Scheffe tests. Results: There were no significant differences in asymmetry. There was significantly reduced midlingual concavity for the stroke patients (F[1, 13] 5 8.78; P , .05). There was also a within-subjects effect for consonant (F[4, 50] 5 14.26; P , .01). Post hoc testing with Scheffe tests indicated that the consonant [k] had significantly lower grooving than the other consonant sounds (P , .05). Conclusions: The hemilateral paresis affected not the symmetry but the midlingual grooving. Residual ipsilateral innervation in the hemiparalyzed tongue may help patients compensate. More research is needed to assess the impact of the intrinsic deformation of the tongue on speech acceptability and intelligibility in patients with a lingual hemiparalysis. Key Words: Stroke—tongue—ultrasound— speech—hemiparalysis. Ó 2015 by National Stroke Association
Introduction
From the *Department of Speech-Language Pathology, St. John’s Rehabilitation Program, Sunnybrook Health Sciences Centre; †Department of Speech-Language Pathology, University of Toronto, Toronto, Ontario, Canada; and ‡Department of Neurology, Rheinisch-Westf€ alische Technische Hochschule Aachen, Aachen, Germany. Received August 26, 2014; revision received September 29, 2014; accepted November 23, 2014. Address correspondence to Tim Bressmann, PhD, Department of Speech-Language Pathology, University of Toronto, 160-500 University Avenue, Toronto, ON M5G 1V7, Canada. E-mail: tim. [email protected]. 1052-3057/$ - see front matter Ó 2015 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.11.027
834
This study addressed the impact of a lingual hemiparalysis on coronal tongue shape in speech in stroke patients. The 3 most common neurologic deficits in the area of speech-language pathology described in the literature after first-ever ischemic stroke in the acute stage are dysphagia in more than 55% of patients,1 dysarthria in between 25%2 and 42%3 of patients, and aphasia in between 25%2 and 35%4 of patients. Umapathi et al5 examined 300 patients with acute unilateral ischemic stroke and found that 29% of the stroke victims and 5% of their control participants showed a deviation of the tongue to the weak side. The authors also described associations of tongue deviation with brachiofacial paresis, dysarthria, and dysphagia.
Journal of Stroke and Cerebrovascular Diseases, Vol. 24, No. 4 (April), 2015: pp 834-839
TONGUE SHAPE IN STROKE PATIENTS
Unilateral upper motor neuron damage caused by a stroke can cause paralysis, weakness, or incoordination of the speech musculature, which may result in dysarthria.6–10 Arguably, the most important speech articulator is the tongue, which receives its motor supply from the hypoglossus, the 12th cranial nerve.11 All extrinsic muscles (genioglossus, hyoglossus, and styloglossus muscles) and intrinsic muscles (transverse, vertical, and superior and inferior longitudinal muscles) of the tongue are innervated by the hypoglossal nerve. Speech movements of the tongue are controlled by the primary motor cortex in the precentral gyrus.11,12 Patients with a unilateral supranuclear lesion of the hypoglossal nerve may show weakness or paralysis of the contralateral side of the tongue and a deviation of the protruded tongue to the contralateral side of the lesion.13 The deviation of the tongue on protrusion is considered an important clinical sign.10,14 Duffy and Folger15 investigated speech characteristics in patients with single upper motor neuron lesion and found that the most common characteristic was an imprecise production of consonants. Urban et al9 described slow articulatory movements and reduced speech rate as well as a reduced modulation of pitch and intensity. The severity of the speech characteristics can vary from patient to patient. Dysarthria due to a monohemispheric stroke is often transient and mild to moderate.9,16,17 Muellbacher et al16,17 have argued that the decussation of the hypoglossal fibers in the pyramidal tract is incomplete and that uncrossed corticolingual pathways explain the often mild clinical course of the lingual hemiparalysis after a unilateral upper motor neuron lesion. Lingual hemiparalysis is usually evaluated with a tongue protrusion and lateralization task during the neurologic examination. However, it is not known whether and how the lingual hemiparalysis may affect the symmetry and intrinsic deformation of the tongue during speech. It is difficult to inspect these features of tongue movement during speech because the tongue is concealed in the oral cavity. Ultrasound imaging of the tongue allows a noninvasive assessment of tongue movement.18 A particular advantage is the ability to image the coronal plane of the tongue.19,20 In the present study, coronal tongue shapes of stroke patients with lingual hemiparalysis were compared with healthy control speakers. Because there were no previous reports on this topic, the hypotheses and outcome measures were informed by previous research about the effects of lateral lingual resections in tongue cancer patients.21,22 The hypotheses were that, because of the lingual hemiparalysis, the stroke patients’ tongue shapes would be (1) more asymmetrical and (2) less grooved than the tongues of the control speakers.
835
Method Participants Nine clinical participants (5 women and 4 men; mean age, 68 years; age range, 49-83 years) with a monohemispheric cerebrovascular accident as the primary disease were recruited from the Inpatient and Ambulatory Care Rehabilitation Department at St. John’s Rehabilitation Hospital of the Sunnybrook Health Sciences Centre in Toronto, Canada. All clinical participants were in an early chronic stage in terms of their disease progression. The clinical participants were identified as candidates for this study by a speech-language pathologist. The inclusion criterion for the clinical participants was a lingual hemiparalysis stemming from a stroke. Exclusion criteria for this study were difficulties in attention, vigilance, and cognitive and/or motor abilities. Data recordings took place after a regular appointment with a speechlanguage pathologist. Control speakers (3 females and 3 males; mean age, 25 years; age range, 23-35 years) were recruited through the Department of Speech-Language Pathology, University of Toronto, Faculty of Medicine. All control speakers had normal speech and hearing. They spoke fluent English with the standard Canadian English accent that is common to Southern Ontario. All research procedures were approved by the Research Ethics Review Boards of the St. John’s Rehabilitation Hospital of the Sunnybrook Health Sciences Centre and the University of Toronto.
Materials and Recording Procedure Recordings for the clinical participants took place at the St. John’s Rehabilitation Hospital of the Sunnybrook Health Sciences Centre, Toronto. Recordings for the control speakers took place at the University of Toronto. The procedures were the same in both locations. The participants were seated on a chair with their forehead against a headrest fashioned from a drum practice pad (RF-12G; HQ Percussion, Farmingdale, NY) mounted on a cymbal stand (Pearl BC 1030; Pearl Cooperation, Nashville, TN). Their chin was resting on an ultrasound transducer, which was stabilized with a lockable holder (Manfrotto Magic Arm, Manfrotto, Cassola, Italy). A portable B-mode ultrasound transducer with a center frequency of 5 MHz (Interson USB Probe; National Ultrasound, Pleasanton, CA) with a 90 microconvex array and a netbook (Acer Aspire One; Acer Canada, Mississauga, ON) running the Interson SeeMore imaging software (National Ultrasound, Pleasanton, CA) was used to record tongue movements of all participants in the coronal plane of the tongue. The frame rate of this system was 15 frames per second (fps). The transducer was placed in a central scanning position to get an image of the middle dorsum of the tongue in the coronal plane as displayed in Figure 1.
T. BRESSMANN ET AL.
836
Data Preparation
Figure 1. Ultrasound image of the middle dorsum of the tongue in the coronal plane for consonant /s/.
The video output from the ultrasound assessment was captured from the computer’s video graphics array output using a Sabrent converter TV-PC85 (Sabrent, Chatsworth, CA), which converted the signal to a National Television Systems Committee RS-170A signal. The video stream was recorded with a digital video disc recorder (Sony DVDirect VRD-MC6; Sony Canada, Toronto, ON). Although the ultrasound transducer produced 15 fps, the movie on the video disc had a frame rate of 30 fps, which means that the ultrasound images were doubled. The acoustic signal was recorded to the same digital video disc using a microphone (Apex, 435; APEX Electronics, Pickering, ON) phantom-powered by an ART Tube MP preamplifier (ART Proaudio, Pickering, ON). The participants in this study produced a series of vowel-consonant-vowel (VCV) sequences. All VCV sequences were repeated 3 times. Five target consonants in 3 different vowel contexts were chosen for this task, resulting in a total number of 45 individual VCV sequences that were produced by each participant. The target sounds for consonants were the velar stop [k], the alveolar stop [t], the sibilants [s] and [!] (the ‘‘sh’’ sound), and the approximant [ɹ] (North American ‘‘r’’). These consonants were chosen because they demonstrate different features of tongue motility, such as midlingual groove in [s] and [!], tongue dorsum elevation in [k], tongue tip elevation in [t], and retroflexion and midlingual groove in [ɹ]. Also, the consonant targets represented the alveolar, postalveolar, and velar articulation zones. The vowel contexts were [a], [i], and [u]. These vowels were chosen because they represent the extreme lingual positions for vowel production in the English language.22 The stimuli were presented visually on a paper printout held at eye level for the participant. They were also modeled by the examiner for the participant to repeat.
The ultrasound movies of the tongue movement were copied from the digital video discs to a computer and converted to Windows Media Video format. For the segmentation of the movies into single VCV sequences, the Sony Screenblast 3.0 software (Sony Canada, Toronto, ON) was used. Because the video stream lagged behind the audio, the duration of the lag was established based on the release of plosive sounds in the video and the audio channels. The event in the video image occurred 8 frames after the corresponding audio event (note that this corresponded to 4 unique ultrasound images because of the image doubling of the 15 fps ultrasound in the 30 fps movie). The delays for patients and control speakers were confirmed anew in every movie before further editing and analysis. The surface of the tongue in selected frames from the ultrasound movie was traced by the first author by hand using a graphics tablet (Bamboo; Wacom Technology Corporation, Vancouver, WA). The centers of the initial and final vowels in each VCV sequence were determined based on the sound oscillogram. The end point of the tongue excursion for the consonant sound was based on visual inspection and the acoustic landmarks in the sound oscillogram, such as plosive releases or noise segments of fricatives. The hand-drawn lines were measured using the Ultrasonographic Contour Analyzer for Tongue Surfaces software (University of Toronto, Toronto, ON).19 The resulting data described the distance between the anchor point and the tongue surface in millimeters along radiating gridlines in 5 intervals. The measurements were saved for descriptive and inferential statistical analyses.
Data Analysis All data for this study were processed with Microsoft Excel 2007 (Microsoft Canada, Mississauga, ON). In a first step, the intrarater reliability was calculated to verify the accuracy of the tracing of the tongue surface. A subset of 5% of the data was remeasured by the first author. Averages across 3 repetitions of each target consonant in each vowel context were calculated. In case of missing data in a measurement, the average was calculated from the remaining 2 repetitions. To facilitate comparison of cases, the data from the paralyzed side of the tongue were always assigned to the right side. Therefore, the data from the speakers with a paralysis on the left side of the tongue were flipped in the spreadsheet. The tongue shape in the coronal plane was evaluated using quantitative measures for asymmetry and concavity, based on Bressmann et al.21,22 To calculate these measures, the tongue height at the center of the tongue and at 2 grid angles 10 left and 10 right of the center were used. Asymmetry was calculated by subtracting the tongue height on the paralyzed side (right) at 10 from the nonparalyzed (left) side at 10 . For the control
TONGUE SHAPE IN STROKE PATIENTS
837
Table 1. Cumulative percentages for direct agreement of the ultrasound measurements between 10 left and 10 right and mean differences in mm Measurement differences
210 left
25 left
0 center
5 right
10 right
57.6 87.4 95 97 100 1.09 .89
60.7 89.1 96 97.7 100 1.04 1.11
64.7 90.1 97 98.7 100 .93 .89
58.1 84.5 93.7 97.4 100 1.10 .96
62.5 87 97 99 100 .99 1.03
,1 mm (%) ,2 mm (%) ,3 mm (%) ,4 mm (%) $4 mm (%) Mean difference in mm Standard deviation
speakers, the height of the right side of the tongue was subtracted from the height of the left side of the tongue. Concavity of the tongue was calculated by averaging tongue height at the left and right 10 grid angles and subtracting the tongue height at 0 .
Statistical Analysis For the analysis, the Number Cruncher Statistical System 8 (NCSS LLC, Kaysville, UT) was used. Partial missing data were noted for 3 patients because of recording errors. Since these missing data constituted only 3.6% of the overall data set, they were treated with case-wise deletion. Two repeated-measures analyses of variances (ANOVAs) were run to determine the possible differences in asymmetry and concavity between the 2 groups. For the repeated measures–ANOVAs, P values were considered significant if they were lower than .05 without further Bonferroni adjustment.23
Results Intrarater Reliability Table 1 represents the results for test–retest reliability. The results indicated that 80% of the repeated measurements were within a 2 mm range between measurements.
Asymmetry The results for the asymmetry measure are summarized in Table 2. A repeated-measures ANOVA with group (patients and control speakers) as the betweensubject factor and the vowels [/a/, /i/, and /u/] and the consonants [k, t, !, s, and ɹ] as the within-subject factors was not significant.
Concavity The results for the concavity measure are summarized in Table 2. A repeated-measures ANOVA with group (patients and control speakers) as the between-subject factor and the vowels [/a/, /i/, and /u/] and the consonants [k, t, !, s, and ɹ] as the within-subject factors was calculated. A main effect for group was found F(1, 13) 5 8.78, P less than .05. Figure 2 illustrates this finding. There was also a within-subjects effect for consonant F(4, 50) 5 14.26, P less than .01. Post hoc testing with Scheffe tests indicated that the consonant [k] had significantly lower grooving than the other consonant sounds (P , .05). Figure 3 illustrates this finding.
Discussion The present study compared stroke patients to control speakers with regard to their coronal tongue movement
Table 2. Results for asymmetry and concavity for patients and control speakers
Speech sound [a] [i] [u] [k] [t] [!] [s] [ɹ]
Patients mean (SD) asymmetry
Patients mean (SD) concavity
Control speakers mean (SD) asymmetry
Control speakers mean (SD) concavity
1.17 (2.03) .16 (2.67) .97 (2.31) 1.01 (3.07) 1.15 (2.97) 2.07 (2.58) .69 (3.64) 1.05 (2.02)
.85 (1.51) 2.04 (1.70) 2.22 (1.68) 22.54 (1.96) 1.35 (1.47) .41 (1.64) 1.67 (1.58) 1.53 (2.10)
.49 (1.98) 2.05 (2.15) .93 (1.34) 2.32 (1.86) .18 (1.56) 2.10 (1.99) .20 (1.75) 2.37 (2.09)
2.79 (2.00) 3.03 (3.43) 2.03 (2.55) 2.36 (3.32) 3.94 (2.19) 3.10 (2.01) 3.65 (1.67) 2.95 (3.43)
Abbreviation: SD, standard deviation.
T. BRESSMANN ET AL.
838
Figure 2.
Boxplot for concavity of patients and control speakers.
in speech. The results refuted the first hypothesis that the stroke patients’ tongues would be more asymmetrical but confirmed the second hypothesis that the stroke patients’ tongues would be less grooved than the tongues of the control speakers. For the purposes of the present study, the B-mode ultrasound imaging provided sufficient information. The assessment was well tolerated by participants and the setup and recording time was reasonably short. The intrarater reliability was acceptable, and 80% of the repeated measurements were in a 2-mm range difference. This result replicates previous findings in patients with lateral partial glossectomies.24
The results for asymmetry of the tongue were not significantly different between patients and control speakers. In general, the movement of the lateral free margins of the tongue in speech is limited.20 Because the tongue mass does not atrophy in a central paralysis, it is plausible that the paralyzed lateral free margin would occupy the same space in the oral cavity as before. It would be interesting to compare these results to participants with peripheral hemiparalysis of the tongue where one would expect to see increased asymmetry and atrophy. Previous research has demonstrated that a moderate amount of asymmetry of the tongue is normal in speech.25–27 Hamlet et al27 distinguished spatial and temporary asymmetry when using electropalatography to study the production of consonants /s/ and /l/. Because of the frame rate of the ultrasound system used in the present study, only the spatial aspect could be evaluated. A significant difference between patients and control speakers was found regarding concavity of the tongue. Results indicated a lower extent of midlingual grooving for patients compared with control speakers. The muscle thought to be responsible for midlingual grooving is the genioglossus. It can be speculated that either the loss of bilateral innervations of this muscle or instability in the paralyzed lateral free margin of the tongue led to the observed reduction in midlingual grooving. Bressmann et al21 found that patients with a partial resection of the lateral free margin of the tongue also showed a decrease in midlingual grooving. The within-subjects effect for the consonant sound showed a significantly reduced groove in the dorsum of the tongue during the production of [k]. Because the dorsum of the tongue is elevated during the production of this sound, a convex tongue shape is to be expected. The other 4 consonants [t, !, s, and ɹ] have more anterior constriction locations, so the more posterior aspects of the tongue dorsum and root will show grooving. A number of caveats should be kept in mind while considering the results. The control speakers of this study were considerably younger than the patients. It is not known whether younger speakers would show deeper midlingual grooves and if the grooving of the tongue decreases with age. The temporal resolution of the transducer was relatively low. However, the 15 fps scan rate did not affect the quality of the results for the vowels and consonants. In future research, it would be worthwhile to employ a 3D transducer that would allow data acquisition from multiple coronal planes. Finally, the differences between patients and control speakers were relatively small in the present study. This may indicate that there is indeed some residual ipsilateral innervation in the hemiparalyzed tongue that helps the speaker compensate.16,17
Conclusion Figure 3. Boxplot of concavity for different consonant sounds.
The present study provides first information about the impact of lingual hemiparalysis on tongue shape in the
TONGUE SHAPE IN STROKE PATIENTS
coronal plane. The results from the ultrasound data showed no differences between stroke patients and control speakers with regard to symmetry of the lateral free margins of the tongue. However, the stroke patients showed significantly less midlingual grooving. More research is needed to assess the impact of the intrinsic deformation of the tongue on speech acceptability and intelligibility in patients with a lingual hemiparalysis stemming from a stroke. Acknowledgments: This research was funded in part by an operating grant from the St. John’s Rehabilitation Hospital Research Foundation to the authors T.B., A.R., and J.S. We would like to thank the patients and control speakers for participating in our study. Also, we are grateful to Professor Dr. Klaus Willmes for statistical advice.
839
13.
14.
15.
16.
17.
18.
References 1. Martino R, Foley N, Bhogal S, et al. Dysphagia after stroke: incidence, diagnosis, and pulmonary complications. Stroke 2005;36:2756-2763. 2. Lubart E, Leibovitz A, Baumoehl Y, et al. Progressing stroke with neurological deterioration in a group of Israeli elderly. Arch Gerontol Geriatr 2005;41:95-100. 3. Lawrence ES, Coshall C, Dundas R, et al. Estimates of the prevalence of acute stroke impairments and disability in a multiethnic population. Stroke 2001;32:1279-1284. 4. Tsouli S, Kyritsis AP, Tsagalis G, et al. Significance of aphasia after first-ever acute stroke: impact on early and late outcomes. Neuroepidemiology 2009;33:96-102. 5. Umapathi T, Venketasubramanian N, Leck KJ, et al. Tongue deviation in acute ischemic stroke: a study of supranuclear twelfth cranial nerve palsy in 300 stroke patients. Cerebrovasc Dis 2000;10:462-465. 6. Darley FL, Aronson AE, Brown JR. Differential diagnostic patterns of dysarthria. J Speech Hear Res 1969;12:246-269. 7. Darley FL, Aronson AE, Brown JR. Cluster of deviant speech dimensions in the dysarthrias. J Speech Hear Res 1969;12:462-496. 8. Darley FL, Aronson AE, Brown JR. Motor speech disorders. Philadelphia: W.B. Saunders 1975. 9. Urban PP, Hopf HC, Fleischer S, et al. Impaired corticobulbar tract function in dysarthria due to hemispheric stroke. Functional testing using transcranial magnetic stimulation. Brain 1997;120:1077-1084. 10. Duffy JR. Motor speech disorder: substrates, differential diagnosis, and management. St. Louis: Mosby 2013. 11. Loh C, Menahem Maya M, Go JL. Cranial nerve XII: the hypoglossal nerve. Semin Ultrasound CT MR 2002;23: 256-265. 12. Walker HK. Cranial nerve XII: the hypoglossal nerve. In: Walker HK, Hall HD, Hurst JW, eds. Clinical methods:
19.
20.
21.
22.
23. 24.
25.
26.
27.
the history, physical, and laboratory examinations. Boston: Butterworths 1990:331. Schuenke M, Schulte E, Schumacher U. Atlas of anatomy: head and neuroanatomy. Leipzig, Germany: Thieme 2010. Murdoch BE. Acquired speech and language disorders: a neuroanatomical and functional neurological approach. Sussex, UK: John Wiley & Sons 2010. Duffy JR, Folger WN. Dysarthria associated with unilateral central nervous system lesion: a retrospective study. J Med Speech Lang Pathol 1996;4:57-70. Muellbacher W, Artner C, Mamoli B. Motor evoked potentials in unilateral lingual paralysis after monohemispheric ischaemia. J Neurol Neurosurg Psychiatry 1998; 65:755-761. Muellbacher W, Artner C, Mamoli B. The role of the intact hemisphere in recovery of midline muscles after recent monohemispheric stroke. J Neurol Neurosurg Psychiatry 1999;246:250-256. Menard L, Aubin J, Thibeault M, et al. Measuring tongue shapes and positions with ultrasound imaging: a validation experiment using an articulatory model. Folia Phoniatr Logop 2012;64:64-72. Bressmann T. 2D and 3D ultrasound imaging of the tongue in normal and disordered speech. In: Maassen B, Van Lieshout P, eds. Speech motor control: new developments in basic and applied research. New York: Oxford University Press 2010:351-369. Bressmann T, Flowers H, Wong W, et al. Coronal view ultrasound imaging of movement in different segments of the tongue during paced recital: findings from four normal speakers and a speaker with partial glossectomy. Clin Linguist Phon 2010;29: 589-601. Bressmann T, Ackloo E, Heng CL, et al. Quantitative 3D ultrasound imaging of partially resected tongues. Otolaryngol Head Neck Surg 2007;136:799-805. Bressmann T, Thind P, Uy C, et al. Quantitative threedimensional ultrasound analysis of tongue protrusion, grooving and symmetry: data from 12 normal speakers and a partial glossectomee. Clin Linguist Phon 2005;19: 573-588. Perneger TV. What’s wrong with Bonferroni adjustments. BMJ 1998;316:1236-1238. Rastadmehr O, Bressmann T, Smyth R, et al. Increased midsagittal tongue velocity as indication of articulatory compensation in patients with lateral partial glossectomies. Head Neck 2008;30:718-726. Stone M, Shawker T, Talbot L, et al. Cross-sectional tongue shape during the production of vowels. J Acoust Soc Am 1988;83:1586-1596. Stone MA. Three-dimensional model of tongue movement based on ultrasound and x-ray microbeam data. J Acoust Soc Am 1990;87:2207-2217. Hamlet SL, Bunnell HT, Struntz B. Articulatory asymmetries. J Acoust Soc Am 1986;79:1164-1169.