The problem of three-dimensional visualization of a human lens in vivo has been solved by a technique of volume rendering a transformed series of 60 rotated Scheimpflug (a dual slit reflected light microscope) digital images. The data set was obtained by rotating the Scheimpflug camera about the optic axis of the lens in 3 degree increments. The transformed set of optical sections were first aligned to correct for small eye movements, and then rendered into a volume reconstruction with volume rendering computer graphics techniques. To help visualize the distribution of lens opacities (cataracts) in the living, human lens the intensity of light scattering was pseudocolor coded and the cataract opacities were displayed as a movie.
©1998 Optical Society of America
The development of the light microscope is closely linked to advances in biomedicine. The story of the early microscopic imaging of the ocular lens is less well known. Human lens fibers were the subject of microscopic investigation by Leeuwenhoek about 350 years ago. Leeuwenhoek used his single lens microscope to examine the ocular lens from several species. His studies included the human lens.
The adult human lens is an oblate, spheroid which is composed of concentrically arranged shells of lens fiber cells. In the adult human the number of fiber cells may be in the range of 3.5 million. The size of an adult human lens is about 3.6 mm from the anterior to the posterior pole (transverse diameter) and about 9 mm on the equatorial axis (equatorial thickness). The human lens increases in weight and thickness throughout the human life span. The ocular lens changes its shape on accommodation which is the process by which the refraction of the eye is changed in order to focus an image on the retina. Cataract is defined as an opacity in the ocular lens. If the cataract is sufficiently dense and on the optic axis it may impair vision[1–2].
The enhanced optical sectioning capability of the reflected light confocal microscope has been used to investigate the morphology of the unfixed, unstained, ex vivo rabbit lens [3–7]. Reflected light confocal microscopy and electron microscopy have been correlated in a morphological study of lens epithelial cells (imaging nuclei with their nucleoli, lens fibers with their surface undulations, and vacuolar elements and fiber disorder in regions of the cataract in the ex vivo human ocular lens .
The ability to observe the in vivo human ocular lens is an important factor in lens and cataract research . The problem is formidable since the lens is thick, and therefore it is difficult to obtain a sharp two-dimensional image of the entire lens with the slit lamp . However, it is possible to image regions of the in vivo lens with various optical instruments. The lens epithelium has been observed in vivo by specular illumination with a the slit lamp microscope . The specular microscope has been used to image specific regions of human lenses in vivo [10–12]. A scanning laser ophthalmoscope was used to examine the human ocular lens in vivo .
The development of the Scheimpflug camera (a dual slit reflected light microscope) resulted in a new capability to obtain reflected light images across the full thickness of the human ocular lens in vivo. The Scheimpflug principle requires that the plane containing the slit beam and the image plane intersect at one point, with the corresponding angles equal.
The Scheimpflug camera is shown in Fig. 1. The unique feature of this camera is its ability to form an image of the back scattered light from the full thickness of the ocular lens. The Scheimpflug principle is shown in Fig. 1. A noncoherent light source at S projects an image of a slit onto the ocular lens. This slit beam is shown by the thick, vertical arrow on the ocular lens. The back scattered light from the cornea, the iris, and the ocular lens is imaged onto the image plane which is shown in Fig. 1. by the thick, horizontal arrow in the image plane. If the planes of the slit beam and the image plane (containing the photodetector) are arranged as shown in Fig. 1. then the image will contain the backscattered light from the cornea and the full thickness of the ocular lens.
The traditional Scheimpflug camera acquires, on either film or an electronic camera, a back scattered light image in a sagittal plane which traverses the full thickness of the ocular lens . The image is a record of the regions of high back scatter (cataract) and regions that are more transparent.
An early advance was the use of a rotating slit image camera which rotated the image plane on the optic axis of the eye and obtained a series of images, each image was tilted a few degrees from the adjacent images . The next advance was the combination of rotating the Scheimpflug camera on the optic axis of the eye during image acquisition and the use of digital image processing to reconstruct a series of coronal sections (coronal sections are orthogonal to sagittal sections) . Finally, the stack of reconstructed coronal sections were combined to form a three-dimensional image of the ocular lens [16–17].
A major limitation in the current technology of imaging the ocular lens is that a single Scheimpflug slit image of the backscattered light from the ocular lens is very sensitive to sampling errors. This is still a problem even in the case of multiple Scheimpflug slit images acquired in parallel planes, or a series of rotated planes. This sampling problem is only solved with a three-dimensional visualization of the opacities in the volume of the ocular lens. A second limitation is that the current technology is not capable of repeated studies of lens opacities over time since it is very difficult to reposition the camera to image identical regions of the ocular lens. A three-dimensional visualization of the backscatter from the ocular lens does not have the problem of exact repositioning; the volume of the ocular lens is reconstructed.
The current study differs from the previous studies in several ways [16–17]. A new interpolation procedure was written and coded as a computer program to convert the rotated set of Scheimflug images into a new aligned orthogonal set of 332 images. A new volume rendering software program is used for rendering, viewpoint control, and opacity control of the rendered three-dimensional lens . The opacity of the reconstructed ocular lens was adjusted so that the distinct regions of opacity within the volume of the ocular lens could be visualized and measured. Finally, pseudocolor was mapped onto the gray level images to aid in the topological visualization of the opacities in the in vivo human lens. The combination of pseudocolor and rotatory motion of the three-dimensional ocular lens enhances the visualization of the internal opacities and presents a new diagnostic advance in cataract research.
2. Materials and Methods
The detailed experimental procedures have been previously described [19–20]. The images used in this study were acquired in the Department of Ophthalmology, Kanazawa Medical University, Uchinada, Japan, under the medical supervision of Dr. K. Sasaki. The subject was a 68 year old male with incipient cataract. Informed consent was given by the subject for the imaging of the lens. The subject had anterior and posterior cortical opacities which were judged to be grade 1 by the Japanese Cooperative Cataract Epidemiology Study Group System. The subject’s pupil was dilated with 0.5% tropicamide and 0.5% phenylephrine hydrochloride prior to the Scheimpflug imaging.
A set of 60 Scheimpflug images were acquired with the Anterior Eye Segment Analysis System (Nidek, EAS-1000). This new type of Scheimpflug camera is based on a CCD camera with a computer analysis system. The Scheimpflug camera was rotated about the optic axis of the patient’s eye in three degree increments. Each digital image was 640 by 400 pixels. The digital images had a dynamic range of 8-bits or 256 gray levels. The data acquisition time was 20 minutes.
The acquired Scheimpflug images were not in registration due to eye movements during the image acquisition process. The Scheimpflug images were realigned interactively using a “four-point registration” feature of that program written by S. Senft . One image was chosen as the “fiducial,” and two easily recognizable landmarks were identified on that section. A corresponding pair of landmarks were picked on an adjacent section, and the program used the two line segments to register the images. The alignment consisted of an xy-shift (to superpose the midpoints of the two line segments), a rotation (to superpose the line segments) and a scaling (to superpose the pairs of points themselves). Application of these transformations to each pixel in the target image resulted in registered images. This set of operations was applied successively to the 60 Scheimpflug images in the data set.
Image resampling refers to the procedure for converting the aligned set of 60 rotated Scheimflug images into a new set of 332 images which can be directly imported into the volume rendering software, VoxelView . The new resampled set of images are orthogonal to the optic axis (coronal sections), and consist of slices from the anterior pole to the posterior pole of the ocular lens.
The aligned, resampled data set was processed with Voxel View (Vital Images, Fairfield, Iowa) software to form three-dimensional volume reconstructions of the in vivo human ocular lens. This software is based on the volume rendering technique .
The voxels in the final three-dimensional reconstruction were pseudocolored to aid in visualization of small differences in opacity of the different regions of the ocular lens. The use of pseudocolor is an important feature of our procedure because of its help in the visualization discrete regions of lens opacities. The pseudocolor coding of the intensity of light scattering from the three-dimensional volume reconstruction of the in vivo human lens is as follows. Dark colors (purple, blue, blue-green) represent low light scattering voxels (high lens transparency) and bright colors (green, yellow, orange, red) represent high light scattering voxels (low lens transparency). Voxels are small volume elements which comprise the digital, three-dimensional volume.
The volume reconstruction of the in vivo human ocular lens is a three-dimensional digital data set. It can be viewed from any perspective and direction. In order to “look inside” the three-dimensional volume the transparency any subset of the voxels forming the object can be increased. For example, in order view the opacities in the anterior regions of the ocular lens we have increased the transparency of the voxels which form the lens epithelium. When the volume is rendered under these conditions, with a viewpoint located in front of the ocular lens, we can “look inside” the lens and see the cataract located posterior to the lens epithelium.
The movie shows the three-dimensional volume of the ocular lens that was accessible to the rotated Scheimflug camera through the dilated pupil. The three-dimensional volume rendering includes the full volume of the lens from the anterior surface to the posterior surface and also a circular ring of the dilated iris. The peripheral regions of the ocular lens that are situated behind the dilated iris are not shown in the reconstruction since these regions of the lens were masked by the iris during the data acquisition phase. The highly scattering iris forms a circular structure on the anterior plane of the lens and thereby serves as a reference landmark. The reconstructed ocular lens is placed in the center of a cubic wire frame which completely enclosed the reconstruction. During the various rotations around two orthogonal axes of rotation the cubic wire frame (yellow color on a black background) helps to visualize the lens from various viewpoints.
In addition to various rotations of the ocular lens the opacity of the volume reconstruction is continuously varied in order that the observer could “see” inside of the lens. This is necessary since the strong regions of light scatter due to high opacities just posterior to the anterior surface (below the lens epithelium) tend to mask the regions of opacity which are deeper in the lens. As the three-dimensional volume reconstruction is rotating about a vertical axis the observer can view the regions of opacity in the posterior regions of the ocular lens. When the reconstructed lens is aligned with the coronal plane the small regions of opacity are readily observed. Although not shown in the movie loop each of the voxels forming the three-dimensional reconstruction are color coded on a scale of 256 levels of scattered light; therefore the various regions of opacity could be measured. In addition, the volume of each region of opacity within a given level of light scattering can be measured.
The three-dimensional, pseudocolored, volume rendering of the in vivo human ocular lens is an advance in the noninvasive techniques of three-dimensional ocular imaging. The visualization of the opacities of the human lens in three-dimensions, with arbitrary cutting planes of any angle and direction, provides an improvement over the two-dimensional Scheimpflug images. Since the full volume of the ocular lens is available as a three-dimensional digital volume it can be viewed in any subregion and from any angle or viewpoint. The traditional Scheimpflug images are limited to information contained within the plane of the image. The use of pseudocolor is helpful to visualize discrete regions of opacity in three-dimensions.
The image resampling technique used in this paper obtained 332 images from the initial set of 60 rotated images. It is clear that an increased number of rotated images would result in a closer approximation of the scattering from the entire volume of the ocular lens. In the resampling technique more voxels must be interpolated as the distance from the optical axis of the eye increases. Therefore, the reconstruction method has a higher fidelity near the center of the ocular lens and it decreases with distance orthogonal to the optic axis.
Since the Scheimpflug camera was mechanically rotated between each image acquisition and then realigned and focused on the subject’s lens, it was not feasible to increase the number of acquired imaged due to the limitations of the subject. The time of data acquisition was 20 minutes for the 60 rotated images. However, a future development of a new camera with rotating optical prisms could shorten the image acquisition time and increase the number of rotated images.
This paper illustrates a new technique to use the Scheimpflug camera and computer visualization techniques to reconstruct a three-dimensional visualization of the human ocular lens in vivo. The Scheimpflug camera is used to generate two-dimensional images of the backscattered light form the ocular lens. The resolution of the images could be increased with a CCD camera which contains a larger number of pixels. A more important improvement would be a camera with an increased dynamic range. Since the size of the opacities in the ocular lens are of the order of millimeters, the current resolution is sufficient for clinical studies of opacities. The optical distortions and aberrations that are inherent in the Scheimpflug camera were not corrected in the three-dimensional visualization since it is based on the rotated set of two-dimensional images. As a final test of the image fidelity of the three-dimensional visualization, the initial set of 60 rotated Scheimpflug images was carefully compared with the three-dimensional reconstruction. There was a one-to-one correspondence in the position of opacities from the two techniques.
The technique of volume rendering has the advantage of visualizing the entire (unmasked) human ocular lens. This volume rendering is best shown dynamically as a movie. In the case where this technique is not feasible two alternative techniques have been developed to present the three-dimensional reconstruction of the human ocular lens in vivo. The first technique presents the light scattering of the ocular lens as sets of orthogonal slices through the lens . The second technique presents the three-dimensional ocular lens as a sequential series of coronal slices from the anterior pole of the lens to the posterior pole . In both of these techniques the use of pseudo color to code the intensities of the scattered light are a marked advance of the previous methods which only used gray scale image to depict the intensities of light scattering in the cataract [16–17].
The techniques presented in this paper have an important advantage for cataract research over the acquisition of Scheimpflug images along a single meridian or even on a series of meridians. For time dependent studies on the progression of cataracts, or for clinical studies on drug induced regression of cataracts, it is critical to acquire scattered light Scheimpflug images from the same region in the ocular lens. This is a very difficult task to perform in a reproducible manner using the standard Scheimpflug technique. However, the technique described in this paper visualizes the full volume of the ocular lens (the masked regions behind the dilated iris are not observable in either technique). Therefore, over a prolonged period of time the observable volume of the ocular lens can be visualized and the volume and intensity of scattered light from the opacities can easily be measured. This is the most important feature of the technique of three-dimensional visualization of the living human lens in vivo.
The use of three-dimensional imaging and pseudocolor visualization of the in vivo human lens may be useful in the early detection of human cataracts, and improve our understanding of the aging process of the normal and diseased ocular lens.
This work was supported by a grant from NIH EY-06958. The author thanks K. Sasaki, S. Senft, M. Forster, J. Czégé, and the Biomedical Instrumentation Center at USUHS.
References and links
1. N. S. Jaffe and J. Horwitz, Lens and Cataract, Vol. 3, Textbook of Ophthalmology,S. M. Podos and M. Yanoff, eds. (Gower Medical Publishing, New York, 1992).
2. N. A. P. Brown and A. J. Bron, Lens disorders: a clinical manual of cataract diagnosis, (Butterworth-Heinemann Ltd., Oxford, 1996).
3. G. Q. Xiao, G. S. Kino, and B. R. Masters, “Observation of the rabbit cornea and lens with a new real-time confocal scanning optical microscope,” Scanning 12, 161–166 (1990). [CrossRef]
4. B. R. Masters, “Two and three-dimensional visualization of the living cornea and ocular lens,” Machine Vision and Applications 4, 227–232 (1991). [CrossRef]
5. B. R. Masters, “Confocal microscopy and three-dimensional reconstruction of thick, transparent, vital tissue,” Scanning Microscopy 6, 71–79 (1992).
7. B. R. Masters, “Three-dimensional confocal microscopy of the lens,” Ophthalmic Res 28(2), 115–119 (1996). [CrossRef]
8. B. R. Masters, G. F. J. M. Vrensen, B. Willekens, and J. Van Marle, “Confocal light microscopy and scanning electron microscopy of the human eye lens,” Exp. Eye Res. 64 (3), 371–377 (1997). [CrossRef]
9. N. Brown, “Macrophotography of the anterior segment of the eye,” Br. J. Ophthal. 54, 697–701 (1970). [CrossRef]
10. A. J. Bron and K. Matsuda, “Specular microscopy of the human lens,” Trans. Ophthalmol. Soc. UK, 101, 163–169 (1981).
11. A. J. Bron, “Specular microscopy of human and animal lenses in vivo and vitro,” Ophthalmology 2, 229–232 (1985).
12. M. Ayaki, W. Tung, J. K. Wolfe, T. Shinohara, N. Ibaraki, H. Oharazawa, K. Ohara, and L. T. Chylack Jr., “New non-contact specular microscope for lens epithelium visualization,” Exp. Eye Res. 65, 143–146 (1997). [CrossRef] [PubMed]
13. J. H. Massig, M. Preissier, A. R. Wegener, and G. Gaida, “Real-Time Confocal Laser Scan Microscope for Examination and Diagnosis of the Eye In Vivo,” Appl. Opt. 33, 690–694 (1994). [CrossRef] [PubMed]
14. O. Hockwin, K. Sasaki, and S. Lerman, “Evaluating cataract development with the Scheimpflug camera,” in Noninvasive Diagnostic Techniques in Ophthalmology, B. R. Masters, ed. (Springer-Verlag, New York, 1990), 282–318.
15. V. Dragomirescu, O. Hockwin, H. R. Koch, and K. Sasaki, “Development of a new equipment for rotating slit image photography according to Scheimpflug’s principle,” Interdis. Top. Gerontol 13, 1–13 (1978).
16. Y. Emori, Y. Yasuda, H. Fukuda, and K. Sasaki, “Application of image processing technique for analyzing the optical system of the eye,” in Advances in Diagnostic Visual Optics,A. Fiorentini, D. L. Guyton, and I. M. Siegel, eds. (Springer-Verlag, Berlin, 1987), 55–60.
17. Y. Sakamoto and K. Sasaki, “Computed tomographic images and three dimensional expression of crystalline lens findings from multiple slices of Scheimpflug slit images,” Ophthalmic Res 27 (suppl 1), 94–99 (1995). [CrossRef]
18. M. Levoy, “Display of Surfaces from Volume Data,” PhD dissertation, (University of North Carolina, Chapel Hill, 1989).
19. B. R. Masters, K. Sasaki, Y. Sakamoto, M. Kojima, Y. Emori, S. Senft, and M. Foster, “Three-dimensional volume visualization of the in vivo human ocular lens showing localization of the cataract,” Ophthalmic Res, 28(2), 120–126 (1996). [CrossRef]
20. B. R. Masters and S. L. Senft, “Transformation of a set of slices rotated on a common axis to a set of z-slices: application to three-dimensional visualization of the in vivo human lens,” Comput. Med. Imag. Graph. 21(3), 145–151 (1997). [CrossRef]
21. B. R. Masters, “Optical tomography of the in vivo human lens: three-dimensional visualization of cataracts,” J. Biomed. Optics, 1(3), 289–295 (1997).
22. B. R. Masters, “Three-dimensional visualization of human cataract in vivo,” German J Ophthalmol, 5(6), 532–536 (1997).