Nikole L. Himebaugh, OD
Annette R. Wright, OD
Arthur Bradley, PhD, FAAO
Carolyn G. Begley, OD, MS, FAAO
Larry N. Thibos, PhD, FAAO

School of Optometry
Indiana University
Bloomington, IN


Abstract  |  Introduction  |  Methods  |  Results  |  Discussion  |  References


Purpose: The aim of the current study was to develop quantitative methods to assess optical aberrations caused by tear film disruption.

Methods: We used standard fluorescein imaging (FL) and a novel retroillumination (RI) method to image tear film disruption in 12 eyes. Using a clinical slit lamp biomicroscope, we alternated between wide-field blue and narrow beam white light to obtain an interleaved series of FL and RI images of the time course and pattern of tear film break-up. We develop an optical analysis which indicates that the RI image should be proportional to the spatial derivative of the FL image. Intensity fluctuations in the RI images are due to thickness changes in the tear film, whereas intensity fluctuations in FL images are directly determined by tear film thickness.

Results: As predicted by optical analysis of RI, the spatial distribution of gaps in the tear film seen with fluorescein appeared as pairs of light and dark contours in the RI images, and a precise correspondence between the spatial derivative of the FL image (slope) and the RI image was found. Both methods show a gradual spreading of the tear disruption during blink suppression that varies tremendously among eyes in both time and spatial pattern. Resumption of normal blinking did not produce an immediate reconstitution of the normal tear film, and areas of tear break-up created during blink suppression remained abnormal for up to several minutes of normal blinking.

Conclusions: Our analysis indicates that both FL and RI have the potential to quantify optical changes occurring during tear break-up. These results support an interpretation of RI as an intensity-based method for mapping the highly irregular optical aberrations of the eye produced by tear film disruption

Key Word: Tear film, aberration, retroillumination, fluorescein, tear break-up


The pre-ocular tear film is the most anterior refracting surface of the eye. With each blink, an optically smooth tear film is spread across the microscopically uneven surface of the cornea. However, during periods between blinks, the tear film may not be stable,1-4 especially in dry eye patients. It appears to thin or disrupt locally, a phenomenon clinically termed tear film break-up. These non-uniformities in thickness introduce irregularities in the air-tear interface. Since the change in refractive index at the air-tear interface is far greater (e.g. Dn=0.37) than at any other optical interface in the eye, irregularities in the anterior surface of the tears have the potential to significantly degrade the optical quality of the eye.

Psychophysical studies5-7 confirm that retinal image quality is degraded by prolonged periods of non-blinking. Also, physical measures of image quality6, 8 have demonstrated a significant reduction in retinal image quality after refraining from blinking either with or without soft contact lenses. Recently, wavefront sensing technology has been used to document a significant increase in higher order aberrations and scatter after prolonged periods of non-blinking.9-11 Finally, clinical studies12-17 have reported a wide range of visual symptoms (reduced visual acuity, blurry, disturbed or fluctuating vision) consistent with degraded retinal image quality in dry eye patients and contact lens wearers.

Although theory and experimental observations both show that the disruption of the human pre-corneal tear film can and does have a detrimental effect on retinal image quality, little is known about the precise nature of the tear film changes that lead to image degradation. This is due to the complexity of measuring a dynamic thin film and the paucity of suitable experimental techniques to study this system. A standard clinical technique for evaluating tear film break-up involves instilling sodium fluorescein dye into the eye. The dye diffuses into the tear film, and fluoresces green when irradiated with a blue light. Decreases in tear film thickness produce decreases in fluorescence intensity. Clinicians typically use the fluorescein method to measure tear break-up time (TBUT), which is the time from the last blink to the first appearance of a dark spot (i.e. absence of fluorescence). The fluorescein method has also been used to monitor the time course and spatial patterns of tear film disruption during periods of non-blinking.18, 19 Although fluorescein is a simple and convenient method for documenting changes in tear film stability, it is invasive. The instillation of fluid and dye may alter the composition, volume, and thickness of the tears, which can influence tear stability and break-up.20 Interpreting fluorescein patterns is potentially problematic in that fluorescence intensity is influenced by more factors than tear film thickness (e.g. fluorescein concentration and irradiating source characteristics).21 In addition, the fluorescein method cannot be used to assess the tear film in patients wearing soft contact lenses because the lens absorbs fluorescein dye.

The aim of the present study was to evaluate retroillumination (RI) as an optical method for monitoring changes in tear film thickness during periods of non-blinking. We have employed the RI technique previously to visualize the spatial pattern of refractive changes induced by thinning of the tear film and to correlate those patterns with aberration maps and fluorescein maps of tear film thickness.6, 9, 11 Here we extend those observations by first developing and then testing experimentally a quantitative model of the RI image and its dependence on variation in tear film thickness. The comparison between RI and fluorescein images support an interpretation of RI as an intensity-based method for mapping the highly irregular optical aberrations of the eye produced by tear film disruption.


Principle of the RI technique

RI is commonly used by clinicians to observe changes in opacification of the crystalline lens. RI can be achieved by back-illuminating the pupil with reflected light from the fundus, resulting in a “red reflex.” The RI technique used in the current experiment is a modified version of the clinical technique in which the tear film (instead of the crystalline lens) is imaged to examine refractive changes (rather than density changes) that occur in the tear film during periods of non-blinking. The technique has many optical similarities to the Foucault edge-test22 and, more specifically, to eccentric photorefraction.23-26

The RI technique, shown schematically in Fig. 1, was described briefly by Tutt et al.6 An external light source uniformly illuminates a patch of retina which extends from point A to point B. Reflected light from this patch which passes through the eye’s pupil and enters the camera lens is focused in a plane conjugate to the retina, but then continues on to the film plane of the camera to produce a blurred image of the retinal source. Since the film plane is optically conjugate to the tear film, which is located only a few millimeters anterior to the eye’s iris, the image captured by the camera is a patch of light nearly the same size and shape as the eye’s entrance pupil. The key to understanding the technique is to understand the distribution of light within this circular patch and how this distribution is related to the presence of optical imperfections of the tear film. We argue below that the intensity of the RI image corresponding to any given point on the cornea is determined by the proportion of rays passing through that corneal point which enters the camera aperture.


Figure 1. Schematic diagram of the retroillumination technique. (A) A light source centered on the camera axis uniformly illuminates a patch of retina extending from point A to point B. Light reflected out of the eye from point A is indicated by dashed rays and light reflected from point B is indicated by solid rays. The portion of the cornea extending from R to U is thus retro-illuminated by multiple cones of light emanating from each point in the extended retinal source. In this example, all of the rays emerging from the cornea are captured by the camera to produce a uniformly illuminated RI image (see insert). (B) An eccentric light source causes some reflected rays to miss the camera aperture, resulting in a non-uniform intensity profile across the RI image (see insert). (C) A gap in the tear film alters the number of rays entering the camera aperture, thus producing local intensity fluctuations in the RI image (see insert).

Consider first the configuration shown in Fig. 1A in which the light source is centered on the camera axis. Each point in the extended retinal source illuminates the cornea with a cone of rays that pass through the eye’s pupil. Two such cones are depicted in Fig. 1 to illustrate how each point R, S, T, U on the cornea is illuminated by a multitude of rays, each with a different angle of incidence. If all of the light rays emerging from every point on the cornea are captured by the camera’s entrance aperture then the resulting RI image of the tear film would be uniformly illuminated from R’ to U’ (see insert in Fig. 1A). In practice, the light source is positioned eccentric to the entrance pupil of the camera as shown in the configuration of Fig. 1B. This lateral positioning of the light source relative to the camera aperture produces an eccentric retinal image. Light reflected from this retinal source again retro-illuminates the cornea from R to U, but now some of the rays fail to enter the entrance aperture of the camera. As illustrated in Fig. 1B, all of the rays exiting the cornea at U on the side opposite the eccentric light source will enter the camera aperture, but none of those exiting the cornea at R on the same side as the eccentric light source will enter. For any point S or T lying between U and R, a portion of the rays will enter the camera aperture depending on how close each corneal point is to the lower and upper extremes. Thus, the proportion of rays exiting the cornea and entering the camera aperture will increase monotonically across the pupil resulting in a smooth intensity gradient across the image of the tears in the camera. For an extended light source that produces a defocused retinal image, the intensity gradient will be a linear ramp26, which may be modulated by any directionality in the retinal reflection (a reverse Stiles Crawford Effect).27

Consider now the effects of tear film disruption on the RI image. In Fig. 1C we present a simple model of tear film break-up as a local gap or thinning of the tear film resulting in a pair of opposite sign prisms (see insert in Fig. 1C). Rays exiting the eye at the prisms will be bent around the base of the prisms, causing the local ray bundles to be deviated either towards or away from the entrance pupil of the camera, thus either increasing or decreasing the proportion of each ray bundle entering the camera aperture. Figure 1C shows the rays exiting at S being deviated toward the base of the prism and away from the camera aperture. As a result of this deviation, no rays exiting from point S in this example enter the camera system. This produces a darker than average local area in the corresponding point of the RI image. Conversely, rays exiting the eye through point T will be deviated around the base of the oppositely oriented prism, deviating toward the camera aperture and resulting in a lighter than average local area in the RI image at this point. In this way, optical deviations of the exiting ray bundle produced by local variations in tear film thickness create dark and light zones or intensity fluctuations in the RI image. The sign of the intensity change (darker or brighter) indicates the sign of the slope of the tear film profile. For example, as shown in this schematic, a gap in the tears would create a pair of dark and light contours, dark on the side of the eccentric light source.


Twelve normal subjects, 9 females and 3 males, ages 24-45, participated in this study. The study followed the tenets of the Declaration of Helsinki and was approved by the Institutional Review Board at Indiana University. Informed consent was obtained from all subjects.

A 20SL Zeiss biomicroscope and video system was used for both the RI and fluorescein techniques. To obtain the RI image of the full pupil, the light source on the biomicroscope was positioned immediately adjacent to the aperture of the channel containing the video recorder. To displace the very bright 1st Purkinje image as far from the pupil center as possible, the narrow slit of white light was positioned at the edge of the subject’s pupil.

The traditional clinical fluorescein (FL) method was used to visualize tear film thickness. A sterile fluorescein strip (Barnes-Hind 0.6mg Ful-Glo® sterile strip) was moistened using non-preserved saline, shaken once to remove excess fluid, and applied to the inferior bulbar conjunctiva. The tear film break-up was observed in the biomicroscope (8x magnification) using a wide beam illumination with a cobalt blue light source (approximately 30 degree beam angle) viewed through a Wratten #8 filter.

To maximize the central area of the tear film observed with RI during the experiment, one drop of 0.5% tropicamide was instilled into the eye to dilate the pupil. Both eyes were anesthetized with 0.5% proparacaine to minimize discomfort and to prevent reflexive tearing and blinking. This was necessary to observe the spatial and temporal characteristics of tear film break-up beyond the first area of disruption which would normally initiate a blink. Prior to testing, sodium fluorescein was instilled into the eye being observed.

The subject’s head was positioned and stabilized using a standard biomicroscope head and chin rest. Subjects were asked to blink three to four times and hold their eyelids open as long as they could comfortably do so (1-3 minutes). Each trial began with the RI view of the pupil. When obvious tear film disruptions (intensity fluctuations) were observed in the RI image, the lighting was changed to observe tear film fluorescence. Tear film break-up was then alternately monitored by the two techniques at approximately fifteen-second intervals. The spatial and temporal characteristics of tear film break-up were captured with a CCD video camera (Sony DXC-107AP, Tokyo, Japan) and recorded at 30 frames/second on an S-VHS video recorder (Mitsubishi HS-U69, Mitsubishi Electronics America, Cypress, CA) with a video image resolution of 720 x 486 for later display and analysis. A minimum of three pairs of RI and FL images were obtained during each trial. At the end of each trial, the tear film was monitored for up to 5 minutes during normal blinking.

Single frame FL and RI images were obtained by digitizing the recorded video using Adobe Premier 4.2 (Adobe Systems, Inc., Mountain View, CA). A Matlab (Mathworks, Inc., Natik, MA) computer program was used to align and mask the images to allow for direct comparison of only the area within the pupil.


Figure 2 shows an example of a time series of images collected for one eye during one period of non-blinking. The images have been masked to show only the pupil. The central bright spot in the FL images and the vertical bright strip and two bright spots in some of the RI images are the 1st Purkinje image artifacts. Immediately after the blink (t=0), the RI image of the tear film is a uniform intensity ramp. However, after only 5 seconds of non-blinking, intensity fluctuations in the inferior pupil are apparent in the RI image indicating that a disruption of the tear film has occurred. The FL image recorded 11 seconds later shows a large irregular area lacking fluorescence inferiorly. Areas lacking fluorescence (dark areas) are interpreted as local gaps or thinning of the tear film. With increasing time after the blink, the light/dark intensity fluctuations in the RI image (t = 35 and 70 sec) spread across the entire pupil. The interleaved FL images recorded 50 seconds and 87 seconds after the blink show little remaining fluorescence in the lower two thirds of the pupil indicating continued break-up of the tear film. There are thin, vertical strips of fluorescence remaining across the central cornea. The area of fluorescence present superiorly at t=50 seconds is due to a partial blink between t=35 and 50 seconds. The larger area of fluorescence superiorly at t=87 seconds is due to another partial blink between the FL image recorded at t=50 seconds and the RI image recorded at t=87 seconds. The pattern of intensity fluctuations in the RI image recorded 35 seconds after the blink exhibited clear similarities in spatial structure to the intensity fluctuations in the FL image at t=16 seconds. Also, the spatial pattern of intensity fluctuations in the RI image recorded at t=70 seconds are very similar to the vertical strips of remaining fluorescence in the FL images recorded before (t=50) and after (t=87) this RI image.

Figure 2. Time course and spatial topography of tear film break-up for subject AL obtained during a 90 second period of non-blinking. The time immediately after a blink is t=0 sec. R=retroillumination image; F=fluorescein image.

Additional examples of paired images (RI on the left and FL on right) from 4 other subjects are shown in Fig. 3. Although spatial patterns of thinning and break-up of the tear film were highly idiosyncratic among subjects, pairs of RI and FL images obtained from the same eye share the same spatial structure. Tear film thinning or small, localized areas of tear break-up occurred in the eyes of five subjects (e.g. Fig. 3A). However, seven subjects showed large areas of break-up that extended across a large portion of the pupil (e.g. Fig. 3C & D).

Several features of the RI and FL images in Fig. 3 confirm the model predictions outlined in the methods. First, the three oblique dark lines in the FL image (Fig. 3A) indicate three gaps in the tear film. In the same location within the RI image are three pairs of dark/light bands. On the light source side of gap in the tears (right in this image) local prismatic deviation has reduced the amount of light entering the slit lamp, and conversely on the opposite side of the gap in the tears to the eccentric light source RI intensity is increased. Second, from the optics of RI we predict image intensity modulation due to changes in the slope of the tear film, but no such modulations if the tear film is uniformly thick or uniformly absent. This prediction is borne out in Fig. 3B, where the FL image indicates tears present in the central lower pupil and generally absent in the upper third, but the RI image has the same intensity in both regions. However, fluctuations in the RI image are seen at the transition between these two zones. Another example of this is shown in Fig. 3C. Third, if there is a local presence of tears (as opposed to a local absence), the light/dark banding in the RI image should have the opposite phase (light on the side of the eccentric source) to that seen with a gap. A good example of this can be seen in Fig. 3D where most of the cornea lacks tears.

Figure 3. Qualitative comparison of tear film break-up monitored with RI and FL for four subjects (A) CM, (B) DC, (C) DR, and (D) AW. All images were obtained while the subject refrained from blinking for an extended period of time. The time immediately after a blink is t=0 sec. R=retroillumination image; F=fluorescein image.

The qualitative correspondence seen in the RI and FL images is not surprising since both methods are monitoring the same tear break-up. However, before comparing the two images quantitatively, it is worth reassessing the precise theoretical relationship between these two types of images. For example, if we assume that reductions in fluorescence intensity only reflect reduced tear film thickness, and the relationship between fluorescence intensity and tear film thickness is linear which it should be for such a thin film,21 then intensity of the FL image will be directly proportional to tear film thickness. The constant of proportionality is determined primarily by the concentration of FL, the irradiance and wavelength of the illuminating light source, and the fluorescence efficiency of the fluorescein.21

We can now infer the optical impact of tear film break-up since reduced tear film thickness will produce a reduced optical path-length in the tears, thus producing an exiting wavefront with advanced phase.11 Therefore, if variations in fluorescence are due solely to variations in tear film thickness, then the pattern of fluorescence intensity should be directly proportional to the wave-front deviations introduced by tear film break-up. The slope of this wavefront corresponds to the ray deviation present at each location in the tear film.11 According to our model of RI described in the methods, intensity fluctuations in the RI images are also determined by ray deviation produced by changes in tear film thickness or slope of the tear film surface. Therefore, the spatial derivative of the FL images along the x direction should mirror the RI intensity fluctuations in the horizontal meridian. That is, we expect dIF/dx = kIR, where IF = intensity of the FL image and IR = intensity of the RI image.

Figure 4. Quantitative comparison of differentiated FL images (solid line, left ordinate) and RI images (dashed line, right ordinate) for subjects (A) & (B) CM and (C) AL. White lines in image above each graph show area of tear film used for analysis. R=retroillumination image; F=fluorescein image.

We have compared the differentiated FL images to those obtained with the RI method. Three samples are shown in Fig. 4. Images used for analysis in Fig. 4A and 4B are from subject CM (see Fig. 3A) and were chosen to represent discrete areas of tear break-up occurring shortly after a blink. Images used for analysis in Fig. 4C are from subject AL (see Fig. 2 t=70 sec and t=87 sec) and were chosen as an example of tear break-up that occurred after an extended period of non-blinking. In this example there are local regions of tears within large areas free of tears. The areas used in analysis are indicated in the Fig. 4 inserts by white lines. In general, the agreement is excellent. The number, location, sign and magnitude of the intensity fluctuations in the RI images correspond almost perfectly with the differential of the corresponding FL images. Such close agreement is remarkable given that the RI and FL images were not obtained at precisely the same time, and there are many unknown parameters of the slit lamp optics that will determine the exact relationship between RI intensity and local slope of the tear film.

Figure 5. Quantitative comparison of FL images (solid line, left ordinate) and integrated RI images (dashed line, right ordinate) for subjects (A) & (B) CM and (C) AL. White lines in image above each graph show area of tear film used for analysis. R=retroillumination image; F=fluorescein image.

The above analysis demonstrates that the local slope of the tear film can be measured directly via RI or calculated by differentiation of the FL images. Based upon the same arguments, it is possible therefore to integrate the RI data to obtain a measure of tear film thickness, which should correspond directly to the experimental measure of thickness obtained from the FL images ( IR * dx = kIF). Prior to integration along the x dimension (the eccentric light source was decentered horizontally), we isolated changes in the RI image due to tear thinning by subtracting the t=0 image from the RI image after break-up has occurred. The same three tear films that were compared in Fig. 4 are also compared in Fig. 5, but this time we compare the experimental measures of tear film thickness obtained with FL to those computed by integration of the RI images. Again, the correspondence between the two methods is striking, again confirming the underlying relationship between these two methods. In each case, the gap in the tears measured directly by FL is observed in the integrated RI data. Thus both methods can be used either to assess tear film thickness or local slope of the tear film.

Figure 6. Recovery of tear film with normal blinking for subjects (A) AL and (B) CM. Images were obtained after normal blinking was resumed. Arrows in fluorescein image indicate areas of fluorescein staining of the corneal epithelium. R=retroillumination image; F=fluorescein image.

In addition to monitoring the break-up of the tear film during periods of non-blinking, we have also used FL and RI to monitor the tear film during a recovery period after blinking has been resumed. If only thinning or small areas of tear film break-up occurred, a few (sometimes even one) normal blinks were sufficient to re-establish uniform fluorescence and a regular RI image free of local intensity modulations (not shown). However when large areas of tear break-up occurred (e.g. Fig. 2), the FL and RI images did not return to baseline immediately with normal blinking. Two examples of RI and FL images obtained after blinking had been resumed are shown in Fig. 6. The RI and FL images in Fig. 6A were from the same experimental trial which generated the time series shown in Fig. 2. After 1 minute of normal blinking, the tear film had not returned to its pre-trial state. The pattern of break-up seen prior to blinking (e.g. see the lower panels in Fig. 2) is also manifest after blinking (Fig. 6), but it lacks the fine detail seen prior to the blinks. This result clearly shows that the pattern of break-up created by blink suppression continues to manifest itself after blinking has been resumed. This suggests that the tear film is not wetting the surface of the cornea uniformly immediately after normal blinking is resumed. In the second example shown in Fig. 6B, there is fluorescein staining on the inferior cornea where tear break-up occurred during the period of non-blinking (see Fig. 3A). The RI image shows tear film disruption corresponding to the areas of epithelial staining seen in the FL image.

Figure 7. Time course of tear film break-up and recovery. (A) Images obtained during 58 second period of non-blinking. The time immediately after a blink is t=0 sec. R=retroillumination image; F=fluorescein image. (B) Images obtained after normal blinking was resumed. Fluorescein image at t=0 sec (blinking resumes) shows fluorescein staining superiorly (bright arc) corresponding to initial tear break-up (arrow) in the RI image at t=12 sec.

A series of RI and FL images collected from one eye over a five minute period are shown in Fig. 7. Blinks were suppressed for the first 56 seconds, and during this time similar patterns can be seen in the FL and RI images characteristics of tear film break-up. The subject first blinked at 58 seconds, and we continued to monitor this eye during four minutes of normal blinking to examine the time course and spatial characteristics of the tear film recovery. Two interesting observations can be seen during the recovery period. First, along the arc-shaped gap that appeared initially in the tear film in the upper half of the pupil (RI at t=12 sec and FL at t=22 sec) there was corneal fluorescein staining seen in the FL image immediately after blinking resumed (see Fig. 7B, t=0 sec). The staining clearly shows that some epithelial damage occurred along this gap in the tears. The fluorescein staining was still present at the end of the 4 minutes of normal blinking monitored in this study. More surprising perhaps is the observation that the same arc can be seen in the post blink RI images up to about three minutes after the blink. This suggests that once epithelial damage has occurred, even after multiple blinks, either the tear film or corneal surface is not smooth over the damaged site. This intensity modulation in the RI image finally disappeared after four minutes of blinking even though epithelial staining was still visible. These results reveal a potential advantage of RI over the conventional FL technique. If fluorescein staining is present on the surface of the cornea, it is difficult to assess the stability of the tear film overlying this area using FL.


In this study, we determined the relation between two methods for examining tear film disruption, retroillumination and sodium fluorescein. We were able to argue on theoretical grounds that the two methods should produce highly correlated images, the RI being the spatial derivative of the FL since RI intensity is determined by local slope of the tear film and FL intensity by tear film thickness. Both qualitative and quantitative analysis confirmed this theoretical model.

Although we have described our RI and FL results in optical terms, there are several important experimental constraints that limited our ability to use the RI or FL images to compute optical quality of these eyes. As discussed by Himebaugh et al.9 and Thibos and Hong,11 standard optical techniques such a Shack-Hartmann (SH) aberrometry can be used to measure the slope of the wavefront exiting the eye and by integration compute the actual wavefront from which image quality can be calculated. However, due to limitations in our ability to calibrate the FL and RI methods, we only know that the FL intensity is proportional to the wavefront deviation, and the intensity of the RI is approximately proportional to the slope of the exiting wavefront. In order for these methods to be employed to assess the full optical impact of tear film disruption, we must be able to identify these constants of proportionality by controlling such parameters as FL concentration, irradiation spectrum and intensity, geometry of the RI optical system and exiting beam intensity in the RI method.

The current RI method only employed a fixed single eccentric light source located at the horizontal edge of the slit lamp aperture. Because of this, the RI method can only measure the horizontal dimensions of any tear film thickness changes. In order to get a complete two-dimensional map of the local slope changes two orthogonal light sources are required. Also, in order to compare the RI and FL images more completely it is essential to collect each image simultaneously rather than in an interleaved series such as we used in this study. With such modifications, we may discover an exact correspondence between the differentiated FL image and the RI image. Also, we would be able to quantify in absolute terms the slope of the wavefront exiting the pupil and the wavefront deviation. One major difference between these two methods is that the FL method reflects only the aqueous portion of the tears, but the exiting beam of the RI method is affected by the optics of the entire tear film and eye.

We have also used the Shack-Hartmann aberrometer to measure the exiting wavefront during tear film disruption.9 Although this approach has the advantage of being well established, theoretically sound, and sufficient to compute retinal image quality, it is limited in resolution by the size and thus density of the small lenslets used to assess local slope of the exiting wavefront. For most typical aberrations seen in human eyes,28, 29 low sampling densities are adequate to capture the generally gradual changes in wavefront errors that occur across the pupil when tears are intact. However, as can be seen from Figs. 2 and 3, tear film break-up can be highly localized and create a spatially detailed pattern that could not be adequately sampled by a SH system. In such cases, the high sampling density available with the FL or RI methods (e.g. in our examples, we have more than 10,000 pixels in each pupil image) might prove an advantage.

In this study, we also examined the temporal and spatial characteristics of tear film break-up during blink suppression and recovery after blinking resumed. Tear film break-up occurred in all subjects during the extended period of non-blinking recorded in our experiment. The tear film break-up patterns were highly idiosyncratic varying in both their spatial and temporal properties between eyes but generally showing a progressive deterioration during blink suppression. The extent of tear film break-up ranged from very small and localized dots or streaks to geographic areas covering a large percentage of the cornea. In order to compare the FL and RI techniques and observe the tear break-up beyond the first area, we instilled mydriatics, anesthetic, and sodium fluorescein into the eye. All three pharmaceutical agents may have affected the tear film stability and thus the time course, distribution, and extent of tear break-up.

When normal blinking was resumed, immediate recovery of the tear film did not occur for all subjects (Fig. 6 & 7) and some features of the break-up pattern lingered on in the tear film for several minutes after blinking was resumed. One hypothesis to explain this lingering effect is that prolonged periods of blink suppression not only cause a redistribution of the aqueous portion of the tears, but also dry the tear mucins. If drying does occur, optical quality may not return to normal until the mucins have re-hydrated. Alternatively, either dried mucins and/or damaged epithelial cells may have decreased affinity for water producing poor wettability of the corneal surface.

Retroillumination has long been used by clinicians to examine the crystalline lens for cataracts. The technique requires only standard clinical equipment and could be used as a non-invasive clinical method for assessing the quality of the tear film as well as the optical effects of tear film break-up. In our study, RI was an invasive technique because we dilated the pupils of our subjects. However, if an infrared light source is used, under dark conditions the pupil will dilate naturally resulting in RI of a large area of the precorneal tear film, but limited by the size of the pupil. Another advantage of RI is that it can be used to assess the tear film in patients wearing soft contact lens. Further development of the non-invasive RI technique would be very beneficial for assessing the tear film in dry eye patients as well. Tear break-up time is often used to assess tear quality and aid in the diagnosis of dry eye. However, instillation of fluorescein into the eye may alter the dynamics of the tears, which can influence tear stability and break-up. With the correct infrared light source or filter, RI could be used non-invasively to assess tear film break-up and recovery with or without a contact lens.

In conclusion, RI may be used clinically in the diagnosis and management of dry eye, particularly in the assessment of the tear film in soft contact lens wearers. With further development, this technique also has the potential to be used measure optical quality and quantify the optical effects of tear film break-up.


We gratefully acknowledge the support of Xin Hong and Kevin Haggerty for computer programming used in data analysis. This project was supported by NEI Grant R01-EY05109 (LNT) and a Bausch & Lomb/American Optometric Foundation William C. Ezell Fellowship (NLH).

Presented, in part, at the Annual Meeting of the American Academy of Optometry, December 1999.


  1. Norn MS. Desiccation of the precorneal film: Corneal wetting-time. Acta Ophthalmol 1969;47:865-80.

  2. Korb DR, Baron DF, Herman JP, Finnemore V.M, Exford JM, Hermosa JL, Leahy CD, Glonek T, Greiner JV. Tear film lipid layer thickness as a function of blinking. Cornea 1994;13:354-9.

  3. Rengstorff RH. The precorneal tear film: breakup time and location in normal subjects. Am J Optom Physiol Opt 1974;51:765-9.

  4. Mengher LS, Bron AJ, Tonge SR, Gilbert DJ. A non-invasive instrument for clinical assessment of the pre-corneal tear film stability. Cur Eye Res 1985;4:1-7.

  5. Thai LC, Ridder WH, Tomlinson A. The effect of tear film break-up on contrast sensitivity. Invest Ophthalmol Vis Sci 2001;42:S596.

  6. Tutt R, Bradley A, Begley C, Thibos L. Optical and visual impact of tear break-up in human eyes. Invest Ophthalmol Vis Sci 2000;41:4117-23.

  7. Timberlake G, Doane M, Bertera J. Short-term, low contrast visual acuity reduction associated with in vivo contact lens drying. Optom Vis Sci 1992;69:755-60.

  8. Albarran C, Pons AM, Lorente A, Artigas JM. Influence of the tear film on optical quality of the eye. Contact Lens and Anterior Eye 1997;20:129-35.

  9. Himebaugh N, Bradley A, Begley C, Thibos L. The effect of tear film break-up on the optical aberrations and light scatter of the eye. Invest Ophthalmol Vis Sci 2002; submitted.

  10. Koh S, Maeda N, Hori Y, Nishida K, Watanabe H, Inoue Y, Tano Y, Mihashi T. The effect of tear film break-up on higher-order aberrations measured with wavefront sensor. Invest Ophthalmol Vis Sci 2001;42:S900.

  11. Thibos LN, Hong X. Clinical applications of the Shack-Hartmann aberrometer. Optom Vis Sci 1999;76:817-25.

  12. Begley CG, Caffery B, Nichols KK, Chalmers R. Responses of contact lens wearers to a dry eye survey. Optom Vis Sci 2000;77:40-6.

  13. Toda I, Fujishima H, Tsubota K. Ocular fatigue is the major symptom of dry eye. Acta Ophthalmol 1993;71:347-52.

  14. Bjerrum KB. Test and symptoms in keratoconjunctivitis sicca and their correlation. Acta Ophthalmol Scand 1996;74:436-41.

  15. Vitali C, Moutsopoulos HM, Bombardieri S. The European community study group on diagnostic criteria for Sjögren's syndrome. Sensitivity and specificity of test for ocular and oral involvement in Sjögren's syndrome. Ann Rheum Dis 1994;53:637-47.

  16. Lee SH, Tseng SCG. Rose bengal staining and cytologic characteristics associated with lipid tear deficiency. Am J Ophthalmol 1997;124:736-50.

  17. Goto E, Shimmura S, Yagi Y, Tsubota K. Decreased visual acuity in dry eye patient during gazing. Invest Ophthalmol Vis Sci 1998;39:S539.

  18. Bitton E, Lovasik JV. Longitudinal Analysis of Precorneal Tear Film Rupture Patterns. In: Sullivan DA, Dartt DA, Meneray MA, ed. Lacrimal Gland, Tear Film, and Dry Eye Syndromes 2 : Basic Science and Clinical Relevance, vol. 438. New York: Plenum Press, 1998:381-9.

  19. Begley C, Renner D, Himebaugh N, Guo Y, Wilson G. Tear break-up during eye opening in dry eye and normal subjects. Optom Vis Sci 2001;78:121.

  20. Mengher LS, Bron AJ, Tonge SR, Gilbert DJ. Effect of fluorescein instillation on the pre-corneal tear film stability. Cur Eye Res 1985;4:9-12.

  21. Szabo A. Fluorescence principles and measurement. In: MG G, ed. Spectrophotometry and spectrofluorimetry-A practical approach, New York: Oxford University Press, 2000:33-67.

  22. Geary J. Introduction to Wavefront Sensors. vol. TT18. O' Shea D ed. Bellingham: SPIE Optical Engineering Press, 1995.

  23. Bobier W, Braddick O. Eccentric photorefraction: optical analysis and empirical measures. Am J Optom Physiol Opt 1985;62:614-20.

  24. Howland H. Optics of photoretinoscopy: results from ray tracing. Am J Optom Physiol Opt 1985;62:621-5.

  25. Roorda A, Campbell M, Bobier W. Geometrical theory to predict eccentric photorefraction intensity profiles in the human eye. J Opt Soc Am A 1995;12:1647-56.

  26. Roorda A, Campbell M, Bobier W. Slope-base eccentric photorefraction: theoretical analysis of different light source configurations and effects of ocular aberrations. J Opt Soc Am A 1997;14:2547-56.

  27. Marcos S, Burns S. Cone spacing and waveguide properties from cone directionality measurements. J Opt Soc Am A 1999;16:995-1004.

  28. Thibos LN, Hong X, Bradley A, Cheng X. Statistical variation of aberration structure and image quality in a normal population of healthy eyes. J Opt Soc Am A 2001;submitted:

  29. Porter J, Guirao A, Cox I, Williams D. The human eye's monochromatic aberrations in a large population. J Opt Soc Am A 2001;18:1793-803.