The development and validation of the eyelid pressure measurement system has been reported in detail.
13 The carrier rigid contact lens was designed based on the average corneal topography data of 100 young, healthy subjects, with a back optic zone radius of 7.8 mm and an eccentricity of Q = −0.25.
20 The contact lens diameter of 15 mm was chosen so that the upper and the lower eyelids rested on the lens. A specific tactile pressure sensor (Tekscan; Boston, MA) was chosen for this application because it is thin (approximately 0.17 mm), has a low pressure limit (5 psi), has good sensitivity, and is suitable for insertion into the eye. The sensor contains piezoresistive ink, which responds to mechanical deformation. The change in electrical resistance is recorded through a cable linked to a computer.
Before use, the pressure sensor attached to a rigid contact lens was conditioned to improve the consistency of its response. The sensor was conditioned with four loads of 25.9 mm Hg for 1 minute (with 30-second intervals between loads) less than 60 minutes before use. A calibration procedure was completed to convert the output from the sensor to actual pressure units. Loads of 1, 2, 2.5, 3, 3.5, 4, 5, 6, 8, and 10 mm Hg were each applied twice, and a linear fit was applied to the average pressure score between 10 and 30 seconds after loading began. The mean coefficient of determination (R 2) ranged from 0.68 to 0.96, with an average for the 11 subjects of 0.77.
The sensor-contact lens combination was placed on the eye, with the tears filling between the back surface of the contact lens and cornea (
Fig. 1). To obtain a stable measurement of eyelid pressure, the subject was required to maintain constant fixation during measurement, with limited eyelid movement. The most stable measurements were achieved when the non–measurement eye was manually held closed by the subject while fixation in the tested eye was maintained in a direction of approximately 10° downward gaze. Eyelid pressure measurements (each at 5 Hz) were taken until at least two, and preferably three, successful measurements were captured (that is, with the eyelid in the correct position on the sensor for at least 20 seconds). With a mean of 2.8 recordings per subject, each for approximately 12 seconds at 5 Hz, there was a total mean of 168 pressure readings per subject.
Given that the region of the eyelid contacting the sensor could not be visualized during pressure measurement, a front-view video of the eye was recorded to estimate the position of the eyelid margin (
Fig. 1). This was exported as.jpg images at 5 Hz and later analyzed with custom-written image processing software. The eyelid position (i.e., eyelid margin and estimated contact region with the ocular surface) relative to the lower edge of the pressure cell was determined (
Fig. 2).
A computer program (written for MATLAB; MathWorks, Natick, MA) was used to filter the sensor's pressure output data dependent on eyelid position with respect to the pressure cell. The first 10 seconds were discarded because of the initial noisy sensor response.
13 Outliers (>1.96 SD from the mean) were removed, most often because of eyelid twitches (i.e., small errors of eyelid movement detected in the front video recording). Assuming that the maximum pressure value occurs when the eyelid is in the correct position on the sensor, 50% of the power of the maximum pressure was used to define a lower boundary limit of data acceptance (upper half criterion). This is equivalent to 1/√2 of the signal's maximum amplitude or the logarithmic −3 dB criterion commonly used in acoustics
21 and the half-width point spread function used in retinal image metrics.
22 The raw scores within the upper half criterion were averaged, and the mean raw score was converted to actual pressure units (mm Hg) using the corresponding calibration equation. An example of an eyelid pressure measurement and the corresponding eyelid position is shown in
Figure 3. Although the eyelid moved steadily downward over the sensor during the measurement (dashed line), the corresponding pressure measurement was approximately a step function (solid line). Eyelid pressure was zero (−1 on the graphed normalized scale) after 52 seconds, when the primary contact point of the eyelid moved downward past the pressure cell. Pressure was no longer measured by the sensor (now well under the eyelid), suggesting that a band of the eyelid margin applied pressure to the ocular surface, whereas the remaining eyelid applied very little pressure (below the measurement noise of this technique).