Optical Sectioning of the Comea with a New Confocal In Vivo Slit--scanning Videomicroscope

Wolfgang Wiegand, MD, PhD, 1 Andreas A Thaer, PhD, 2 Peter Kroll, MD, 1

Otto-C. Geyer, MD,2 Alberto]. Garcia, MD3

Purpose: The purpose of this article is to introduce a newly developed confocal in vivo slit-scanning microscope for continuous recording and real-time imaging of the various corneal subsegments of the patient's eye with high microscopical resolution and adequate contrast.

Methods: One-dimensional confocal slit-scanning videomicroscopy of the human cornea was performed with an instrument mainly consisting of a scanning module, an image-intensifier video camera, a video monitor, and a synchronization unit for matching optical scan and video cycle with respect to frequency and phase. Light intensity or fluorescence intensity profiles through the cornea could be obtained by microphotometric recording of part of the imaging light. An immersion contact technique using an isotonic tear replacement liquid with thixotropic properties avoids any mechanical contact be­tween the front lens of the microscope objective and the corneal surface.

Results: In normal human eyes, the corneal micromorphology could be made visible with satisfactory lateral and axial resolution and with good contrast. The separately focussed sections of the cornea showed the endothelial cells, the superficial, interme­diary, and basal cells of the epithelium, as well as stromal keratocytes and nerves. Even in eyes with significant corneal opacities resulting from corneal edema, the endothelial pathology could be imaged with sufficient contrast.

Conclusion: The in vivo slit-scanning videomicroscopy offers real-time noninvasive and noncontact serial imaging of corneal subsegments with resolution and imaging contrast. Thus, an important step toward using confocal scanning microscopy for corneal diagnosis seems to be done. Ophthalmology 1995;102:568-575

Microstructural analysis is of growing importance for ophthalmologic diagnosis. This has become particularly

true for the anterior segment of the eye, namely for the cornea and its subsegments, because photorefractive ker­atectomy is performed worldwide. A significant barrier for the in vivo use of microstructural findings in the nor­mal eye, in eyes with corneal diseases, and in eyes after photorefractive keratectomy, however, is the restricted optical resolution and magnification range of the slit lamp being sufficient for the macroscopic range only. Attach­ments to the slit lamp for biomicroscopy or special ophthalmomicroscopes, which have been used success­fully for the examination of the corneal endothelium, do not meet the requirements for an in vivo microstructural analysis of the different corneal sections because light re­flected and light scattered by structures above and below the focussing plane of the cornea impair the image contrast significantly. To satisfy these requirements, therefore,

Originally received: August 4, 1994. Revision accepted: December 31, 1994. 1 Department of Ophthalmology, University of Marburg, Germany. 2 Institute for Medical Visual Aid, Wetz1ar, Germany. 3 Ophthalmology Practice, Emden, Germany.

Presented in part as a poster at the American Academy of Ophthalmology Annual Meeting, Chicago, November 1993.

Supported by grants of Bundesministerium ftir Forschung und Tech­nologie (BMFT), Bonn, Germany.

Dr. Andreas A. Thaer is consultant to the company Helmut Hund GmbH, Wetzlar, Germany.

Reprint requests to Wolfgang Wiegand, MD, PhD, Department of Oph­thalmology, Philipps University, Robert-Koch-Str 4, D-35033 Marburg, Germany.

568

Wiegand et al · In Vivo Optical Sectioning of the Cornea

J/A

,.···-· ..

FIELD DIAPHRAGM

ILLUMINATED OBJECT AREA RESTRICTED BY

........

many attempts have been made in the last few years to develop confocal microscopes. 1.4-9

The purpose of this article is to introduce a newly de­veloped confocal in vivo slit-scanning microscope for real­time serial imaging of the microstructures of the cornea, making use of the known advantages of this new light­microscopic principle with respect to optical resolution, image contrast, and optical sectioning capability.

Instrumentation and Methods

In contrast to conventional microscopy, where the illu­minated object area is restricted by a field diaphragm only and the incident light thus covers a large area of the imaged tissue almost homogeneously, confocal microscopy is based on the principle that the incident light is focussed within a small region of the focal plane of the microscope objective leading to a steep axial light gradient in the tissue (Fig l ). This reduction of the illuminated object area min-

I :1

J/AL . . . . . . .

........ ........

SLIT· OR POINT .SHAPED CONFOCAL DIAPHRAGM

Figure 1. Axial light-intensity gradient 0/ A) in an axially extended object such as the cornea in case of ordinary (le&) and confocal (right) light microscopy (schematic).

imizes the influence of light on the image which is scat­tered by structures outside the focussed object plane. The minimizing effect is supported further by inserting an ap­erture in the plane of the image (i.e., in a confocal position with respect to the illuminating aperture). Thus, all light reaching the image plane outside of this aperture is pre­vented from influencing the image of the object and therefore prevented from reducing the image contrast. Both confocal diaphragm apertures scan equidirectionally across the imaged object area with identical speed and amplitude.

In principle, confocal microscopy can be performed either by two-dimensional spot scanning1- 8•10- 12 or by one­dimensional slit scanning9•13•14 (Fig 2). In our instrument, the one-dimensional slit scanning instead of the two-di­mensional spot scanning is used among other things be­cause of its easy applicability in combination with video equipment. The optical contrast in one-dimensional scanning is only slightly lower than in two-dimensional scanning if the effective slit aperture is small enough.

Figure 2. Optical sectioning of a three-dimensional mi­croscopic object (schematic) by two-dimensional spot· scanning (le&) and one-di­mensional slit-scanning (right).

569

Ophthalmology Volume 102, Number 4, April1995

Figure 3. Optical arrangement for in vivo confocal slit-scanning microscopy and mi­crophotometry of the cornea.

J/A

,, '• I 1 :I I : 1 I

CONFOCAL SCANNING SLITS LIGHTSOURCE ,' I I I . · .. _______ .. ': SYNCHRONIZED WITH A VIDEO IMAGE ;ICK-UP CYCLE ~

....... ~

/I I I

SPECTRAL FILTER ! \ FLUO:~~ENCE b'fm'm~ Z-SCAN (1 mm) t I I ~

--,t..:-J-.::-:-~----; ~-+---~- --~(?]~~'~'o ..__~--..,..-r I -- - · I \

PHOTOSENSOR 1 : - - -i- - - - _ t;:. : I t : • -, ---· :!f• : ~ x~ ~EAM SPLITTER

1 I / 50/50 I ' I

: ~ DICHROMATIC BEAMSPLITTER ~ FOR FLUORESCENCE

,-,-----. <4t-r--------­~----------

IMAGE INTENSIFIER VIDEO CAMERA

D VIDEO MONITOR

The instrument (Fig 3) mainly consists of a conven­tional light source (halogen lamp 12 VI 100 W /DC), a scanning module for lateral confocal scanning containing

the illuminating and imaging optics with the respective scanning slits combined in a fixed distance to each other in an electromagnetically driven double-slit arrangement, an image-intensifier video camera, a video monitor, and a synchronization unit for frequency and phase-sensitive matching optical scan and video cycle. The scanning fre­quency is 25 Hz. A microprocessor-controlled stepping motor can move the focus of the microscope optics in the direction of the optical axis with a velocity of I mm/

---CORNEAL EPITHELIUM

TEAR LIQUID

Figure 4. Friction-free optical contact between the front lens of the mi­croscope objective of the confocal scanning system and the corneal surface by use of a thixotropic gel (VIDISIC-Gel, Dr. Mann Pharma, Berlin).

570

Figure 5. Videomicrograph of the rabbit corneal endothelium without (left) and with (right) synchronization between optical slit-scan and video­image cycle.

Wiegand et al · In Vivo Optical Sectioning of the Cornea

Figure 6. Normal human corneal endothelium (objective, 25/0.60).

second and thus scan the whole cornea within approxi­mately 0.5 second, the exact value depending on the actual thickness of the cornea. Thus, a sequence of lateral sec­tions at different depths of the cornea (z-scan) can be re­corded and stored on a videotape. In addition, the z-scan permits the photometric recording of light-intensity pro­files through the whole cornea (or parts of it) without lat­eral scanning but with an axial resolution given by the focal depth of the objective used. For fluorescence mea­surements, spectral filters (excitation and barrier filter) and a high-pressure mercury lamp ( l 00 W /DC) instead of the halogen lamp can be added.

For eliminating image blurring due to eye movements and internal (e.g., caused by the scanner) or external ("seismic") vibrations, each half of the full video image is generated by a single scan only. Its duration is 20 mse­conds (one of the half images) or 40 mseconds (full image), the scanning frequency being 25 cycles in both cases. The time for imaging a microstructural detail still resolved by the video camera is given by its individual exposure time on the photocathode and thus by the width of the slit

Figure 7. Human corneal endothelium with cellular polymegatism (ob­jective, 25/0.60).

Figure 8. Human corneal endothelium with cellular precipitate on its rear side due to inflammation (objective, 25/0.60).

image and its speed across the photocathode. The normal slit width used corresponds with l/30 of the scanning am­plitude or approximately 20 lines of the 600 lines of the video camera. Thus, the exposure time of each resolved microsiructure in each slit-shaped area is not more than 20 mseconds/30 = 0.66 msecond! For using the full video line number, an interline CCD camera is used. A one­step image pre-amplifier with optical fiber coupling to the ceo camera reduces the necessary illumination intensity for keeping the light exposure of the cornea as low as possible. An accurate synchronization between the image pick-up cycle of the video cameraand the optical slit scan is obtained by using the V -signal of the video camera for triggering the scan control unit.

In addition to the microscopic imaging of corneal structures, corneal profiles based on light scattering or on fluorescence can be recorded photometrically. Part of the imaging light (20%) is therefore used for a microphoto­metric registration of scattered light or fluorescence in­tensity profiles throughout the cornea (see Fig 3) by means

Figure 9. Human corneal endothelium shows an advanced "cornea gut­tata" (objective, 25/0.60).

571

Ophthalmology Volume 102, Number 4, April1995

Figure 10. Human cornea; stroma with keratocytic nuclei (objective, 25/ 0.60).

of the z-scan. An immersion contact technique (Fig 4) using a commercial isotonic tear replacement liquid (VI­DISIC-Gel, Dr. Mann Pharma, Berlin, Germany) with a refractive index of J.to = 1.350 provides an optically almost homogeneous immersion between the front lens of the objective and the anterior chamber. It also avoids any mechanical contact and shearing forces between the mi­croscope objective and the corneal surface. The low vis­cosity and thixotropic property (low refractive index dif­ference at the interface gel/corneal surface) of the sub­stance is caused by a 0.2% polyacrylic acid content with a molecular weight of approximately 4 million. Thus, the immersion drop behaves like a ball with very low internal convection.

A 25/0.60 or a 40/0.75 water-immersion objective is used as standard lens ofthe instrument. For highest optical resolution, a 50/1.00 water-immersion objective can be used. The use of the latter is restricted to imaging the anterior corneal subsegments for reason of its smaller free­working distance. The effective lateral optical resolution

Figure 11. Human cornea; nerve branching in the anterior stromal region (objective, 25/0.60).

572

Figure 12. Human cornea; nerve fibers without myelin close to Bowman membrane (objective, 25/0.60).

of the system is between 1 and 2 J.tm, depending on the objective.

Application and Results

Before using the system for in vivo real-time examinations of human corneas, it has been tested thoroughly in human volunteers and in rabbits. Before examination, an in­formed consent was signed by the patients. At the begin­ning of the examination, a drop of topical anesthetic was placed into the patient's eye and the eye then was posi­tioned in front of the objective lens of the instrument. All investigations described in this article were performed with the immersion contact technique and either the 25/0.60 water immersion objectives or the 50/1.00 water immer­sion objective. Due to the immersion fluid on top of the objective lens, the corneal epithelium was not touched during the investigation, and the cornea thus was not ap­planated. From all patients, initially a z-scan of the whole

Figure 13. Human corneal epithelium with cells before desquamation (objective, 25/0.60).

Wiegand et al · In Vivo Optical Sectioning of the Cornea

Figure 14. Human corneal stroma with keratocytic nuclei surrounded by cytoplasmic contours usually becoming visible in case of stromal swell­ing (objective, 50/1.00).

cornea was taken and the interesting corneal sections then were imaged subsequently. All images were stored on vid­eotapes, and Figures 5 through I 7 were taken from the video screen.

It generally can be noticed that all photographs of strongly reflecting sections show the best imaging contrast of the structures in the center of the video screen, whereas the cell contours vanish close to the edges of the photo­graphs. This is due to the normal curvature of the cornea ~hich is not applanated by a contact element. Figure 5 Illustrates the effect of synchronization between optical scan and video signal in the rabbit cornea. Figures 6 to 9 show video micrographs of the human corneal endothe­lium taken with a 25/0.60 water-immersion objective. The norm~l endothelium (Fig 6) as well as endothelial poly­megatism due to extended contact lens wearing (Fig 7) and cellular precipitates on the rear side of the endothe­li~~ in a patient with anterior uveitis (Fig 8) are clearly visible. Due to the confocal principle, the corneal endo­thelium could be imaged even in patients with significant corneal opacities resulting from corneal edema or corneal dystrophies. An example of severe endothelial pathology ("cornea guttata") is presented in Figure 9.

Figure 15. Human corneal stroma; structured keratocytic nuclei with notches (objective, 50/1.00).

Figure 16. Human cornea; basal cell layer of the epithelium (objective 50/1.00). '

Within the co_rneal strom~ keratocytic nuclei (Fig 10), nerve branches m the an tenor stromal regions (Fig II) and close to Bowman membrane (Fig 12) can be delin­eated us_ing a 25/0.60 water-immersion objective also. In sup~rficial corneal epithelium, the cells before desqua­matiOn show a much brighter reflection than the sur­rounding cells (Fig 13). . The i~p~ove~ lateral resolution and selectivity of op­

tical sectwnmg m case of a 50/1 .00 water-immersion ob­jective is _demon~trated in Figures 14 to 17. They show keratocytic nuclei surrounded by cytoplasmatic contours (Fig 14) or structured keratocytic nuclei with notches (Fig 15) .. Some other corneal subsegments like the basal epi­th~hal cell layer (Fig 16) or the intermediate cell layer (wmg cells) of the epithelium (Fig 17) also become visible at a higher numeric aperture only.

None of the video micrographs shown in the figures are_decreased in contrast or blurred due to eye movement which proves the possibility of real-time in vivo imaging of the human cornea at the cellular level. According to

Figure 17. Human cornea; intermediate cell layer of the epithelium (ob­jective, 50/1.00).

573

Ophthalmology Volume 102, Number 4, April1995

INTERFACE I I

TEARFILM I EPITHELIUM BASAL LAMINA

Figure 18. Microphotometric light intensity profile by use of confocal z-scanning throughout the human epithelium recorded in vivo (objective, 50/1.00).

IMMERSION LIQUID

I l I J

...

-

I 1.. T

50J.Im -;;

I I J

... -~ ~I.. l1~~t r•

'•I

-

llljll r.\j II: A~~r

STROMA ~

Ll .it.ra-' ~ .. "' lr.1 .,.1.

J

our experiences, the short exposure time reduces the por­tion of images with blurred structural details down to or below 5% (i.e., to approximately 1 frame of 25 frames/ second). This is true for image blurring due to both lateral or axial eye movement. Thus, usually a ratio of video images unaffected by eye motions of more than 90% is gained even for the highest magnification and optical res­olution applied. Evidence for a sufficient selectivity of the optical sectioning also is given by the light-intensity profile through the human corneal epithelium (Fig 18) recorded in vivo with a 50/1.00 water-immersion objective and z­scan movement of the focus.

Discussion and Conclusion

During the last two decades, confocal scanning light microscopy1- 3 has established the basis for an "optical sectioning" of translucent microscopic objects whose thickness exceeds the depth of focus of the microscope optics by many times. Sequential imaging of small point­shaped or slit-shaped areas of the focussed object planes and the confocal optical arrangement, both led to an ex­cellent image contrast. During the last 10 years, such op­tical arrangements have been used successfully for a sys­tematic investigation of the corneal microstructure of enucleated eyes and excised corneal tissue.4- 6•8•10- 12 Al­though confocal scanning microscopy also has been ap­plied to in vivo examination of the cornea,6•9•13•14 its in­troduction into corneal diagnosis is hampered still by im­age blurring due to eye movements, unacceptable light intensity of the scanning light beam, problems encoun­tered in optically interfacing microscope lens and eye sur-

574

face, and by problems obtaining laterally or axially ar­ranged image series for direct observation, image storing, and image documentation. Some contributions to the so­lution of these problems have been discussed previously elsewhere. 15- 17

Essential technical features of our device are the con­focal one-dimensional scanning by use of slit-shaped dia­phragms in the illuminating and imaging light paths co­inciding in the focussed object plane, the use of a sensitive image-intensifier camera, and use of an exact frequency and phase-dependent synchronization of optical scanning and video recording. This allows the imaging of lateral optical sections of the cornea with very high resolution and results in sharp images unaffected by eye motion. Due to the video .equipment, the images can be recorded, stored, and repeatedly displayed. The use of an immersion contact technique greatly diminishes the infection risk compared with the use of contact elements. The device has been very effective and reliable in clinical use.

The real-time in vivo video micrographs shown above prove that an important step toward the use of confocal scanning microscopy for corneal diagnosis has been done. Thus, in vivo corneal "histology" and "histopathology" seem to be feasible not only for diagnostic purposes in humans but also for noninvasive tests in animals. The images can be achieved very quickly (with 25 exposures per second) and without exposing patients or test persons to stress exceeding that of a slit-lamp examination. Con­cerning optical resolution and contrast, the video micro­graphs in Figures 6 to 1 7 are comparable approximately with selectively exposed in vivo confocal slit-scanning photomicrographs taken by using a shutter at 1/60-second exposure time, which were published recently.9 However, due to a lower number of pixels per area made available

Wiegand et al · In Vivo Optical Sectioning of the Cornea

by the video cathode, our video micrographs show a more pronounced grain of the background and a field restriction compared with photomicrographs exposed on 35-mm film. The high image sampling rate, on the other hand, compensates for this field reduction.

The possibility of studying the corneal microstructure dynamically without any preparational or staining arti­facts adds further important information to corneal mi­croanatomy and corneal· physiology. Processes such as wound healing as a function of time thus can be inves­tigated. The new methods of refractive surgery by anterior or intrastromal laser ablation or by keratomileusis in situ doubtlessly will profit by this possibility. In addition to the microstructural information, in vivo measurements of the scattered light intensity and permeation measure­ments with nontoxic fluorescent stains can be performed, which give further information on the actual functional state of the cornea and its subsegments.

References

1. Petmn M, Hadravsky M, Egger MD, Galambos R. Tandem­scanning reflected-light microscopy. J Opt Soc Am 1968;58: 661-4.

2. Wilson T, Sheppard C. Theory and Practice of Scanning Optical Microscopy. London: Academic Press, 1984.

3. Wilson T. Confocal light microscopy. Ann NY Acad Sci 1986;483:416-27.

4. Lemp MA, Dilly PN, Boyde A. Tandem-scanning (confocal) microscopy of the full-thickness cornea. Cornea 1985-86;4: 205-9.

5. Dilly PN. Tandem scanning reflected light microscopy of the cornea. Scanning 1988;10:153-6.

6. Cavanagh HD, Jester JV, Essepian J, eta!. Confocal mi­croscopy of the living eye. CLAO J 1990;16:65-73.

7. Cavanagh HD, Petroll WM, Alizadeh H, eta!. Clinical and diagnostic use of in vivo confocal microscopy in patients with corneal disease. Ophthalmology 1993;100:1444-54.

8. Masters BR, Paddock S. In vitro confocal imaging of the rabbit cornea. J Microsc 1990;158:267-74.

9. Koester CJ, Auran JD, Rosskothen HD, eta!. Clinical mi­croscopy of the cornea utilizing optical sectioning and a high-numerical-aperture objective. J Opt Soc Am [A) 1993;10:1670-9.

10. Jester JV, Cavanagh HD, Lemp MA. Confocal microscopic imaging of the living eye with tandem scanning confocal microscopy. In: Masters BR, ed. Noninvasive Diagnostic Techniques in Ophthalmology. New York: Springer-Verlag, 1990; chap. 11.

II. Masters BR, Paddock SW. Three-dimensional reconstruc­tion of the rabbit cornea by confocal scanning optical mi­croscopy and volume rendering. Appl Optics 1990;29:3816-22.

12. Masters BR, Kino GS. Confocal microscopy of the eye. In: Masters BR, ed. Noninvasive Diagnostic Techniques in Ophthalmology. New York: Springer-Verlag, 1990; chap. 10.

13. Koester CJ, Roberts CW, Donn A, Hoefle FB. Wide field specular microscopy. Clinical and research applications. Ophthalmology 1980;87:849-60.

14. Koester CJ, Roberts CW. Wide-field specular microscopy. In: Masters BR, ed. Noninvasive Diagnostic Techniques in Ophthalmology. New York: Springer-Verlag, 1990; chap. 7.

15. Masters BR, Thaer AA. Real-time scanning slit confocal microscopy of the in vivo human cornea. Appl Optics 1994;33:695-70 1.

16. Masters BR, Thaer AA. Confocal microscopy of the in vivo cornea. In: Ophthalmic and Visual Optics: Summaries of Papers Presented at the Ophthalmic & Visual Optics Topical Meeting, February 19-20, 1993: Noninvasive Assessment of the Visual System Meeting, February 21-23, 1993. Washington, DC: Optical Society of America, 1993;133-6 (Technical digest series 1993;3).

17. Wiegand W, Thaer AA. In-vivo-Hornhautmikroskopie mit einem neuen konfokalen Spalt-Scanning Video-System. In: Robert YCA, Gloor B, Hartmann Ch, Rochels R, eds. 7. KongreJ3 der Deutschsprachigen Gesellschaft fiir Intraok­ularlinsen-Implantation. Berlin: Springer-Verlag, 1993;484-8.

575