Confocal Microscopy of the Cornea

3. Evolution of biomicroscopy of the eye

3.1 The development of the ophthalmoscope
3.2 The development of the slit lamp
3.3 The development of the specular microscope
3.4 The development of the clinical confocal microscope
3.5 Practical requirements for a clinical confocal microscope

3. EVOLUTION OF BIOMICROSCOPY OF THE EYE

3.1 The Development of the Ophthalmoscope

The mathematician Babbage, who is well known for his work on the computer, designed a direct view ophthalmoscope based on a glass mirror with a central hole (Duke-Elder, 1962; Masters, 1994b). The world still had to wait five additional years for the genius Helmholtz, who in 1851, introduced a practical device to permit the clinician to observe the living retina (Helmholtz, 1851, 1924). The Helmholtz ophthalmoscope was based on a stack of glass plates which were set at an angle in order that light from an external source could be reflected into the eye. In 1852 Ruete used a slivered concave mirror and a strong convex lens to form an indirect ophthalmoscope for observing the retina (Ruete, 1852). In 1937 Goldmann modified the slit lamp to obtain optical sections of the retina and thereby determined the depth of retinal and subretinal opacities.

In 1949 Ridley pioneered the development of the television ophthalmoscope. Ridley’s invention of a scanning spot ophthalmoscope is based on a cathode ray tube which served as a scanning point source of light for retinal illumination (Ridley, 1952). This concept was implemented in the modern development of the scanning laser ophthalmoscope. Ridley’s contribution to point scanning of the retina is discussed further in Sec. 3.4.

3.2 The Development of the Slit Lamp

The development of the slit lamp, a microscope that uses oblique illumination and microscopic observation, provided oblique sectioned views of the cornea, the ocular lens, and the retina in the living eye (Berliner, 1966; Tate and Safir, 1991). The slit lamp is a long working distance microscope for observation of the living eye. A slit of light from a lamp, hence the name, is projected onto the cornea, the lens, or the retina. A viewing microscope with a long working distance objective is focused on the same focal region as the image of the illuminated slit. The angle between the observation microscope and the viewing optical system may be varied to maximize the contrast of the ocular image.

Gullstrand received the Nobel prize in 1911 for his work on the dioptrics of the eye. In 1911 Gullstand presented the slit lamp as a practical clinical instrument to the Versammlung der deutschen Ophthalmologischen Gesellschaft in Heidelberg (Gullstrand, 1911a 1911b). The original apparatus devised by Gullstrand required the clinician to hold an illuminating lens in one hand and an single or binocular loupe in the other hand.

Henker mounted the Gullstrand illumination system on an articulated arm; a headrest and the Czapski microscope were mounted to a table. In 1916 Zeiss manufactured a Czapski slit lamp which contained a Gullstrand slit lamp mounted on Henker’s articulated arm and a binocular microscope. Vogt was the first to use the slit lamp for the examination of the corneal endothelium using the slit lamp in the specular reflection mode. In 1933 Goldmann developed a slit lamp in which the vertical and horizontal adjustments of both the lamp and the slit beam were mounted on a single mechanical stage (Goldmann 1938, 1940). This instrument with the addition of a joy stick to position the microscope with respect to the patient’s eye was marketed by Haag-Streit company as model 360 slit lamp.

3.3 The Development of the Specular Microscope

The specular microscope is a reflected light microscope based on the specular reflection at the interface between the endothelium and the aqueous humor. The large difference in refractive index between the cornea and the aqueous humor results in a large reflectivity at this interface. The conditions for specular reflection are that the angle of incidence and the angle of reflection are equal. A light source and the observer can arranged to view this specular reflection. This method was first used by Vogt to image the corneal endothelium. Vogt observed the cell borders of the corneal endothelial cells. The contrast which permitted the observation of the cell borders is the difference of reflectivities of the cell borders and the endothelial cells themselves. These observations are remarkable since these cells are semi-transparent, four microns thick, and located 500 microns below the surface of the cornea. Vogt also observed the epithelial cells on the ocular lens. More recently, a new contact glass has been described for observing the specular reflection of the corneal endothelial cells with slit-lamp examination (Eisner et al., 1985).

The slit lamp suffers from the fact that there is a shallow depth of field and that the reflectivity of the interior cornea is very weak. However, the reflection from the anterior and posterior surfaces are much larger than the internal reflections. Goldmann offered a clever solution to this problem (Goldmann, 1940). His modification of the Gullstrand slit lamp used a photographic system which moved on the optic axis. Both the camera and slit beam mechanically moved forward during the exposure, at the same time the film traverses synchronously behind the slit, therefore, the image on the film is maintained in continuous focus. This innovative technique permitted an integrating system that could integrate the images from the various optical sections into a composite image. This concept of moving the focal plane and integrating the small fields of view into a composite image of narrow depth of field is the basis of the future works of Maurice (wide-field specular microscope), and Koester (wide-field specular microscope for in vivo use).

David Maurice had an interesting experimental problem. How to observe the cell patterns and the cellular details in the corneal endothelium of living eyes? This problem was solved by Maurice (Maurice 1968, 1974, 1984). He coined the term "specular microscopy" from Vogt’s original term ”Spiegelmikroskopie,” and developed a working instrument. The microscope aperture was divided; one side used for illumination of the cornea with the projected image of a thin slit, and the other side was used for observation. In order to separate the strong reflection from the tear- cornea interface from over whelming the weak specular reflection from the endothelium it was necessary to use very narrow slits. This resulted a narrow field of view in the plane of the endothelial cells. Only a few cells could be observed. If the slits were widened, then the optical sectioning of the specular microscope was degraded.

The latter problem was also solved by Maurice. Based on the experimental work of Goldmann a new type of specular microscope was developed which used very narrow slits. However, a large field of view was obtained on the integrated photographic image. The solution to the experimental problem was as follows. The ex vivo eye and the film in the camera were moved in tandem. Therefore, the narrow image of the corneal endothelium was integrated into a set of adjacent narrow imaged which formed an image of high contrast and a large field of view. The disadvantage of this instrument was that it could only be used on ex vivo eyes. It was not suitable for in vivo observation of the corneal endothelium. The work of Maurice laid the foundations of all modern developments in specular microscopy of the endothelium-optically sectioning the cornea. In the early papers of Maurice the previous work of Goldmann and Baer were acknowledged (Masters 1996b).

It was during the next twenty years that others developed clinical specular microscopes for the examination of the in vivo cornea. Leibowitz, Laing and Sandstrom in 1975 demonstrated specular microscopy of the in vivo rabbit endothelium. They used photography to ”freeze” the motion of the eye and to record the endothelial specular reflection. This followed the important contributions of Niesel, and later Brown who were able to photograph the slit-lamp image of the anterior chamber (Niesel, 1966; Brown, 1970). In 1976, Laing reported on a clinical instrument for photographing the in vivo human endothelium (Laing et al., 1976, 1979).

Subsequent contributions to instrument development were made by Bourne and Kaufman (1976); Bourne and Enoch (1976); Thaer and Bigar (1978); Bigar (1982), Hartmann; Hirst; and by Sherrard and Buckley (1991) and Koester and Roberts (1990). These systems used an applanating cone to flatten the cornea and to help stabilize the cornea from the axial motion which occurred with the pulse. These instruments used divided microscope objectives as did Maurice with his instrument; the field of view was very narrow. While the image showed high contrast, only a small number of cells were in the image. Following the example of Vogt, the specular microscope has been used to observe the anterior region of the in vivo human lens. Both lens epithelium and lens fibers could be observed (Bron and Matsuda 1981).

The experimental problem of the last 40 years still remained. How to optically section the living cornea and obtain high contrast images of narrow optical thickness? In addition, how to obtain images, with the former constraint, which showed large numbers of endothelial cells? The solution to this problem was devised by Koester. His early papers cite the works of Baer and Maurice. Koester modified the principle of Baer and Maurice to produce a wide field specular microscope suitable for in vivo examination of the cornea (Koester, 1980). The slits were made narrow. The divided aperture was used for the microscope objective. An applanating cone was used to flatten the cornea. What was new was the use of an oscillating three-sided mirror. The mirror scanned and synchronously descanned a narrow beam of light across the corneal endothelium. This clever solution solved the long lasting experimental problems the designs of specular microscopes. Koester was later involved with further developments and refinements of his wide field specular microscope. There improvements involved increasing the numerical aperture of the applanating microscope objective (Koester et al., 1993).

Improvements in the specular microscope resulted in images containing thousands of cells; the analysis of these cell shapes and sizes presented a new difficulty. Recently, the application of fast Fourier transform analysis was applied to the morphological characterization of the in vivo human endothelium (Fitzke, Masters, Buckley and Speedwell, 1997).

3.4 The Development of the Clinical Confocal Microscope

Petran and co-workers developed a real-time, direct view confocal microscope which was based on a spinning Nipkow disk (Petran et al. 1968) (Sec 2.2.1). They were able to observe and photograph thin optical sections of the ex vivo animal cornea (Egger et al., 1969). Corneal epithelial cells, nuclei of stromal keratocytes and endothelial cells were observed and photographed in ex vivo eyes.

The Petran tandem scanning Nipkow disk confocal microscope was used to observe the ex vivo cornea by Lemp, Dilly and Boyde (Lemp et al., 1986;). Lemp subsequently arranged to have a Petran tandem scanning microscope mounted on a head rest and applied it to observations of the in vivo human cornea. Lemp working together with Jester and Cavanagh produced a series of studies on the rabbit eye and the in vivo human cornea. They used a low numerical aperture applanating microscope objective developed for specular microscopy. The in vivo cornea was flattened by the applanating microscope objective. The disadvantages of the system are high noise in the intensified video camera and scan lines on the single images. Post processing and frame averaging reduced the noise and removed the scan lines; however, the instruments were no longer real-time. We define real-time as the acquisition and display of usable video frames. If the data acquisition is real-time (video) but post processing is required to enhance the quality of the images, then the system cannot be called real-time.

A clinical confocal microscope based on a Nipkow disk with an intensified video camera as a detector was developed by the Tandem Scanning Corporation, Inc. in the U.S.A. It used a higher numerical objective than was used in the first system that Lemp used at Georgetown University. Their later version of the instrument contained an internal focusing lens which varied the depth of focus while the applanating microscope objective was held stationary on the surface of the deformed cornea. This design was first proposed by Masters (Masters, 1990a).

At the same time Masters and Kino coupled the real-time one sided Nipkow disk confocal microscope with a new detector, a cooled, slow scan CCD camera, to obtain images of the ex vivo rabbit eye (Masters and Kino, 1990, 1993; Xiao et al. 1990). There were no scan lines, the dynamic range was 14 bits, and the confocal system was suitable for both ex vivo eyes and in vivo animal studies.

The use of a clinical confocal microscope based on the Nipkow disk has severe inherent problems. The transmission of the typical Nipkow disk is less than 1%. This means that only 1% of the incident light is transmitted through the disk on the illumination side. The cornea has a very low reflectivity and of the small amount of light that is reflected from the cornea only about 1% of this light passes through the disk from the microscope objective to the ocular or detector.

Masters developed a confocal line scan system for obtaining intensity profiles throughout the depth of the cornea of in vivo rabbits. Both reflectance and fluorescence (NAD(P)H redox fluorometry) were obtained in vivo. The microscope objective was mounted on a piezoelectric driver, and computer controlled the position of the focal plane as it was scanned on the optic axis. The main part of the confocal system was a modified specular microscope. Two sets of slits were used; one on the illumination side and one on the detection side in the eye piece. the system used the divided aperture in the objective first used by Maurice.

The main unsolved problem was how to deal with eye motion. The solution of Masters was to use a rapid line scan. This was provided by a piezo- electric driver rapidly scanning the microscope objective alone the optic axis of the eye. The instrument produced line scans of reflected light or fluorescence light from the different depths in the cornea (Masters, 1988; 1995b).
A new, real-time, scanning slit confocal microscope was developed by Thaer for the observation of the in vivo human cornea (Masters and Thaer, 1994b; Masters et al., 1994). The image of a slit is scanned over the back focal plane of the microscope objective. The slit width can be varied in order to optimize the balance of optical-section thickness and image brightness. The instrument is based on the double-sided mirror which is used for scanning and descanning. This confocal microscope used a halogen lamp for illuminating the slit. The detector is a video camera that acquires images at video rates. This confocal microscope can image basal epithelial cells and the adjacent wing cells in the living human cornea due to its high light throughput. This design was first developed into a real-time confocal microscope over twenty years ago (Svishchev, 1969, 1971). Svishchev designed and constructed a real-time confocal microscope based on a oscillating two-sided mirror (bilateral scanning) and used this microscope to observe living neural tissue in the reflected light mode (reprinted papers in Masters, 1996b).

The following design parameters were incorporated into the real-time, scanning slit confocal microscope (Masters, 1989, 1990b, 1991a, 1995b).

• The use of nonapplanating, high numerical aperture, water immersion microscope objectives, Leitz 50X and Leitz 100X microscope objectives.
• The microscope objective would use a methylcellulose gel to optically couple the tip of the microscope objective to the cornea. There was no applanation or direct physical contact, which deforms the cornea, between the objective and the surface of the cornea.
• One half of the numerical aperture was used for illumination, and one half of the numerical aperture was used for collection of the reflected and fluorescence light.
• Optical sectioning in the plane of the cornea was obtained with two sets of conjugate slits. The slit heights are variable and adjustable.
• An oscillating, two sided mirror (bilateral scanning) was used for scanning the image of the slit over the back focal plane of the microscope objective, and for descanning the reflected and back scattered light collected by the microscope objective from the focal plane in the specimen.
• The light source is a 12 volt halogen lamp. For fluorescence studies a mercury arc lamp or a xenon arc lamp can be used (Thaer et al., 1990).
• The scanning was synchronized with the read-out of an interline CCD camera in order that the full vertical resolution of the intensified CCD camera could be utilized.

The optical layout of the real-time, scanning slit confocal in vivo confocal microscope is shown in Fig. 2. The real-time, scanning slit in vivo confocal microscope is based on two sets of adjustable conjugate slits. An oscillating two sided mirror is used for both scanning and descanning. This is similar to the design and construction of the Svishchev microscope (Masters, 1996b).

The microscope used standard nonapplanating microscope objectives with RMS threads that are interchangeable. Several different microscope objectives can be used which permits the use of various magnifications and fields of view. Typically a Leitz 50X, 1.0 NA, water immersion objective is used. When a larger field of view is required a Leitz 25X, 0.6-NA water immersion microscope objective is used.

An intensified video camera (Proxitronic) with video output to a Sony S-VHS tape recorder is used. The synchronization of the bilateral scanning and the read-out of the intensified video camera is described by Thaer (Wiegand et al., 1995). The PAL video format provides 625 lines. The S-VHS video recorder provides high band width. In parallel with the video recording, there is a video monitor in addition, so the operator can observe the confocal images of the subject’s eye in real time. This real-time, scanning slit confocal microscope does not require any frame averaging for producing the image quality and contrast shown in this paper.

Another type of scanning slit confocal microscope was developed by Koester who modified his original design of the wide-field specular microscope. The Koester wide field specular used two conjugate slits and was a true confocal microscope. It suffered from poor optical sectioning capability due to the low numerical aperture of the original applanating cone objective (0.33 NA). This resulted in poor efficiency of light collection. To observe images the slits had to be opened and the resulting images had large depth of focus. While the corneal epithelium and the corneal endothelium could be easily observed, the wing, basal and stromal details were very difficult to image.

Recent development of a new applanating objective with an effective numerical aperture of 0.75 resulted in an improved wide-field specular microscope for clinical observation of the human cornea (Koester et al., 1990, 1993). This new instrument uses a divided aperture, applanating microscope objective with an improved numerical aperture. The divided aperture objective design uses one half of the objective for illumination, and the other half of the objective for light detection. This scheme results in the effective numerical aperture of 0.38 in the meridian perpendicular to the obscuration divider in the center of the objective, and an effective numerical aperture of 0.75 in the meridian parallel to the obscuration divider. The transverse resolution differs in the two perpendicular meridians, reflecting these two different numerical apertures. This instrument can image basal epithelial cells in the normal in vivo human eye. In the initial design, a photographic camera was used as the detector; therefore images are not obtained in real-time and require negative development and printing after image acquisition. A more recent design uses a CCD camera. The microscope has an optical section thickness of about 40 µm. This new modification of the previous Koester wide-field specular microscope can image the normal basal epithelial cells of the in vivo human cornea. However, it requires an applanating microscope objective which in addition to helping to stabilize the in vivo cornea induces artificial deformation induced ridges in the stroma and in Descemet’s membrane (Auran et al., 1994).
What are the advantages of using a scanning slit confocal microscope such as is described and demonstrated in this paper? Slit scanning confocal microscopes have a much higher light throughput than confocal microscopes based on Nipkow disk. This has two consequences. Firstly, the illumination incident on the patient’s eye can be much less. This allows for a much longer duration of the use of the confocal microscope on the patient’s eye without the severe patient discomfort and high light intensity that is necessary with the use of the confocal microscope based on the Nipkow disk. Secondly, it is possible to image the low reflecting layer of wing cells that are immediately adjacent to the basal epithelial cells in the normal human cornea. This layer of wing cells has been imaged, in real-time, as single video frames without the need for any analog or digital image processing using the real-time scanning slit confocal microscope. No other real-time confocal microscope has been able to image these wing cells in the normal, in vivo human cornea. It is extremely difficult to image in real-time the basal epithelial cells of the normal in vivo human cornea. Only one group in Kyoto, using a Nipkow disk based confocal microscope, has succeeded in imaging the normal basal epithelial cells (Tomii and Kinoshita, 1994). The modified wide field specular microscope of Koester based on slits can image basal epithelial cells in the normal in vivo human cornea; however this is not in real-time since it is a photographic system.

Confocal microscopes based on slit systems have other advantages. The slit height can be varied to change the depth of the optical section and the amount of light throughput. If the cornea is very clear the slits can be closed down to yield a thinner optical section. However, if the cornea is cloudy, one can open the slits to pass more light.

What are the disadvantages of slit scanning confocal microscope as compared to confocal microscope based on a Nipkow disk containing pinholes? The resolution of a pinhole based confocal microscope is higher than that based on slits. This does not seem to be an important factor for in vivo confocal microscope of the human cornea. The transverse resolution of a slit scanning confocal microscope varies in the x-y plane according to the direction of the slits. A confocal microscope based on pinholes would not have this directional variation in transverse resolution.

The most important design features of the real-time scanning slit in vivo confocal microscope described in this article is the use of non-applanating, long working distance, high numerical aperture water immersion microscope objectives. For a decade, we have advocated and used Leitz water immersion microscope objectives of 50X and NA 1.0. This high numerical aperture microscope objective is very efficient in collecting the light from the weakly reflecting corneal structures. Since we use normal RMS threads we can readily change the microscope objective in order to change the magnification and the field of view. This important condition of readily changeable, standard RMS microscope objectives (from any manufacturer) is not available on the other confocal microscopes designed for in vivo imaging of the cornea.

The confocal microscopes that have been described were designed to observe the in vivo human cornea. We now describe a unique instrument has been developed to image the in vivo retina.

In 1949 Ridley pioneered the development of the television ophthalmoscope, and point scanning of the retina using a cathode ray tube as a scanning point source of light for retinal illumination (Ridley, 1952). The spot of light on the screen of a cathode ray tube was raster scanned and imaged into the retina of a subject’s eye. The spot of light scattered from the retina was imaged onto a photomultiplier tube and the two dimensional image of the retina was displayed in real-time on a television monitor. Ridley correctly pointed out that the use of point scanning, in which each spot on the retina is sequentially illuminated and the reflected and scatted light from that point imaged onto a detector, greatly improved the contrast of the image. This is a general principle which is valid for all confocal microscopes (Davidovits and Egger, 1971; Ridley, 1952).
A recent development in clinical confocal microscopy is the scanning laser ophthalmoscope developed by Robert Webb and coworkers (Webb, 1990, 1996; Webb et al., 1980, 1987). A laser beam is raster scanned over the retina at video rates and the image of the retina is viewed on a video monitor. Video rate scanning is achieved by the combination of a rotating polygon mirror and a scanning galvanometer mirror situated in orthogonal axes. A digital laser scanning fundus camera based on a similar design was developed in Germany (Plesch et al., 1987).

3.5 Practical Requirements for a Clinical Confocal Microscope

What are the practical requirements for a clinical confocal microscope designed to image the anterior segment of the living human eye (Masters and Böhnke 1998) ? Patient safety is of paramount importance. The use of the confocal microscope must not harm the patient. The head rest must be designed to prevent an involuntary head jerk from injuring the patient’s eye. The electrical system must be properly grounded. There should not be hot regions on the instrument which could inadvertently burn the patient’s hands. The light intensity incident on the patient’s eye must not exceed the standard safety limits of intensity and exposure time. The wavelengths of light used must not be phototoxic. Phototoxicity could be enhanced with the use of drugs that are photosensitizers.

The light source should be either a broad band source such a xenon lamp or a multiwavelength source such as a series of different colored diode lasers. The light should fill the back focal plane of the microscope objective with even intensity of illumination. The intensity of the light should be easily and quickly adjustable.

A range of water immersion microscope objectives with standard RMS threads should be available. Since there is a reciprocal relation between the magnification of the microscope and the free working distance it is important to have a set of microscope objectives that can image from the anterior to the posterior surface of the cornea. The major manufactures of optical microscopes have recently made available a series of water immersion microscope objectives with high numerical apertures and long working distances and minimal optical aberrations. These designs and their optical quality are superior to the custom designed microscope objectives that have been required for some of the commercially available clinical confocal microscopes, e.g., Tandem Scanning Corporation, and the Koester wide field clinical confocal microscope.

The temporal resolution should be sufficient so that each frame that is recorded by the instrument is devoid of motion artifacts. The spatial resolution should be sufficient to observe the smallest details of the cornea that are important for clinical evaluation.
The image quality of the clinical confocal microscope is critical for examination of the cornea. Image quality depends on the signal to noise ratio of the image, the contrast of the image, and the magnitude of the optical aberrations in microscope and in the microscope objective.

Contrast refers to the ability of the microscope to detect an object in the specimen from its surroundings based on an intensity difference that can be seen in the image. Even with sufficient spatial resolution an object can not be observed from its surrounding without sufficient contrast. Contrast is degraded by optical aberrations, stray light in the microscope, and noise in the detection system.

There are two main areas of design modifications of the nonapplanating, real-time, scanning slit, in vivo confocal microscope described in this paper. In order to investigate the tear film in the human cornea it is desirable to have a non-contact instrument. Since the tear-film is a dynamic structure is it important to sturdy its naturel dynamic structure. When a microscope objective contacts the intact tear film it disturbs the structure of the tear film. Although the real-time, scanning slit microscope described in this paper used a gel between the tip of the water immersion microscope objective and the surface of the cornea in order to minimize the contact with the eye the structure of the tear film is still altered. Only a non-contact confocal imaging system can image the natural tear film dynamics.

The second important modification is to develop a microscope for imaging the human ocular lens. It would be important to image to the posterior capsule of the crystalline lens in the in vivo human eye.