Professional Documents
Culture Documents
DOI: 10.1007/s004550000024
Mario Prosiegel, MD,1 Michaela Heintze, MA,1 Edith-Wagner-Sonntag, MA,1 Thomas Schenk, PhD,2 and
Alexander Yassouridis, PhD3
1
Neurologisches Krankenhaus München; 2Neurologische Klinik, Klinikum Grosshadern, Ludwig-Maximilians-Universität München; and
3
Department of Statistics, Max-Planck-Institut für Psychiatrie, Munich, Germany
Abstract. To examine whether kinematic analysis of la- Patients undergoing such rehabilitation often have to
ryngeal movements (which are closely linked to pharyn- learn new motor tasks, e.g., techniques like the Mendel-
geal swallowing) can differentiate between normal and sohn maneuver [1,2] which requires “the patient to in-
disturbed swallowing, we used a three-dimensional ul- crease concentration on swallowing [3].” Mathematical
trasound movement recording system to measure the models [4] predicted that skilled movements following
movements of the larynx during swallowing in 32 pa- successful motor recovery are characterized by single-
tients with neurogenic dysphagia caused by central ner- peaked, bell-shaped velocity profiles [5,6]. This has been
vous system lesions and in 32 age- and sex-matched confirmed by measuring the kinematics of hand and arm
healthy individuals. At the beginning of an inpatient re- movements [7–10] and of movements of hand and fin-
habilitation swallowing program, laryngeal movements gers during writing [11]. Kinematic analysis (i.e., the
in 24 patients were highly disturbed in terms of velocity analysis of the velocity and acceleration) of swallowing
curve irregularities. After rehabilitation, the majority of movements enables the researcher to study in more detail
patients with hitherto irregular velocity profiles exhibited the motor recovery that takes place during motor learn-
laryngeal kinematics that were indistinguishable from ing. To the best of our knowledge, the kinematics of
those of 32 healthy subjects. Kinematic analysis of la- swallowing movements before and after rehabilitation of
ryngeal movements, therefore, is suitable for monitoring patients with dysphagia have not been investigated.
motor recovery of swallowing disturbances in patients Laryngeal movements play a pivotal role in the
with neurogenic dysphagia while undergoing swallowing pharyngeal phase of swallowing. One of the swallowing
rehabilitation. techniques, the Mendelsohn maneuver [1,2], is based on
the volitional prolongation of anterior–superior larynx
Key words: Neurogenic dysphagia — Laryngeal movement, which widens the pharyngoesophageal seg-
movements — Kinematic analysis — Ultrasound move- ment (PES) and decreases the pressure within the PES
ment recording device — Motor recovery — Rehabili- [12–14]. To analyze swallowing movements, we mea-
tation — Deglutition — Deglutition disorders. sured laryngeal kinematics in patients with neurogenic
dysphagia as well as in healthy controls. Because of the
high degree of spatial and temporal resolution needed for
As in other areas of motor rehabilitation, motor learning this purpose, a three-dimensional (3D) ultrasound move-
plays a pivotal role in functional swallowing therapy. ment recording system was used.
from 28 to 87 years (mean ⳱ 57.7 years). Inclusion criteria were In order to assess “normality” or “abnormality” of laryngeal
neurogenic dysphagia resulting from disease of the central nervous movements, we compared the duration and velocity profiles of down-
system (CNS) as defined by history, clinical examination, and magnetic ward movements in patients with neurogenic dysphagia before swal-
resonance imaging (MRI); dependence on nasogastral or PEG feeding lowing treatment with those of healthy controls. Duration of movement
on admission; age ⱖ18 and ⱕ90 years; acquisition of the Mendelsohn is an obvious choice, because it describes the efficiency of a movement
maneuver indicated in all cases because of PES dysfunction as shown in a simple way. If the same movement can be performed in less time,
by videofluoroscopy [15]. Characteristics of the patient sample are this is clearly an advantage. Since the velocity profiles of skilled and
shown in Table 1. Thirty-two age- and sex-matched healthy individuals overlearned movements are highly stereotyped, the velocity profile is
with no known history of significant oral/pharyngeal anomaly served as an attractive yardstick for distinguishing between normal and abnormal
controls. movements. In the case of prehensile movements, handwriting, and
saccadic eye movements, it has been shown that the velocity profile for
a single movement (i.e., one prehensile movement toward an object,
Data Acquisition one upward or downward stroke in handwriting, one saccade) has a
characteristic bell-shaped form, is almost symmetrical, and has only
one peak [7,11,17]. In contrast, during skill acquisition [18] or in patho-
Position and movement of the larynx during swallowing were tracked logical cases [19], the velocity profile is multipeaked and irregular.
using a 3D movement registration system (CMS 50, Zebris, Isny, Ger- In the present study, we investigated whether kinematic mea-
many). This system uses small markers (diameter ⳱ 1.0 cm, weight ⳱ sures that have been applied successfully for distinguishing normal
1 g) which continuously emit an ultrasonic signal at a sampling fre- from abnormal movements in other domains of human movements can
quency of 200 Hz every 5 ms. Three microphones arranged within a also be used to characterize normal or abnormal swallowing move-
plane at fixed distances to each other and mounted on a stationary ments. For quantitative characterization of the velocity profile, we used
frame (40 × 40 × 5 cm) receive the signals. The system measures the the number of positive peaks, the duration of movement, and the peak
transmission time from the markers to the three microphones and uses amplitude of the velocity profile (peak velocity).
the differences in transmission time for the different microphones to We restricted our analysis to the downward movement for three
calculate the spatial position of the markers. The system works reliably reasons: First, healthy individuals did not exhibit substantial differ-
in a workspace of 2 × 2 × 2 m and has a spatial resolution of 0.1 mm. ences in kinematics between posterior–anterior (forward), inferior–
Movements are registered and the data stored for later offline analysis superior (upward), and superior–inferior (downward) movements. Sec-
on a PC (Pentium II fitted with PC card for the ISA Bus, developed by ond, the Mendelsohn maneuver requires volitional prolongation of up-
Zebris). We used one marker attached with double-sided tape to the ward and forward movements of the larynx. Thus, the downward
skin just at the area of the thyroid notch midline. The distance between movement is least susceptible to the effect of this swallowing tech-
the frame and the subject was 1 m. This method does not permit
nique. Third, the signal–noise ratio is much worse for movements in the
measurement of the full range of laryngeal movements because the
posterior–anterior axis than for those in the superior–inferior axis, since
marker, which is attached to the skin and not to the larynx itself, cannot
the range of movement in the forward direction is less than that in the
fully follow the movement of the larynx. Nevertheless, it is still pos-
downward direction.
sible to measure the velocity characteristics of the larynx.
The same device used in this study has already been used for
kinematic analysis of hand and arm movements [9,10].
The 3-D movement registration system is illustrated in Figure 1. Procedure
Laryngeal kinematics were measured twice for each patient, the first
Data Analysis
time prior to commencing the rehabilitation program and the second
time at discharge. The patients as well as the controls were instructed
The positional data coming from the CMS 50 are associated with a to sit as comfortably as possible and to try to swallow normally. In
random spatial error of approximately ±0.58 mm. If unfiltered, this order to avoid significant aspirations, only dry swallows were per-
results in a dynamic error of 82 mm/s and 11,600 mm/s2 for the first formed by each person five times during each session. After each
and second derivatives, respectively. We used a nonparametric estima- swallow, the acquired data were stored. During offline analysis, the
tion of regression functions based on kernel estimates [16]. This beginning and the end of movement (i.e., velocity is zero at the begin-
method has been evaluated using simulated and real data sets. The ning and at the end of the movement under analysis) were marked in
results obtained with the kernel estimates are very similar to those order to ensure that the curve really did reflect a dry swallow. The time
obtained with more conventional filter methods [far infrared (FIR) and needed for this procedure took about five minutes and was, therefore,
butterworth filters]. Although kernel estimates show a slightly elevated long enough to enable the patient to gather saliva for the next swallow.
bias for the second derivative, they reduce the residual variance in the Positive peaks as indicators of irregularities within velocity
derivatives more efficiently than both FIR and butterworth filters. In curves were counted by two blinded raters for both healthy controls and
addition, filtering with kernel estimates offers a very fast method of patients (Kendall’s tau-b ⳱ 0.937 and 0.975, respectively). The mean
data smoothing. This filter method was initially developed for the values of the velocity curve characteristics (number of positive peaks,
analysis of positional data coming from a digitizing board [11], but it duration of downward movement, and peak velocity) over the five dry
can be adapted to the analysis of 3D position data coming from the swallows were chosen for the final assessment of the healthy controls
CMS 50. However, since the error characteristics are different for the and the patients (at the time of first and second measurement). At the
CMS 50 (see above) and the digitizing board, the bandwidths of the end of the rehabilitation program each patient was asked if he/she
filters have to be adapted. Simulation studies showed that the optimal needed to concentrate on his/her swallowing.
bandwidths for data coming from the CMS system are 50 ms/12 Hz, 70 In six patients laryngeal movements were also measured during
ms/10 Hz, and 90 ms/8.5 Hz for the zeroth, first, and second deriva- therapy sessions, i.e., when explicitly instructed to perform the Men-
tives, respectively [11]. delsohn maneuver volitionally.
M. Prosiegel et al.: Motor Recovery in Swallowing Rehabilitation 175
23/9 57.7 ± 16.0 Dorsolateral medullary 96 ± 668 Total oral n ⳱ 22 81.9 ± 76.3
(28–87) infarction (n ⳱ 17) (11–3500) PEG n ⳱ 10 (10–310)
Middle cerebral artery infarction (n ⳱ 8)
left/right: 5/3
Unilateral pontine infarction (n ⳱ 2)
Brainstem encephalitis (n ⳱ 2)
1 postvaccinal (polio)
1 viral of unknown cause
Severe head injury (n ⳱ 2)
Medullary cavernoma (n ⳱ 1)
Results
Table 2. Parameters of velocity curves (mean ± SD and range) obtained by assessment of laryngeal kinematics with a 3D ultrasound movement
recording system in patients with neurogenic dysphagia at the time of admission as well as in healthy controls
Number of positive peaks 6.4 ± 4.5 1.3 ± 0.88 28.06 p < 0.05
(2–22) (0–3)
Duration of movement (msec) 846.3 ± 423.3 336.6 ± 111.4 30.28 p < 0.05
(250–1850) (125–500)
a
Univariate F-tests in MANOVA.
make use of the Mendelsohn maneuver so automatically In summary, kinematic analysis of laryngeal
that they are not really aware of performing this newly movements is a suitable method for distinguishing nor-
learned technique. mal from disturbed swallowing and for monitoring the
Certain limitations of this study need to be men- degree of motor recovery achieved during swallowing
tioned. First, the patient group was not homogeneous and rehabilitation.
the control group consisted of healthy persons. We are
currently conducting studies on a more homogeneous Acknowledgments. The authors wish to thank Anita Wuttge-Hannig,
sample of dysphagic patients with respect to etiology as M.D., and Christian Hannig, M.D., who performed the videofluoro-
well as of nondysphagic neurological patients. Prelimi- scopic examinations.
nary results show that the velocity profiles of the non-
dysphagic neurological patients do not differ from those
of healthy individuals. Second, the specificity and sen- References
sitivity of the method still have to be studied more ex-
tensively, including the measurement of laryngeal up- 1. McConnel FM, Cerenko D, Mendelsohn MS: Manofluoro-
ward and forward movement characteristics, which are graphic analysis of swallowing. Otolaryngol Clin N Am
21:625–635, 1989
strongly expected to be influenced by certain swallowing 2. Logemann JA: Noninvasive approaches to deglutitive aspira-
techniques. Third, since it cannot be ruled out that subtle tion. Dysphagia 8:331–333, 1993
differences in velocity profiles depend on whether the 3. Miller RM, Langmore SE: Treatment efficacy for adults with
medullary pattern generators or the cerebral cortex are oropharyngeal dysphagia. Arch Phys Med Rehabil 75:1256–
affected, further studies with larger patient groups are 1262, 1994
4. Flash T, Hogan N: The coordination of arm movements: an
needed in order to answer this question. Fourth, on the experimentally confirmed mathematical model. J Neurosci
basis of our results, it is not possible to give a definitive 5:1688–1703, 1985
answer to the question of whether normalization of hith- 5. Fitts PM: Perceptual-motor skill learning. In: Melton AW (ed):
erto abnormal curves reflects only motor recovery or also Categories of human learning. New York: Academic Press,
motor learning processes, including automaticity of 1964, pp 243–285
6. Rosenbaum DA: Human motor control. New York: Academic
movements. According to Fitt’s stage theory [5] of skill Press, 1991
learning, during the first (cognitive) stage the basic pro- 7. Morasso P: Spatial control of arm movements. Exp Brain Res
cedures to be followed are learned, e.g., by verbal cues 42:223–227, 1981
with a considerable amount of attention. The second (as- 8. Abend W, Bizzi E, Morasso P: Human arm trajectory formation.
sociative) stage represents a transition from conscious Brain 105:331–348, 1982
9. Hermsdörfer J, Mai N, Spatt J, Marquardt C, Veltkamp R, Gold-
control to more automatic control. In the third (auto- enberg G: Kinematic analysis of movement imitation in apraxia.
matic) stage, movements can be performed quickly and Brain 119:1575–1586, 1996
consistently with little conscious involvement and are 10. Hermsdörfer J, Ulrich S, Marquardt C, Goldenberg G, Mai N:
characterized by single-peaked, bell-shaped velocity pro- Prehension with the ipsilesional hand after unilateral brain dam-
files. Ongoing studies are examining the swallowing pat- age. Cortex 35:139–161, 1999
11. Marquardt C, Mai N: A computational procedure for movement
tern of dysphagic patients while performing parallel (at- analysis in handwriting. J Neurosci Meth 52:39–45, 1994
tentional) tasks (i.e., dual task condition). Preliminary 12. Ekberg O, Olsson R: The pharyngoesophageal segment: Func-
results indicate that normal velocity profiles are posi- tional disorders. Dis esophagus 8:252–256, 1995
tively correlated with automaticity of swallowing move- 13. Pouderoux P, Kahrilas PJ: The pharyngoesophageal segment:
ments as defined above and thus with motor learning Normal structure and function. Dis esophagus 8:233–241, 1995
14. Kahrilas PJ, Logemann JA, Krugler C, Flanagan E: Volitional
processes. Automaticity of swallowing is especially im- augmentation of upper esophageal sphincter opening during
portant as it can be expected to have positive conse- swallowing. Am J Physiol 260:450–456, 1991
quences for the disability and for the handicap level as 15. Hannig C, Wuttge-Hannig A: Radiologische Diagnostik der
defined by the WHO [24]. For example, the patients’ Schluckfunktion. In: Bartolome G, Buchholz DW, Feussner H,
independence of tube feeding and their ability to swallow Hannig C, Neumann S, Prosiegel M, Schröter-Morasch H, Wut-
tge-Hannig A (eds): Schluckstörungen—Diagnostik und Reha-
either without or with only minor concentration would bilitation. München, Jena: Urban and Fischer, 1999, pp 65–110
make them feel safer when dining out with other people. 16. Gasser T, Müller HG: Estimating regression functions and their
Finally, we would like to stress that this study was not derivates by the kernel method. Scand J Statist 11:171–185,
intended to assess the efficacy of swallowing therapy. 1979
Therefore, on the basis of the method we used, no state- 17. Westheimer G: Mechanism of saccadic eye movements. Arch
Ophthalmol 52:710–724, 1954
ment can be made as to whether normalization of swal-
18. Konczak J, Dichgans J: The development toward stereotypic
lowing movements during swallowing therapy was due arm kinematics during reaching in the first 3 years of life. Exp
to swallowing therapy or spontaneous recovery or a com- Brain Res 117:346–354, 1997
bination of both. 19. Eichhorn TE, Gasser T, Mai N, Marquardt C, Arnold G,
M. Prosiegel et al.: Motor Recovery in Swallowing Rehabilitation 179
Schwarz J, Oertel WH: Computational analysis of open loop 22. Martin RE, Sessle BJ: The role of the cerebral cortex in swal-
handwriting movements in Parkinson’s disease: a rapid method lowing. Dysphagia 8:195–202, 1993
to detect dopamimetic effects. Mov Disord 11:289–297, 1996 23. Hamdy S, Aziz Q, Rothwell JC, Hobson A, Thompson DG:
20. Buchholz DW, Bosma JF, Donner MW: Adaptation, compen- Sensorimotor modulation of human cortical swallowing path-
sation, and decompensation of the pharyngeal swallow. Gastro- ways. J Physiol 506:857–866, 1998
intest Radiol 10:235–239, 1985 24. World Health Organization: The international classification of
21. Miller AJ: The search for the central swallowing pathway: the impairments, disabilities, and handicaps. Geneva: World Health
quest for clarity. Dysphagia 8:185–194, 1993 Organization, 1980