Torsional and Vertical Eye Movements during Head Tilt Dynamic Characteristics

Abstract

purpose. As the first response to a Bielschowsky head tilt test (BHTT), a fast transient torsional eye movement in the same direction as the head tilt has been shown with the three-dimensional (3D)-video oculography (3D-VOG) technique, and this movement is paralleled by a transient vertical vergence shift inducing a physiological skew deviation. The purpose of the present study was to investigate these dynamic eye movements further in response to a BHTT paradigm with the magnetic search coil technique.

methods. Ten healthy subjects performed a BHTT (15°, 30°, and 45°) toward each shoulder while (search coil) the monocular eye and head positions were recorded. The same head tilt paradigm was repeated in a second test while (3D-VOG) the binocular eye position was recorded.

results. Subsequent to the initiation of the head tilt (latency, ∼160 ms) a rapid torsional eye movement (mean peak velocity: 40 deg/s; mean amplitude: 4°) was seen in the same direction as the head movement, followed by a somewhat slower return movement. This torsion was synchronous with a vertical vergence eye movement (mean amplitude 3°). The vertical vergence was always with left eye over right eye in the rightward head tilt and in head straightening from the left shoulder. In the left head tilt and in the head straightening from the right shoulder, this movement was always with the right eye over the left eye.

conclusions. A torsional and vertical vergence back-and-forth eye movement induced by a BHTT was confirmed with the search coil technique. Utricular inertia due to an interaural head translation, combined with a stimulation of the vertical semicircular canals, seems to be a plausible explanation for these eye movements.

In humans the sense of balance is preserved by integrating vestibular, visual, and proprioceptive sensory information. If there is no vestibular information, an upright body posture can still be preserved by the two other sensory modalities. However, an appropriate vestibular signal is crucial for maintaining clear vision. When turning the head horizontally or vertically (i.e., yaw or pitch) the vestibulo-ocular reflex (VOR) maintains visual fixation on the object of interest throughout the head movement and thereby reduces the motion of the image on the retina. The semicircular canals in the inner ear detect rotatory accelerations, such as when turning the head, while the otoliths detect linear accelerations during a translation, for instance, and through the earth’s gravitation. ¹ The canals and the otoliths are the anatomic substrates for VOR eye movements. 

The Bielschowsky head tilt test (BHTT) is a clinical test performed in subjects who have vertical diplopia and is used to diagnose motility disturbance of the oblique extraocular muscles. In a previous study, when the BHTT was performed on normal subjects, ² a rapid torsional eye movement was the initial response to the head tilt (roll). This eye movement was in the same direction as the head tilt—thus, in the opposite direction of a presumed compensatory movement. This response has previously been mentioned by some investigators ^{3 4} but has not, as far as we know, been elaborated fully. Furthermore, this rapid torsional eye movement was immediately followed by a return of the torsional position toward the initial reference position. ⁵ Simultaneous with the rapid torsional movement, there was a vertical disconjugate eye movement with the ipsilateral eye depressing and the contralateral eye elevating. This vertical disconjugate and torsional response to a head tilt has not been reported. The etiology of these findings is not known, but a vestibular origin seems plausible. The purpose of this study was to reinvestigate these findings with the same head-tilt paradigm but to supplement the video method used to record eye position in the former study ² we also used the scleral search coil technique to record both the monocular eye position and the head position. 

Methods

Subjects

The research adhered to the tenets of the Declaration of Helsinki. The study was approved by the local ethics committee at the Karolinska Institute. Written informed consent was obtained from all subjects. The group consisted of 10 healthy subjects (3 men and 7 women ranging from 20 to 28 with a mean age of 25). All subjects had a visual acuity of better than 20/20 wearing correction. No subjects with ametropia of more than ±1.0 D were permitted, because correction could not be worn while wearing the video-oculography (VOG) mask. All subjects had normal binocular vision including stereopsis (Lang <550 arc seconds). No subjects had any history of ocular or vestibular disease and none was taking medications that could influence visual or vestibular function. 

Visual Stimuli

The subjects were seated 1.5 m from a back-projected screen with the fixational stimulus displayed straight in front of the them, so that the stimulus was viewed with the eyes in the reference position. Because of the inherent problem of defining the real primary position in the current setting, the subject’s position with the head straight and the eyes looking straight ahead was defined as the reference position. ⁵ The stimulus was in the form of two yellow concentric circles (outer diameter subtending a visual angle of 3° and 18°) on a black background. A monocular horizontal and vertical calibration using a red dot stimulus (visual angle: 23 arc seconds) was performed before test 1 (eight-point calibration in 10° and 20° secondary eye positions) and test 2 (four-point calibration in 10° secondary eye positions). 

Data Recordings

In test 1, the eye movements were recorded (frequency, 25 Hz) binocularly using three-dimensional infrared video-oculography (3D-VOG, SensoMotoric Instruments, Teltow, Germany) on videotape for later off-line analysis. In test 2, the magnetic scleral search coil technique (Skalar, Delft, The Netherlands) was used to record (frequency, 500 Hz) the monocular eye position and the head position simultaneously. The head coil was fixed with surgical tape at the glabella on the forehead. 

Paradigms

Subjects performed two tests under identical conditions, except for the recording technique of the head and/or eye position. In addition, the right eye was anesthetized (tetracaine 1.0%) in test 2 to allow insertion of the silicone annulus. Both test paradigms started with the head in an upright position fixating binocularly with the eyes in the reference position ⁵ for at least 30 seconds before any data were collected. This was followed by head tilts of 15°, 30°, and 45° to the right shoulder and then back to the upright head position. After that, the head was tilted correspondingly to the left shoulder in three identical steps and then returned to the upright position. Each paradigm thus resulted in a total of eight head movements and nine head positions. Each head position was held for approximately 10 seconds. Head movements other than head tilt were restricted by a specially designed tiltable chin rest with a wooden bite bar. All movements of the chin rest were controlled by the investigator’s manually tilting the chin rest. The tilting of head position was performed promptly. Test 2 was performed 10 minutes after test 1 with the same test paradigm as in test 1. 

Definitions

Tilting of the head toward the shoulder was defined as head tilt, and the raising of the head from a head-tilt position was defined as head straightening. The term disconjugate was used to describe eye movements in the same direction but with different amplitudes, and disjugate was used to describe eye movements in contrary directions. 

Data Acquisition and Analysis

The 3D-VOG technique in test 1 resulted in binocular recordings of eye positions, whereas in test 2 with the search coil method, eye positions were recorded monocularly (right coil in the right eye) simultaneously with the head position (left coil on the forehead). The 3D-VOG data were used to determine the conjugacy of the eyes whereas the search coil data were used for calculating the torsional dynamic gain and to describe the characteristics of the torsional movements in response to the head movement. 

3D-VOG Data

The horizontal and vertical eye positions were evaluated by means of the so-called “black pupil technique,”—that is, the geometrical calculation of the center of lowest infrared reflection (center of pupil) using the Fick coordinate system. Ocular torsion was measured by the angular displacement of a defined iris segment. This was achieved by measuring luminance levels of the defined iris segment (profile) and subsequent correlation of the profile with that of each segment for each video frame. The concordance between the initially selected reference profile and that of the same iris segment of each following frame throughout the recording was computed by the VOG software and called “torsion quality.” This torsion quality was represented by a decimal value ranging between 0 (no concordance) and 1.0 (maximum concordance). According to the 3D-VOG instruction manual, only torsion data with a quality value of or above 0.3 should be considered for evaluation, because a lower value does not guarantee correct evaluation of torsion. The recordings were digitized and calibrated into ASCII data for the six channels (right and left eye; horizontal, vertical, and torsional data) and imported into a computer for evaluation (Origin software; Microcal, Northampton, MA). Data containing artifacts such as blinks were identified and removed. The signal was smoothed by adjacent averaging (five samples), the eye position data were then differentiated to obtain the eye movement velocity (deg/s). A velocity criterion was used (≥5 deg/s) to define the initiation of the eye movement to each head tilt, because no head position signal was available with the 3D-VOG. The mean eye position (duration, 1 second) before the defined head tilt was used as a reference position to calculate the amplitude of the induced eye movement. Eye movement conjugacy was determined by calculating the vergence of eye position data (i.e., left eye − right eye). The conjugacy during the head tilt was compared with the vergence position just prior (duration, 1 second) to the head tilt, and a Student’s t-test was performed. 

Search Coil Data

Simultaneous measurements of the three-dimensional eye and head rotations were performed and described in Fick coordinates. The recordings were digitized and calibrated into ASCII data for the six channels (right and left eye; horizontal, vertical and torsional data). The search coil data were then imported into a computer for evaluation (Origin; Microcal). The signal was smoothed by adjacent averaging (50 samples) and differentiated to obtain the eye and head movement velocity (degrees/second). A 2-sample differentiation was performed to obtain the eye and head acceleration (degrees/second squared). The orientation of the eye and head lens coils before the start of the head tilt was calculated with the use of the average head and eye position during the 100 ms before the head tilt. The dynamic torsional gain was calculated from the torsional slow-phase velocity after the torsional quick phases had been removed during the corresponding head movement. The quick phases were identified in the acceleration graph and the corresponding parts of the velocity data were removed. A linear fit tool was performed on the eye (Y) and head (X) velocity plot, and the slope value described the average dynamic gain. An acceleration criterion was set to define the start and the end point for the eye (50 deg/s²) and head movements (10 deg/s²). 

Potential Artifacts Due to Translation of the Head and Eye Lens Coil

Eye and head lens coils were precalibrated on a gimbal device. Because the head tilt paradigm performed consists of both a rotation and a translation of the head in the roll plane (interaural), we measured the effect of translation of the lens coil. Three subjects were photographed with a high-resolution digital camera to measure the amount of rotation and translation of the head during a BHTT. The average translation of the orbits from a head-straight position to the 45° head tilted position was less than 9 cm in the horizontal and less than 7 cm in the vertical direction. The static interaural translation artifact within the central 10-cm cube of the magnetic field, in which the subjects’ head always remained, was less than 0.01° for the torsional and horizontal channels and less than 0,02° for the vertical channel. The torsional deviation from linearity was less than 2.5% over an operating range of ±30°. When the range was extended to 45°, the linearity was 8.3% or less. 

Potential Artifacts Due to Data Filtering

The data were smoothed by a boxcar filter (adjacent averaging) to reduce signal noise and to allow calculation of torsional velocity and acceleration. The filter extended over 160 ms (50 samples at 500 Hz) for the search coil data and over 98 ms (5 samples at 25 Hz) for the 3D-VOG data. A too powerful filtering would interfere by shortening the latencies and would reduce the amplitudes of the torsion peak. The influence of the filtering on the data were estimated by regression analysis of the filtered and unfiltered data. The results demonstrate an interference of the filtering process in the search coil data (B = 0.97; coefficient of determination, >0.99) as well as on the 3D-VOG data (B = 0.95; coefficient of determination, >0.97). We considered the interference minor for the purpose of this study. 

Results

General Findings

The head tilt induced a rapid torsional and vertical vergence back-and-forth eye movement. This rapid torsional response occurred with both the search coil (Fig. 1A) and 3D-VOG (Fig. 1B) techniques, and is hereafter referred to as the torsion peak. The vertical vergence response was found in the binocular 3D-VOG data. 

The Head Movement

The head-tilt movement was analyzed from the left torsion search coil data. The horizontal (yaw) and vertical (pitch) head rotations were generally small (<2°). The 45° head tilt induced the largest deflections from a pure head tilt in the roll plane (≤7°). There was no difference in peak velocity between rightward and leftward head tilts (ANOVA). The average peak velocity in the first head tilt to 15° was 25 deg/s, whereas the head tilt toward 30° and 45° was performed more slowly, 20 deg/s and 15 deg/s, respectively. The fastest head movement was in head straightening, 40 deg/s on average. 

Torsion Peak

Subsequent to the initiation of the head tilt, a torsion peak occurred in the same direction as the head movement (Fig. 2) . This movement was initiated without any tendency to a slow-phase movement in the direction opposite to the head tilt (Fig. 3) . The latency for this peak was approximately 160 ± 30 ms (SD). The amplitude of the peak had a mean of 4° (range, 2°–6°) and the mean peak velocity was 40 ± 6 deg/s (SD; range, 30–50 deg/s). There was a positive and statistically significant (P < 0.01; coefficient of determination: 0.75) correlation between the torsion peak amplitude and velocity. After a time interval of approximately 350 ms from the initiation of the head movement, nystagmus beats (∼3 Hz) were superimposed on the torsion peak, with the slow phase directed in the direction opposite that of the head tilt (see Figs. 1 2 ). There was no significant disconjugacy of the eyes when the average cyclovergence position during the peak was compared with the average cyclovergence position before the head movement (Student’s t-test). 

Dynamic Gain

The dynamic gain was calculated from the data acquired during the head movement. The average dynamic gain was 0.51 ± 0.09 (SD). The coefficient of determination (R ²) of the linear fitting was never below 0.98. 

Vertical Eye Movements during Head Tilt

The head tilt also induced vertical disconjugate eye movements (Fig. 4) . A consistent finding in the 3D-VOG data was a fast vertical vergence shift simultaneous in time with the torsion peak described earlier. The direction of this movement was always with left eye over right eye in the rightward head tilt and in the head straightening from the left shoulder. In the left head tilt and in the head straightening from the right shoulder, this movement was always with right eye over left eye. The amplitude of the vertical vergence shift had a mean of 3° (range, 2°–5°). 

Discussion

A step-wise head tilt in the roll plane induced a rapid shift in the torsional and vertical eye positions. To the best of our knowledge, there is no previous description of this initial rapid response to the head tilt. The static properties of ocular counterrolling and vertical vergence as a response to a head tilt have been described in detail. ⁶  

Methodological Considerations

Because it can be assumed that both the action of gravity and inertia during head movements can lead to movement of the VOG head mask that could be interpreted as eye movements by the VOG software (i.e., chimera movements), a control experiment was performed by using a phantom eye. ⁶ The control experiments verified that head tilt gives rise to only a small displacement of the VOG mask. The largest vertical mask movement corresponded to a disconjugate 2° vertical chimera movement which in turn corresponds to maximum chimera torsion of 0.5°. The chimera torsion due to the vertical movements of the VOG mask were conjugate and to ranged on average between ±0.1° in the maximum head-tilted positions. There were no fast-twitch movements of the mask during the head tilt that could explain the fast eye movements that occurred. 

Torsional Eye Movements

The finding of a rapid torsional movement in the same direction as the head tilt is similar to findings by other investigators, ^{3 4} although they were never commented on. Of interest, the fast torsional movement shown in Figure 1 (P4), in the study by Averbuch-Heller et al. ³ occurred only unidirectionally in patients with skew deviation (left over right) and left internuclear ophthalmoplegia. 

The finding of a rapid torsional movement in the same direction as the head tilt could be related to the position of the rotational axis of the head. Thus, a BHTT rotates the head and neck around a sagittal axis located in the upper part of the chest. The position of this rotational axis is not fixed during the head tilt and depends on several factors such as the angle of the head tilt and stiffness of the neck and spinal column. The position of this rotational axis moves upward (i.e., rostral) with increased angle of head tilt (as can be seen from the photographic evaluation of orbit position described earlier). Because the axis of head rotation is located below the head, a tilt toward the right shoulder induces a rightward interaural translation (Y-direction) of the vestibular complexes. The BHTT paradigm used in the present study differs from the head movement paradigms used in the similar studies on vestibular eye movements in head roll where the roll axis is positioned between the eyes. ^{7 8} A head roll paradigm performed in a rotation chair with the axis of rotation positioned between the eyes of the subject primarily induces a divergent vertical translation (right ear downward and left ear upward in a rightward head roll) that would primarily stimulate the saccular maculae. The head tilt performed in this study differs from that in the studies mentioned earlier, ^{7 8} as the utricular mass initially moves in the reverse direction and initially bends the direction-sensitive receptor cilia in the utricular mass in the direction opposite to the translation. ⁹ The anticipatory torsional movement reported by Aw et al. ⁷ preceded the head movement and therefore was not considered to be a vestibular induced eye movement. 

Vertical Eye Movements

In the present study, a head tilt induced vertical disjugate or disconjugate eye movements (i.e., vertical vergence). Similar to the findings of a torsion peak we also found an initial rapid vertical vergence response corresponding to a physiological skew deviation. This vertical vergence response consisted of a left eye-over-right eye deviation in the rightward head tilt and in the head straightening from the left shoulder. Correspondingly, we found always a right eye-over-left eye deviation in the left head tilt and in the head straightening from the right shoulder. The etiology of this fast vertical vergence response is currently unknown but probably can be explained by a vestibular mechanism similar to that for the torsion peak. A vertical skew in response to both a static ¹⁰ and a dynamic ^{11 12} head roll has recently been reported. However, the stimuli in these studies were not similar to those used in the present study, in that we used a step-wise tilting of the subject, whereas the other two studies ^{11 12} used continuous sinusoidal stimulation in the roll plane. 

We presume that these rapid changes in torsional and vertical position of the eye are due to a combined stimulation of the vertical semicircular canals and to the inertia of the utricular mass during the acceleration of the head. A stimulation of the canals alone would not elicit the quick phase that we found in the data but rather the slow phase to stabilize the retinal image. The otoliths seem to induce a torsional movement not only during low-frequency head roll, but also during step-wise high-frequency head roll. Groen et al. ¹³ suggested that the otolith–ocular system acts to stabilize the eye position in space rather than to prevent retinal blur. Whether the torsion peak is due to stabilizing the retinal image or can be explained by an idiosyncrasy of the vestibulo-ocular system is not known. Further studies are necessary to obtain better understanding of the eye movement dynamics in response to a head tilt. 

 Supported by the Sigvard and Marianne Bernadotte Research Foundation for Children’s Eye Care, Huddinge University Hospital, Stockholm, Sweden.

 Submitted for publication February 3, 2003; revised March 11, 2003; accepted March 16, 2003.

 Disclosure: T. Pansell, None; H.D. Schworm, None; J. Ygge, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

(A, B) Graphs displaying the torsion peak from the search coil recording (A) and from the 3D-VOG recording (B). Note the nystagmus beats superimposed on the peaks as well as the similar appearance of the peaks in the two graphs. One step on the axes corresponds to 1° (y-axis) and 0.25 seconds (x-axis).

Figure 2.

 

Graph displaying the simultaneous search coil recordings of the head (dotted) and ocular torsion position (inverted for greater clarity). The shift of head position (bold arrow) was seen before the shift of eye position (small arrow). One step on the axes corresponds to 2° (y-axis) and 0.25 seconds (x-axis).

Figure 3.

 

Graph displaying the eye and head (dotted trace) velocity during head tilt. One step on the axes corresponds to 5 deg/second (y-axis) and 0.25 seconds (x-axis). This velocity plot corresponds to the position plot in Figure 2 . Note there was no tendency for the eye to accelerate in the direction opposite to the head acceleration.

Figure 4.

 

Graph displaying vertical vergence (i.e., left eye position − right eye position) and torsion position (dotted trace) from one representative subject. Positive (+) corresponds to right eye-over-left eye vergence and clockwise torsion. Observe that the fast vergence movements (arrows) were always synchronous with the fast torsion movements. One step on the axes corresponds to 1° (y-axis) and 5 seconds (x-axis).

The authors thank Roberto Bolzani for help with statistical analysis.