High-Speed Videotopographic Measurement of Tear Film Build-up Time

Abstract

purpose. To detect tear film regularity changes in the 15 seconds after a blink, by using a new high-speed videotopographic method.

methods. The new system, based on a standard corneal topograph, allows registration of four photokeratoscopic images per second. Altogether, 15 eyes of 15 healthy volunteers and 7 eyes of 7 patients with dry eye were examined in this prospective preliminary study. The main outcome measures were changes in surface regularity index (SRI), surface asymmetry index (SAI), and corneal power.

results. The corneal surface became more regular in the first few seconds after a blink. In healthy eyes, the trend line of SRI and SAI decreased (improved) significantly (P < 0.001) after a blink, in 10 of 15 eyes for the SRI and in 13 of 15 for the SAI. In the typical cases, the trend line for SRI reached its minimum level, on average, at 7.1 ± 3.9 seconds after a blink and that for the SAI at 5.4 ± 2.7 seconds. Similar trends were found in the dry-eye group. The changes in keratometric measures were small (mean range, <1.5% of the absolute value) and showed no definite trends.

conclusions. High-speed videotopography provides the possibility of quantitative measurement of tear-film dynamics and may have clinical value in the management of ocular surface disorders. After a blink, it takes the tear film approximately 3 to 10 seconds (tear film build-up time) to reach the most regular state. However, despite surface-regularity changes, the measured corneal powers are stable.

The cornea and especially its anterior surface, the thin and optically smooth tear film, is the major refractive interface of the eye. Corneal topography is a sensitive method for measuring and mapping corneal refractive power, ¹ and moreover it reveals the changes in the tear film. ^{2 3 4} The tear film alterations cause surface irregularities that distort the topographic image. ^{2 4 5} These changes can be quantified using the surface regularity index (SRI) and surface asymmetry index (SAI), introduced by Dingeldein and Klyce ⁶ and Wilson and Klyce. ⁷ The indices are dimensionless numbers, and they increase as the corneal surface departs from the ideal smooth surface and overall spherical form. 

The tear film is not stable in time. It builds up quickly after the eyelids are opened. ^{8 9} Evaporation then starts, and the tear film becomes thinner and finally breaks up. ^{10 11} The break-up time (BUT) is a clinically well-established entity with diagnostic implications, ^{10 12} but the process of the tear film build-up is not well known. It has been estimated that after a blink it takes the tear film 2 to 3 seconds to spread evenly on the corneal surface. ⁸ Recently, Owens and Phillips ¹³ have measured the displacement of tear film particles after a blink and found that tear stabilization time, defined as cessation of lateral movement of the tear fluid, was approximately 1 second. However, to our knowledge, no actual measurements of tear film build-up time have been published to date. 

In an earlier study, ³ we found that the corneal surface in most cases is significantly more regular at 5 seconds than at 15 seconds after a blink, which may be attributed to evaporation and thinning of the tear film. However, in a few of the (healthy) subjects there was an opposite trend: The tear film was found to be more regular at the later time. We attributed this contrary effect to the possibility of slow tear film build-up, which results in the tear film’s only reaching its best regularity later than 5 seconds after a blink. 

To allow us to answer the question of whether it takes the tear film a certain time after a blink to build up and achieve the most regular surface, we developed a high-speed videotopographic examination technique. The purpose of our study was to investigate the pattern of rapid changes in tear film surface regularity during the first 15 seconds after a blink. 

Subjects and Methods

Our videotopography system uses the video signal from the photokeratoscope part of a corneal topograph instrument (TMS-1; Computed Anatomy Inc., New York, NY). The National Television System Committee (NTSC) video signal of the photokeratoscope is sent to a supplementary desktop computer equipped with a frame-grabber card, and the special software allows online registration of four images per second during a period of up to 15 seconds. After storage of the image sequence, individual images are transferred back to the system’s computer for analysis in offline conditions. In the present study, before the analysis was performed, the fixation error was registered and corrected by manual centration of the image. The analysis of each topographic image consisted of the derivation of the corneal refractive powers (K₁, K₂, K_min; i.e., the power in the meridian with the greatest curvature and in the meridian perpendicular to it and the power in the meridian with the least curvature) and calculation of Klyce corneal statistics (ver. 1.1), including the SRI and SAI. ^{6 7} Also, for each photokeratoscopic image, the maximum width of the palpebral fissure was measured on the photokeratoscope screen with a millimeter scale, and then the true lid fissure width was calculated by using the on-screen magnification factor. 

The right eyes of 15 healthy volunteers were examined. All subjects had full visual acuity and negative ophthalmic status and did not wear contact lenses. They comprised 12 women and 3 men, aged from 20 to 56 years (mean, 31.5 ± 10.4). A separate group comprising seven eyes of seven female patients with dry eye was also examined. The seven patients (ages 48–72 years; mean, 58.6 ± 7.7) had at least a 6-month treatment history for various dry-eye conditions: Three had primary Sjögren syndrome, one had secondary Sjögren syndrome due to systemic lupus erythematosus, and three had keratoconjunctivitis sicca. The diagnoses were based on subjective symptoms and results of ophthalmic examination and tests (Schirmer 1 test, tear film BUT, fluorescein staining), as well as on rheumatologic and dermatologic findings. The first videotopographic examination was performed after a pause of at least 14 hours in the patient’s ongoing artificial tear therapy, and a second examination was performed 3 minutes after instillation of 1 drop of a proprietary artificial tear solution that contained only physiological saline ophthalmic solution without preservative (Unilarm; Novartis/Ciba Vision, Basel, Switzerland). In all cases (both healthy subjects and patients with dry eye), tear film BUT (determined by the fluorescein-imbibed strip technique) and Schirmer 1 test results at 1 minute and 5 minutes were recorded, but in a session separate from that used for the videotopographic examination(s). All participants were advised of the nature and purposes of the examination, and informed consent was obtained from each person, according to the provisions of the Declaration of Helsinki. 

The protocol used was as follows: the subject placed his or her head against the support rest of the topograph and was asked to look straight ahead and not to move. Fixation and centration were performed. The subject was then asked to make a complete blink and subsequently to keep the eyes open and to fixate continuously. At the same time, the laser aiming beams were switched off, and acquisition of video images was initiated by the frame-grabber software. For each subject, 60 images were registered and stored during the 15-second period, just after the blinks. The first image in which the eye was not closed was considered to be the 0.25-second image. This first image, however, was omitted from the statistical analysis, because in many cases it exhibited partially opened lids; extreme SRIs, SAIs, and Ks; and a high fixation error. 

To analyze the changes with time of the SRI, SAI, and K parameters and the changes of the fixation error and lid fissure width, mathematical statistical modeling was applied. Using the algorithm implemented in commercial software (SPSS, ver. 9.0; SPSS, Chicago, IL), the time series of the SRIs and SAIs were each decomposed into a fourth-order polynomial trend line and a first-order autoregressive (AR-1) random-noise series. The decomposition algorithm applies iterative maximum-likelihood estimates of the component parameters to achieve the final results. ¹⁴ On the basis of the obtained estimates of parameters and their statistical properties, maximum-likelihood estimates and asymptotic significance tests were derived for other indicators, ¹⁵ such as the time position and value of the first minimum of each polynomial trend line (e.g., minimum value tested for significant change from the polynomial initial value). As a further stage of analysis, the estimated AR-1 parts of the SRI and SAI sequences were tested in individual subjects at various time lags for cross correlation with one another and with fixation error, with lid fissure width, and with the sequence of incremental differences between adjacent observations of K. Similarly, all the other pairs of these sequences were tested against each other. 

We analyzed the results obtained in the 15 healthy volunteers in an attempt to reveal correlations between the high-speed videotopographic data (the initial SRI and SAI at 0.5 second after a blink, their minimum levels, the time to reach the minimum, and the coefficients of their fitted polynomial trends) and the results of classic tear tests (Schirmer-test results at 1 and 5 minutes and BUT). In subjects with stable or increasing trend lines of SRI or SAI after opening the eyes, in whom, in fact, no initial decrease was observed, for purposes of statistical analysis, the minimum value of the parameter was taken to be the initial value, and the corresponding time to reach the minimum was taken to be zero. (In the tabulated results, the form of the trend line for such subjects is briefly described in words.) 

Results

After the eyelids were opened, a clear tendency toward changes in SRI and SAI, relatively consistent in different subjects, was evident. In most subjects, an initial improvement of surface regularity was observed, reaching an optimum value at some time after 2 seconds. Subsequent to this, however, great variability was found between subjects (Fig. 1 , Tables 1 2 3 4 ). 

Healthy Subjects

In 11 of 15 healthy subjects, the SRI decreased and reached its minimum in an average of 7.1 ± 3.9 seconds (SD) after a blink. The individual decrease was significant in 10 eyes (P < 0.0001, Table 1 ). The remaining four eyes showed various patterns of SRI change over time, as follows: In eyes 14 and 15, after a short initial relatively stable period (4 seconds), a rapid increase occurred that was followed by a continuous decrease (Fig. 2) . In eye 10, a different pattern was found (continuous increase of SRI, followed by a stable period), and in eye 7 the SRI was very low and stable over the whole examination period. Of the 11 eyes with initial decrease in SRI, in 6 eyes (1, 3, 4, 5, 6, and 8) an SRI increase started at between 4.0 and 8.4 seconds after a blink. In the other five eyes (2, 9, 11, 12, and 13), after the minimum level, the SRI was stable up to the end of the 15-second examination period. 

In 13 healthy eyes the SAI showed a statistically significant decrease after a blink (P < 0.001, Table 2 ). However, in eye 11, there was a short and clinically nonrelevant initial increase before the value started to decrease, but this may have been an artifact due to the well-known type of mathematical modeling error that is common in regression models at the beginning or end of the time domain. ¹⁶ Eyes 10 and 12 had a short initial stable period (∼3 seconds), followed by a quick increase and a subsequent slow decrease, similar to the SRI trend lines in eyes 14 and 15, described earlier. The minimum SAI was reached, on average, at 5.4 ± 2.7 seconds after a blink. After this minimum, 8 of these 13 eyes showed an SAI increase starting at between 3.2 and 7.9 seconds after the blink. Two eyes had a stable SAI, and the remaining two eyes showed a stable SAI followed by a further decrease later on. 

The changes in corneal power (K) were moderate in healthy subjects. The maximum range in an individual eye was 1.5 D, and no definite trends were noted (a typical case is shown in Fig. 3 ). The mean range of K₁ was 0.59 ± 0.30 D (SD), that of K₂ was 0.53 ± 0.34 D, and that of K_min was 0.58 ± 0.39 D. The differences between adjacent observations showed only slight autocorrelation, with correlation coefficients no greater than 0.5 in absolute value. Consequently, the changes of keratometric values can be interpreted as cumulations of autocorrelated random disturbances. 

To explore possible relationships between rapid changes in the different measured or derived values, in each subject, cross correlations at various time lags were evaluated between, for example, the estimated AR-1 part of the SRI time-series and the four other parameter series for fixation error, lid fissure width, the AR-1 part of the SAI, and the sequence of incremental differences between adjacent measurements of corneal power. This process was repeated for all pairs in turn between the four series and for the three representatives of corneal power: K₁, K₂, and K_min (altogether 18 cross correlations; the K₁, K₂, K_min series were not tested against each other). We did not detect any significant correlation patterns between the time series mentioned, with the greatest observed correlations being approximately 0.5 in absolute value. 

The trend lines for SRI and SAI were somewhat similar in appearance in several of the healthy eyes. The minimum SRIs and SAIs on the trend lines showed strong positive correlation with each other (Kendall τ-b = 0.6, P = 0.002). However, there was not enough statistical power to analyze with great precision the coefficients (total number, 10) of the two fourth-order polynomial trend lines, with data from only 15 observed eyes. The results of the classic tear tests (Schirmer test at 1 and 5 minutes and BUT) did not correlate significantly with these 10 trend coefficients; but, as mentioned, the power of the tests was low. A larger study with more eyes is needed to assess these connections with sufficient confidence, although the number of necessary eyes can in principle be estimated from the data in this study. 

In healthy subjects, the times to reach the minimum SRIs in the trend lines (t _min in Tables 1 and 2 ) correlated negatively with the Schirmer test results at 1 minute (Kendall τ-b = −0.421, P = 0.044) and at 5 minutes (Kendall τ-b = −0.418, P = 0.035). However, the corresponding t _mins for the SAI trend lines did not correlate significantly with the classic tear indicators, and there were also no correlations detected between the initial SRIs and SAIs (the indices at 0.5 second after a blink) and the results from the classic tear indicators (Schirmer test and BUT). Similarly, no correlations were detected between the minimum SRIs and SAIs (the minimum level of the respective polynomial trend lines) and the classic tear test results. 

Patients with Dry Eye

In the patients with dry eye, in five of the seven eyes, the SRIs and SAIs behaved in a manner similar to that in the typical healthy subjects. The indices initially decreased significantly and reached a minimum in an average of 4.7 ± 1.6 seconds after a blink for SRI and 5.0 ± 3.1 seconds for SAI. However, two eyes showed different patterns of change in SRI or SAI (Tables 3 and 4) . A typical sequence of images (patient 16) is shown in Figure 4A . Note the pronounced irregularities in the pattern in the initial image, which disappeared as tear film built up over approximately 2 seconds. 

Figure 4B shows the effect of artificial tear fluid in the same patient. The image quality is improved considerably and is now comparable to that in a normal subject. However, the graph in Figure 5 of data from patient 20 shows that, even though a considerable initial improvement of surface regularity (SRI index) may be brought about by use of the eye drops, the improvement may not be maintained at longer times. In this case, at times longer than approximately 8 seconds after the blink, the regularity became progressively worse (higher SRI) than at the corresponding postblink times without use of eye drops. 

The instillation of 1 drop of artificial tear fluid did not produce consistent results in different patients with dry eye. In approximately half of the eyes, the SRI and SAI trend lines were of a shape comparable to that before instillation, although not always with improved indices. In the remaining eyes, a tear film build-up pattern (initial SRI or SAI decrease) became apparent in some patients, whereas the previous typical pattern disappeared in the remainder (Tables 3 4) . 

On average, the initial and minimum levels of the SRI and SAI trend lines in subjects with dry eye were somewhat higher than those in the healthy subjects (Tables 1 2 3 4) . The respective averages of the initial and minimum values of these parameters became lower (and close to the levels in healthy subjects) after instillation of proprietary tear fluid (Tables 3 4) ; however, these changes were not shown to have statistical significance (P > 0.3), possibly because of the small number of patients in the dry-eye group and the inhomogeneity of this group. 

Discussion

We used a newly developed high-speed videotopographic method that allowed online acquisition of photokeratoscopic images four times per second during a period of 15 seconds immediately after a complete blink. The analysis of these high-speed videotopographic images of healthy eyes revealed significant rapid changes in the topographic indices (SRI, SAI), whereas changes in corneal power were very small (<0.6 D, on average, corresponding to <1.5% of the absolute value), close to the measuring accuracy. ^{17 18} The latter finding is in agreement with the findings of Buehren et al., ¹⁹ who also found high stability of corneal topography in the postblink interval for the refractive power in the central (4–5 mm in diameter) corneal regions, although, in the upper and lower regions of the maps, they found statistically significant changes in the corneal power. They attributed these alterations to the mechanical effect of the eyelids. The apparatus uses a target comprising 31 concentric rings. In our study, we used the SRI, SAI, and K parameters, which are calculated from data derived from different extents of the whole target. The SRI is derived from only the central 10 circles and is not influenced by changes outside this area, whereas K is derived only from rings 6, 7, and 8. The SAI parameter is influenced by the more peripheral parts of the image; however, the data are averaged for the whole cornea. Thus, the outcome measures of our study are mainly influenced by changes in the central region of the corneal surface and tear film. 

In most of the eyes, SRI and SAI decreased in the first few seconds after a blink, which implies that after the eyelids are opened it took the tear film some time to build up and reach its highest regularity and optical quality. In 15 healthy eyes, we found that this tear film build-up time was 5 to 7 seconds on average. Later on, the changes in SRI and SAI were variable. Further improvement, stable indices, and worsening were observed in different subjects. Because of the short examination period and because we did not perform the conventional measurements to determine this parameter at the same measurement session, we cannot draw direct conclusions regarding the tear film break-up time; however, the observed increase of the SRI and SAI toward the end of the 15-second measurement period (in 55% and 64% of the subjects, respectively) may be an indicator of imminent tear film break-up. These results are comparable to those of Norn, ¹² who found a tear film break-up time of less than 20 seconds in 44% of normal eyes. 

We found in normal subjects that the time (t _min) necessary to reach the optimal surface regularity (minimum SRI and SAI levels) depends on the quantity of tears: The higher the Schirmer result, the shorter the time interval required for tear film build-up. However, in all other aspects (initial level, minimum level, form of trend line), apart from the connection with t _min, there appeared to be no relationship between the classic tear-test results (Schirmer test, BUT) and the tear film build-up parameters determined with high-speed videotopography. 

Regarding the correlation between t _min and tear quantity, our findings appear consistent with the findings of Owens and Phillips, ¹³ who reported that after the aqueous phase is increased by stimulation with onion vapor, the tear stabilization time decreases significantly. The noncorrelation in other respects may in part be because, as generally believed, the measured BUT correlates poorly with the results of other tear tests. ¹³ Also, tear film build-up time may reflect other characteristics of the tear film than those that are measured by the BUT or by the Schirmer test. 

Some limitations of our study were the relatively small number of subjects, the short observation period, the fixation error, and the alterations in lid fissure width that were found during measurements. However, neither the fixation error nor the lid fissure width correlated with the changes found in SRI, SAI, and K. 

The fixation error during continuous topography was found to be greater than the acceptable limit (0.125 mm) mentioned in the instrument manual, because in the present study we were unable to center the eyeball again after image acquisition had started. To overcome this problem, before analysis of each image, we corrected the fixation error by manual centration of the image. In the future we hope to develop and use a high-speed eye-tracking system that can monitor the fixation and readjust the photokeratoscope before each image acquisition. 

The eyelid fissure width also changed during the examination period, which in principle might affect the tear film surface parameters, because widening of the eyelids causes thinning of the tear film and also increases the evaporation of tear fluid because of the increased surface area. ²⁰ However, in practice we could not detect any such influence, a result consistent with that of an earlier study ¹² in which no correlation was found between tear film BUT and width of palpebral fissure. 

Tear film build-up time is a new clinical parameter introduced in this study. As yet, its clinical significance and the detailed mechanism of it are unclear. 

It has been found in earlier studies, both in vitro ⁹ and in vivo, ⁸ that the tear film covers the corneal surface in two steps. The first step is very quick: During opening of the eyelids the rising upper lid spreads the mucin and water layers of the tear film, and somewhat later the superficial lipid layer relatively slowly spreads over the surface from the lower to the upper part, bringing more water, which makes the tear film thicker. Subsequently, the tear film becomes thicker in the upper part, while continuously thinning in the lower part, ⁸ as was also found by Shimmura et al., ²¹ who performed tests after instillation of nonviscous aqueous artificial tear solution. It has been estimated that the spreading of the lipid layer takes approximately 2 to 3 seconds, ⁸ but, as far as we know, no direct measurements are available to date. 

Owens and Phillips ¹³ measured the displacement of tear film particles just after a blink and found an initially rapid upward movement (7.34 ± 2.73 mm/sec) that soon slows to zero velocity. The time to achieve zero velocity (tear stabilization time) was found to be 1.05 ± 0.30 seconds. These particles, which are thought to be accumulations of newly secreted lipid from the meibomian glands, protrude outward from the surface of the tear film. The tear stabilization time measured by Owens and Phillips reflects only the cessation of mass upward spreading of the tears—that is, the cessation of lateral movement of the protruding lipid particles. However, the particles are still present at this time (otherwise the velocity measurement could not be made). In their opinion, the protruding particles may cause perceptible distortion of a projected grid, which we interpret to mean that surface irregularities are still present that may affect the videotopographic measurements. We hypothesize that a more even surface produced by spreading of the lipid particles to produce a very thin lipid layer may occur in the subsequent time period. 

In our study we were able to follow the improvement in tear film surface regularity in the first several seconds after a blink for a longer period than the tear stabilization time. Our measurement, the tear film build-up time, relates to the regularity of the outermost layer of the tear film, the air–liquid surface of the lipid layer. Our findings may imply an even lipid and water layer of the tear film, but could equally reflect a build-up of an unevenly thick tear film but with a regular anterior surface that could compensate for some irregularities of the corneal surface. It is also possible that, as suggested earlier, the observed build-up time relates to the spreading of the superficial lipid layer from independent droplets to form an even surface. 

Possible factors involved in the tear film build-up time are the movement needed for the lipid layer to spread evenly over the outermost surface and adjustments in the thickness of the water and mucin layers of the tear film to compensate for the small irregularities in the corneal front surface. Our method did not give us insight into the mechanisms, but it clearly showed that the most even surface needs a certain time to build up after a blink. 

Possible reasons for the atypical cases, in which we did not observe signs of tear film build-up after a blink, might be the following: different age of the subjects, different tear-fluid composition or foam formation, initial reflex tearing, and error in the mathematical modeling (e.g., SAI for subject 11). Most of the atypical cases involved the youngest subjects in the study group, and they may have the most stable tear film (e.g., subject 7). Initial tearing may happen after opening the eyelids and forcing fixation and wide lid fissure while viewing the illuminated bright circles of the keratoscope. The flood of excess tear fluid may worsen the SRI and SAI, as in (for SRI) subjects 14 and 15 and (for SAI) subjects 10 and 12 (Fig. 2) , causing a rapid increase (deterioration) in the SRI or SAI. In these four subjects, after the maximum “spike,” a continuous decrease occurred, possibly due to the distribution and evaporation of the tear fluid. A previous study has found that tear film stability changes, even from one blink to another, because blinking does not necessarily spread the tear film uniformly on each occasion. ²² The tear film build-up time may also vary for each blink. Opening of the eyelids for a long period (4 minutes) may cause permanent dry-spot formation in certain cases, ²² which could also affect tear film build-up. In our study, the series of videotopographic images was in some cases recorded only after several trial openings of the lids for up to 30 seconds, which may lead to atypical results in some subjects. 

The clinical significance of the observed build-up time of the tear film in healthy eyes is not great in terms of visual acuity. An improvement in visual acuity of only a maximum of one Snellen line may be expected in the first 3 to 10 seconds after a blink, based on the amount of change in SRI and SAI, ^{7 23} and we believe that to date no subjective or objective recognition of this phenomenon has been reported. Some indirect data have been published in the measurements in a study by Tutt et al. ²⁴ who measured retinal vessel contrast and contrast sensitivity during periods of nonblinking and reported a noticeable decrease in optical quality of the tested eyes soon after a blink, possibly in connection with tear film break-up. However, looking at their detailed results, it seems that this decrease starts on average only at approximately 10 seconds after a blink. Although their presentation of results is not detailed enough to observe quick changes in the first few seconds after a blink, it is interesting that among the presented individual cases, there are some subjects who showed improved retinal vessel contrast at approximately 5 seconds after they blinked. This may correspond to improvement of optical quality after a blink, which would be consistent with our results, indirectly tending to confirm the existence and optical importance of the tear film build-up phenomenon. 

The tear film build-up and the changes in the SRI and SAI did not seem to influence the corneal power measurements, which means that, for keratometric evaluation, any period is equally suitable for data capture. However, for measurement of the SRI and SAI, the most suitable time in healthy eyes seems to be the period after the tear film build-up and before tear film break-up—on average, 5 to 7 seconds after a blink. However, it is difficult to arrive at a standard figure for the optimum time of measurement because of the great interindividual variability (Tables 1 2 3 4) . In principle, the ideal is to observe the subject’s individual trend line and establish the value and time of the optimum surface regularity. Further studies using the high-speed videotopographic method are needed to establish the clinical significance of these changes in the tear film after a blink. Such studies should measure the reproducibility of the tear film build-up time, both between different subjects and in a given subject at different times and under various conditions. Comparative studies with a longer observation period would also be useful, to find the correlation between the conventionally measured values of tear film BUT (invasive ¹² and noninvasive ²⁵ ) with the value determinable using high-speed videotopography. 

It has been reported that the tear mass topographic contour in patients with dry eye can be irregular but can be improved by use of artificial tears. ² In our patients with dry eye we observed these tendencies (Tables 3 4 ; Figs. 4 5 ); however, the small number of the patients in the present study and their heterogeneity did not allow us to statistically demonstrate either of these effects. In the preliminary examination of a few patients with dry eye, we wanted only to show that high-speed videotopography can be used to quantify tear film dynamics in tear-deficiency cases and can also be used to quantify how instillation of artificial eye drops may alter the tear film profile. We found that the tear film build-up phenomenon was observable even in the examined dry-eye cases of moderate severity. Further studies are needed to demonstrate typical differences between normal eyes and those with various pathologic ocular surface conditions. Also, further study could clarify the possible clinical role of high-speed videotopography in the diagnostic examination and follow-up of such patients. 

Our preliminary data on the effect of artificial tears suggest that different subjects may react quite differently to a given type of artificial tear fluid. In our study, we used only physiological saline eye drops. In several cases, the improvement of the surface regularity was meaningful; however, other subjects showed worsening of the tear build-up and surface regularity, which suggests that another type of artificial tear fluid may be needed. High-speed videotopographic examination may be helpful in trials to find suitable eye drops for a given patient. We are planning a prospective study to analyze the effect of different eye drops on tear film build-up in healthy and diseased eyes. High-speed videotopography may also have a place in clinical trials of different eye drops to exclude or disclose potential side effects affecting the tear film and may have a role in finding the optimal formulation of eye drops for different situations. 

 Presented in part at the annual meeting of the European Association for Vision and Eye Research, Palma de Mallorca, Spain, October 2000.

 Supported by Grant T034483 from the Hungarian Scientific Research Fund.

 Submitted for publication July 19, 2001; revised January 11, 2002; accepted January 30, 2002.

 Commercial relationships policy: N.

 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.

 

Subject 5, a typical healthy subject. Initial improvement and subsequent worsening of the surface regularity (SRI) in the first 15 seconds after a complete blink. Dotted line: measured data. Solid line: polynomial trend.

Table 1.

 

SRI and Other Tear Test Results in Healthy Subjects

Subject	Sex	Age (y)	BUT (sec)	Sch1 (mm)	Sch5 (mm)	iSRI	Minimum SRI (Mean)	t min (sec) (Mean ± SE)	Significance (P)
1	F	37	9	9	18	0.58	0.53	3.36 ± 0.31	0.0001
2	F	56	9	1	8	0.29	0.04	11.99 ± 1.23	0.0001
3	M	32	17	11	19	0.56	0.48	6.80 ± 0.80	0.0001
4	F	36	17	—	14	0.96	0.64	5.81 ± 0.47	0.0001
5	F	30	13	9	24	0.62	0.40	4.36 ± 0.24	0.0001
6	F	25	12	7	15	0.32	0.31	1.41 ± 2.88	NS
7	F	20	16	8	17	0.06	Very low and stable SRI throughout
8	F	48	11	3	19	0.55	0.18	5.93 ± 1.03	0.0001
9	F	27	29	7	13	0.87	0.47	8.02 ± 0.38	0.0001
10	F	21	15	10	22	0.51	Continuous increase to 6.5 sec, then stable
11	M	29	21	8	16	0.94	0.37	12.02 ± 0.76	0.0001
12	M	31	16	1	7	0.87	0.67	13.46 ± 1.82	0.0001
13	F	39	14	8	18	0.95	0.58	5.31 ± 0.46	0.0001
14	F	20	9	2	8	0.21	Maximum at 4.00 sec, then decrease
15	F	22	28	25	35	0.56	Maximum at 3.99 sec, then decrease
Mean		31.5	15.7	7.8	16.9	0.59	0.42	7.13
SD		10.4	6.2	6.0	7.1	0.29	0.19	3.87

 Initial SRI at 0.5 second after a complete blink (iSRI), the minimum SRI during the next 15 seconds, and the time (t _min) needed to reach the minimum. All data are for the SRI polynomial trend line. NS, nonsignificant; SE, standard error; SD, standard deviation; BUT, tear film break-up time; Sch1, Schirmer 1 test result at 1 minute; Sch5, Schirmer 1 test result at 5 minutes, —, no measurement.

Table 2.

 

SAI in Healthy Subjects

Subject	Sex	Age (y)	iSAI	Minimum SAI (Mean)	t min (sec) (Mean ± SE)	Significance (P)
1	F	37	0.40	0.35	7.47 ± 0.20	0.0001
2	F	56	0.20	0.15	6.55 ± 0.07	0.0001
3	M	32	0.46	0.36	3.97 ± 0.35	0.0001
4	F	36	0.74	0.60	5.31 ± 1.00	0.0001
5	F	30	0.42	0.41	1.94 ± 0.63	0.001
6	F	25	0.38	0.19	6.68 ± 0.28	0.0001
7	F	20	0.18	0.15	2.82 ± 0.33	0.0001
8	F	48	0.32	0.26	3.33 ± 0.50	0.0001
9	F	27	0.44	0.32	3.67 ± 0.26	0.0001
10	F	21	0.24	Maximum at 2.47 sec, then decrease and later increase
11	M	29	0.47	0.37	9.47 ± 0.84	0.0001
12	M	31	0.41	Maximum at 3.71 sec, then decrease
13	F	39	0.51	0.44	11.15 ± 2.23	0.0001
14	F	20	0.28	0.19	4.57 ± 0.52	0.0001
15	F	22	0.43	0.33	3.65 ± 0.16	0.0001
Mean		31.5	0.39	0.32	5.43
SD		10.4	0.14	0.13	2.72

 Initial SAI at 0.5 second after a complete blink (iSAI), the minimum SAI during the next 15 seconds, and the time (t _min) needed to reach the minimum. All data are for the SAI polynomial trend line.

Table 3.

 

SRI in Patients with Dry Eye before and after 1 Drop of Artificial Tear Fluid

Subject	Sex	Dx	Age (y)	Before Instillation							After Instillation
								iSRI	Minimum SRI (Mean)	t min (sec) (Mean ± SE)	Significance (P)	iSRI	Minimum SRI (Mean)	t min (sec) Mean ± SE	Significance (P)
16	F	pSS	72	1.35	0.83	3.70 ± 0.59	0.0001	0.28	0.04	12.95 ± 0.51	0.0001
17	F	sSS	55	0.52	0.31	4.31 ± 0.79	0.0001	0.50	0.44	2.62 ± 0.45	0.0001
18	F	pSS	56	0.24	0.02	3.04 ± 0.08	0.0001	0.24	Permanent increase
19	F	pSS	60	0.29	0.09	7.18 ± 0.59	0.0001	0.41	0.29	11.70 ± 1.32	0.0001
20	F	KCS	48	1.08	0.99	5.08 ± 1.55	0.001	0.84	0.69	2.78 ± 0.13	0.0001
21	F	KCS	55	0.36	Maximum at 2.66 sec, then decrease, then increase					0.24	0.07	6.68 ± 0.25	0.0001
22	F	KCS	64	1.08	Maximum at 3.27 sec, then stable					1.24	0.86	4.22 ± 0.46	0.0001
Mean			58.6	0.70	0.45	4.66		0.54	0.40	6.83
SD			7.7	0.45	0.44	1.60		0.38	0.33	4.52

 Initial SRI values at 0.5 second after blinking (iSRI), the minimum value during the next 15 seconds, and the time (t _min) needed to reach the minimum. All data are for the SRI polynomial trend-line. Dx, diagnosis; pSS, primary Sjögren syndrome; sSS, secondary Sjögren syndrome; KCS, keratoconjunctivitis sicca.

Table 4.

 

SAI in Patients with Dry Eye dry-eye patients before and after one drop of artificial tear-fluid

Subject	Sex	Dx	Age (y)	Before Instillation							After instillation of one drop of artificial tear-fluid
								iSAI	Minimum SAI (Mean)	t min (sec) (Mean ± SE)	Significance (P)	iSAI	Min. SAI (Mean)	tmin (sec) Mean ± SE	Significance (P)
16	F	pSS	72	2.99	0.50	3.83 ± 0.23	0.0001	0.28	Permanent increase
17	F	sSS	55	0.31	0.23	1.71 ± 0.56	0.0001	0.32	Maximum at 2.61 sec, then decrease, then increase
18	F	pSS	56	0.36	Permanent increase					0.24	Permanent increase
19	F	pSS	60	0.25	0.20	8.40 ± 1.08	0.0001	0.30	0.21	11.70 ± 1.32	0.0001
20	F	KCS	48	0.63	0.51	2.96 ± 0.32	0.0001	0.26	Permanent increase
21	F	KCS	55	0.23	Maximum at 3.78 sec, then decrease					0.28	0.21	6.68 ± 0.28	0.0001
22	F	KCS	64	0.77	0.56	8.20 ± 2.43	0.001	2.27	0.67	4.22 ± 0.46	0.0001
Mean			58.6	0.79	0.40	5.02		0.56	0.36	4.15
SD			7.7	0.99	0.17	3.09		0.75	0.27	0.71

 Initial SAI at 0.5 seconds after a blink (iSAI), the minimum during the next 15 seconds, and the time (t_min) needed to reach the minimum. All values are for the SAI polynomial trend-line. Dx = diagnosis: pSS = primary Sjögren syndrome, sSS = secondary Sjögren syndrome, KCS = keratoconjunctivitis sicca.

Figure 2.

 

Subject 14, an atypical healthy subject. Short initial stable period, followed by a quick increase (possibly due to reflex tearing) and subsequent decrease in the SRI. Dotted line: measured data. Solid line: polynomial trend.

Figure 3.

 

Incremental changes in corneal power in subject 13, a typical healthy subject. Each plotted point represents the difference between an observed K₁ (in diopters) and the immediately preceding K₁.

Figure 4.

 

Two series of videokeratoscopic images in a patient with dry eye (subject 16) before (A) and after (B) instillation of 1 drop of artificial tears. Note the severe distortion in the first image of the (A) series, which rapidly disappears. The (B) series is similar to that found in a typical normal eye. Time of image registration after the blink (t) and the corresponding measured data are shown above and below each image. The two dark dots at approximately the 12th ring on the horizontal meridian are generated by the apparatus. Only the central part of each original image is shown. PVA, potential visual acuity.

Figure 5.

 

Typical patient with dry eye (subject 20). Before application of artificial tear fluid, an initial improvement and subsequent stable period occurred with relatively high SRI in the first 15 seconds after a complete blink. Line A: measured data; line B: polynomial trend. After instillation of 1 drop of artificial tear fluid, initial improvement became more pronounced with lower measurements (with more regular surface) and subsequent worsening of the SRI, which became far higher than before the eye drop. Line C: measured data; line D: polynomial trend.

The authors thank Robert Bernard for comments on the manuscript and for helpful discussions.