The resolution of an optical microscope is considerably less in the direction of the optical axis (z) than in the x-y plane, typically by a factor of 5 or more. This is true of conventional or confocal microscopes. In order to alleviate this problem we have used multiple tilted views to supply the 'missing data' and thus increase the resolution in z. A special tilting microscope stage has been constructed, which allows specimens mounted on thin glass capillary tubes to be rotated through large angles. Through-focal data sets can then be collected from several mutually tilted viewing directions. The relative orientation, translation and z-sampling parameters for the data sets can then be determined by use of a novel phase/wiener cross-correlation function. Finally, the data sets, once brought to a common coordinate system, can be combined using amplitude and phase combination techniques in Fourier space borrowed from X-ray crystallography. We have applied this technique to metaphase chromosomes in intact embryos of Drosophila melanogaster. Images were collected using a liquid nitrogen cooled CCD array camera, and processed using a VAX computer. As judged from the maximum extent of significant power in the Fourier transform, the resolution of the final reconstruction was about 0.25,urn in x and y and better than 0.4,um in z. Although we have shown an application to data collected from a conventional fluorescence microscope, the technique is equally applicable to any other imaging mode or to confocal images.

The development and operation of the digital imaging microscope, a system capable of analyzing the 3D molecular distribution in singe cells is described. It is a system consisting of hardware and software that: 1. obtains 2D or 3D microscope images of the faint signals from fluorescent molecules introduced into single cells; 2. reverses the distortion in such images introduced by the optics; 3. automatically or semi-automatically extracts features of interest from such images, and; 4. displays 2D, 3D images and even higher order information for interactive analysis. The strategies used to accomplish these tasks are discussed and their application to the analysis of the molecular events underlying white blood cell chemotaxis is illustrated.

Recent technological advances now make it practical to record three-dimensional data from biological specimens using fluorescence light microscopy. When three-dimensional images are collected using a conventional microscope, each observed section contains in-focus information from the parts of the sample at the focal plane and out-of-focus information from the remainder of the sample. The imaging process can be characterized as a convolution of the sample with the point spread function (PSF) of the microscope. We have experimentally determined the PSF for an epifluorescence microscope using high numerical aperture oil and water immersion lenses. Several methods for the processing of the observed data are discussed, with the best results obtained by constrainediterative deconvolution methods.

Due to dramatic advances in optical microscope technology, it is now possible to examine biological specimens in three-dimensions with either electron microscopy or light microscopy. Although the algorithms used for generating the three-dimensional reconstructions are well known, much work remains on developing methods for the display and analysis of the resultant complicated three-dimensional data. In the first part of this paper, we will introduce image processing schemes for enhancing features in data and some computational methods for manipulating three dimensional data sets. In the second part, we will discuss the image display system in the context of software design and hardware requirements, which must be considered for convenient data visualization and measurement. Much effort has been spent on developing a generalized display system that can also be used for model building and analysis. By model building we mean the interactive tracking of features in three-dimensional volumetric images. Throughout this chapter we will make a distinction between display methods that directly utilize three-dimensional image data stored as a contiguous set of pixels (also called volumetric data, or voxel data) and those that convert the data into a set of polygon vertices that define a single-level contour surface. Although this latter approach contains much less information than volumetric approaches, it can make stunning pictures. Figure 1 shows a typical procedure for handling three-dimensional data involving image enhancement, data manipulation, data display and analysis.

In this paper, we present a new method for reconstruction of refractive index field of a transparent medium from multi-frame intcrfcrograms obtained by several interferometers in different viewing directions. It is based on double cubic many-knot interpolating splines and A RT techniques. It can offer significant improvement in the calculational accuracy and combine various constrains. The details of the method are discussed in the paper. Experimental results show that the method is effective.

Continuing advances in VLSI technology have made possible image sensors with high resolution. We have developed an ultra-high-resolution image sensor for industrial and scientific applications. The ultra-high-resolution sensor is a full-frame CCD imager consisting of 2048 x 2048 pixels, and measures 1.84 cm x 1.84 cm. The pixel size is 9 microns x 9 microns. The sensor has dual readout registers to increase the data rate. The sensor could be operated in the single or dual readout register mode. In this paper we present the architecture and results of the four-megapixel image sensor.

Improvements in silicon CCD sensitivity in the NIR by using thick, high resistivity epitaxy silicon, backside illumination, and antireflection (AR) coatings are discussed. Quantum efficiencies at 900 nm of up to 42% for frontside illuminated devices and 78% backside illuminated and AR coated devices are reported.

The design and performance of two very high resolution CCD imagers will be described. Both devices utilize a 7.5 micron pixel and are fabricated with three-level polysilicon to achieve high yield. Dynamic range, charge transfer efficiency, spectral response and successful imagery are demonstrated.

Large area image sensors using Charge-Coupled Device (CCD) technology for scientific applications are often thinned and backside illuminated to improve their overall sensitivity and particularly their short wavelength response. The usual method of fabrication is to start with a p-type (high resistivity) epi-layer on a 1)4-type (low resistivity) substrate and chemically etch the device until there is a thin membrane of the desired thickness at or near the p-p+ interface. Variations in the doping concentration at the final back surface because of nonuniform thinning often lead to poor pixel-to-pixel uniformities. Pixel-to-pixel variations of > 100% are not uncommon for incident light at wavelengths of 350 nm on poorly thinned devices. We have developed a thinning process that reduces the pixel-to-pixel variations to < 5% while maintaining a quantum efficiency (QE) of > 40% at 350 nm, when the back surface is charged. Two scientific CCDs are currently made with thiA process, a 404 x 64 array of 52 x 52 gm pixels and a 1200 x 400 array of 27 x 27 gm' pixels. Test results of this process on these two devices will be presented.

A 512 x 512 monolithic PtSi Schottky Barrier Diode (SBD) array that was designed, fabricated and tested at EG&G Reticon will be discussed. The array uses a novel line-addressed charge-accumulation (LACA) CCD in an interline configuration for the vertical registers and a multiple-readout register for high speed readout. The LACA readout structure has two distinct advantages over the conventional interline CCD structure. First, the LACA register occupies a very small area while still providing enough charge handling capacity • for the Schottky barrier detector. As a result, the 30 gm x 30 gm pixels have a fill factor of 54% when designed with a 2.5 Lim minimum dimension CCD process. Secondly, since the register operates in a charge accumulation mode, it has an extremely high effective charge accumulation efficiency. In the horizontal register, four readout structures are used to increase the overall pixel readout rate. A 75 Hz frame rate is achieved when the horizontal clock rate is 5 MHz. A CCD bending structure is utilized at each output to eliminate any coupling from the common clock lines. Each output has an electron sensitivity of greater than 5 0/e. The quantum yield of the platinum silicide SBDs has been between 15 and 25%. The chip dimensions are 748 mil x 748 mil, of which 604 mil x 604 mil is photosensitive.

Charge-coupled devices (CCDs) are becoming increasingly used as the primary detector for many astronomical imaging and spectroscopic instruments. The major limitations of front-illuminated CCDs in this field are their lack of blue and UV spectral response and relatively poor visible light response. Thinned, back-illuminated CCDs can achieve significantly higher quantum efficiency but typically have a warped imaging surface, non-uniform flat-field response, and often do not respond well to backside charging techniques to eliminate the backside potential well. We describe here a novel thinning/polishing technique which produces an optically fiat imaging surface independent of the epitaxial/substrate interface location. Chemical/mechanical polishing can be used to thin a CCD when the final device thickness is greater 20pm or possible a wafer to any thickness. We discuss the effect of silicon thickness and thinning technique on surface warpage. We also discuss the bump bonding mounting method which we are beginning to use to improve the yield of the thinning process, the ease of backside charging and antireflection coating, and the ability to produce mosaics of CCDs for large focal planes such as the new generation of ground-based telescopes.

The importance of ultra-low light level imaging is now generally recognized in a variety of applications. The purpose of this paper is to describe briefly some pioneering applications of image intensification to observations in physics, biology and biophysics.

Ultrasound is used extensively in medical applications for diagnosis, therapy and even surgery. For these applications, the acoustic pressure may be delivered in short pulses at low duty cycles, in long pulses at high duty cycles, in continuous waves, or in the form of high intensity shock waves. The acoustic frequencies vary from about 20 kHz to higher than 10 MHz. If the acoustic pressure amplitude exceeds about 1 MPa, even for microsecond length pulses, then acoustic cavitation can occur in aqueous liquids. The inception of acoustic cavitation has traditionally been detected by measuring acoustic emissions from the cavitation field, such as a shock wave or a subharmonic. However, for short pulses and for high frequencies the acoustic emissions are difficult to detect. In most instances of cavitation bubble collapse, light is also emitted from the cavitation complex via a process called sonoluminescence, in which the internal temperature of the gas is elevated to incandescent levels. One of the classic papers in this area is one coauthored by George Reynolds 1, the subject of this memorial session. This paper extends his pioneering work, describes the phenomenon of sonoluminescence, introduces some new information intending to clarify its physics, and demonstrates how it can be used to assess possible risks associated with the use of medical ultrasound.

Excited states are a way of life on earth. The absorption of the free energy of photons from the sun produces excited states of the antenna pigments of photosynthesis and this "electronic energy" is converted to "chemical energy" for reduction and phosphorylation of pyridine nucleotides. This drives the plant-DNA-replicating engines and the animal (predator) DNA-replicating engines which we call living organisms. Photons captured by other pigment systems provide signals for pnototaxis, morphogenesis and vision, information for the engines to interact with the environment and with one another. Evolutionary pre-biotic photochemical synthesis selected for stable tetrapyrrole'and carotenoid structures whose electronic properties were coincidently ideal for excited state photosensitization, energy transfer and electron transfer. These basic structures are ubiquitous in all living organisms. In this paper we shall be concerned with biological chemiluminescence, the production of excited states of molecules in the cells and tissues of living organisms, from which photons are emitted as the consequence of chemical reactions with quantum efficiencies ranging from 10^-15 to unity.

The IDG law-light video microscope system was designed to aid studies of localization of subcellular luminescence sources and stimulus/response coupling in single living cells using luminescent probes. Much of the motivation for design of this instrument system came from the pioneering efforts of Dr. Reynolds (Reynolds, Q. Rev. Biophys. 5, 295-347; Reynolds and Taylor, Bioscience 30, 586-592) who showed the value of intensified video camera systems for detection and localizion of fluorescence and bioluminescence signals from biological tissues. Our instrument system has essentially two roles, 1) localization and quantitation of very weak bioluminescence signals and 2) quantitation of intracellular environmental characteristics such as pH and calcium ion concentrations using fluorescent and bioluminescent probes. The instrument system exhibits over one million fold operating range allowing visualization and enhancement of quantum limited images with quantum limited response, spectral analysis of fluorescence signals, and transmitted light imaging. The computer control of the system implements rapid switching between light regimes, spatially resolved spectral scanning, and digital data processing for spectral shape analysis and for detailed analysis of the statistical distribution of single cell measurements. The system design and software algorithms used by the system are summarized. These design criteria are illustrated with examples taken from studies of bioluminescence, applications of bioluminescence to study developmental processes and gene expression in single living cells, and applications of fluorescent probes to study stimulus/response coupling in living cells.

Recently developed high photon yielding fibers that are coupled to position sensitive photomultipliers provide good angular precision and good energy resolution in detecting gamma-rays. The UTD/UCLA High Resolution Gamma Ray Telescope consists of a scintillating plastic fiber Compton or Pair-Production Converter, a time projection gas chamber and scintillating fiber calorimeter. The calorimeter is designed to cover 7c-ray energies from 1 MeV to > 20 GeV. The space-based telescope will be used for detecting i-rays from point-like astrophysical sources and diffused cosmic 1-rays. We present here the description of the telescope, and some preliminary results from a prototype Compton converter.

The design of Gallium Arsenide photocathodes has remained almost static since the pioneering work of Antypas et al in 1973/74(1). This is partly because the GaAs photocathode development has been driven by military requirements, and once production has started, the design is frozen to enable products to be standardised. Recent developments in metal organic chemical vapour deposition (MOVPE) enable new cathode structures to be designed. In our case, we are concerned to produce a photocathode with high quantum efficiency over a wide spectral band, to enable us to build photomultipliers and image intensifiers for general scientific applications. This paper describes some of the cathode structures which have been grown at Sheffield University and Epi Materials Ltd. The paper gives data on morphology, photoluminescence and other material parameters. The paper also gives some of the early data on the photo sensitivity and spectral response together with an overview of some of the devices that can be made with these photocathodes.

The results of a computer simulation of restoring truncated fluorescence micrographs is presented. The particular application that is considered is that of three-dimensional (3-D) optical sectioning with a conventional microscope and a photon-counting camera.1,2 Practicalities that are considered in the simulation include quantum-photon noise, the degrading effects of the missing cone in the 3-D optical transfer function (OTF), and the effects of a truncated image. The restoration algorithT used, based on maximum-likelihood estimation, has been presented previously in a nonregularized4form. Herein we present an adaptation of a regularization method developed by others for positron-emission tomography. The restored images indicate that it may be feasible to use this method for optical sectioning with a conventional microscope. We suggest that this might provide an economical alternative to optical sectioning with a confocal microscope.

Electroluminescent lamps can be used as a convenient stable source of luminescence over at least a three thousand-fold range of light intensity. The electroluminescence (EL) output increases with increasing applied voltage. Using a Quantex 3 x 3 inch EL lamp with a green-emitting phosphor, we found that as little as 10 VAC can be used to elicit EL and that the EL emission spectrum remained constant over a wide range of applied voltages (λ re- 508 nm; FWHM = 75 nm). Resolution measurements can be made under conditions of high or low light intensities by placing a resolution target over the EL lamp and manipulating the voltage, or by using a neutral density filter to achieve the desired light intensity, or by combining both procedures to achieve the desired light intensity and image content. The absolute EL flux, per unit area of the lamp, can be determined using the ferrioxalate chemical actinometer, corrected for light transmission and the corrected emission spectrum of the lamp. The EL lamp can thus also be used as a tertiary radiometric standard for quantification, as well as for resolution testing.

Procaryotic and eucaryotic expression vectors which contain a marker gene for selection of transformants linked to genes encoding bacterial luciferase for detection of promoter activated gene expression in vivo were used to transform the appropriate host organisms and drug resistant colonies, cells, or calli were obtained. Bacterial luciferase expression was measured by a luminescence assay for quantitative determination of promoter activation. The cellular localization of bacteria inside the host plant cell cytoplasm was achieved in a single infected plant cell based on the light emitting ability of the genetically engineered bacteria. In addition, the bacterial luciferase marker gene fusions were used to monitor cell type, tissue, and organ specific gene expression in transgenic plants in vivo. To monitor physiological changes during ontogeny of a transformed plant, low light video microscopy, aided by real time image processing techniques developed specifically to enhance extreme low light images, was successfully applied.

We have engineered a two subunit luciferase enzyme into a single functional polypeptide chain using site specific mutagenesis. We have determined, using low light video imaging, that the activity of this novel enzyme is similar to wild type luciferase when synthesized at low temperatures (15-20°C), but that it is sensitive in vivo to higher temperatures. We have used the gene encoding the monomeric bacterial luciferase as a gene marker in prokaryotic and eukaryotic organisms. Combined with low light video image analysis, it is a practical and powerful tool for quantitatively monitoring gene expression in vivo.

Digital imaging fluorescence microscopy of living cells is a valuable method for studying the dynamics of cellular processes. We are interested in distinguishing different pathways for uptake of fluorescently labeled compounds by living cells on the basis of the differences in their kinetics and location within the cell. The rate constants are calculated on a pixel by pixel basis from a series of images taken at intervals over the course of the experiments. This calculation is meaningful only if, during the time course of reaction, the pixels remain spatially invariant. Because the cells are alive during the experiment, the cells changes shape and position in the interval between image acquistion and thereby destroys the spatial invari-ability of the pixels. Quantitative studies of uptake processes in cells are limited to well behaved systems, i.e., nonmoving or fixed cells. Such experiments with fixed cells provide estimates of the rates of passive uptake processes, part but not all of the information needed to intrepret the image data about the active transport processes. It is obvious that, for quantitat-ive interpretation of uptake studies, it is essential to perform a geometric correction operation, by which each cell image in the time series is mapped to the same shape in order to restore spatial invariability of pixels. The objective of this paper is to develop and test the numerical method for mapping the changes in position and shape of the cell back to the initial geometry. Subsequently, uptake rates or other pixel by pixel based calculations can be performed and the results correlated with cell structure. We have used the recursive procedure based on two dimensional polynomials of arbitrary degree using the least squares method, and as a final step, the interpolation method by Akima. There is a significant improvement in pixel by pixel calculations after correction, although some problems, notably edge effects, still remain.

The revival of light microscope techniques and instrumentation has had major impact on biological studies of the mechanism of cell movements. This paper surnarizes some of the contributions of the participants in this session to this important area of biological science and gives a brief literature review of their past work.

The combination of the light microscope with modern electronic imaging, storage, and processing devices has brought about a virtual revolution in microscopy. Dynamic structures in living cells can now be visualized with a clarity, speed, and resolution never before achieved in differential interference. contrast OTC or Nomarsky), fluorescence, polarized light, dark field, and other modes of microscopy (Fig. 1); the gliding motion, and growth and shortening, of individual molecular filaments of microtubules and factin can be followed in real time, directly on the monitor screen, (Fig. 2); and the changing concentration, and distribution. of ions and specific protein molecules can be followed, moment by moment, in physiologically active cells.1,2 In polarized light and phase contrast microscopy, optical sections as thin as 0.1 um are now attainable (Fig. 3). These images can be viewed as through-focal stacks or stereo pairs, revealing 3-dimensional architecture of biological fine structure at very high. resolution (Fig. 4). The basic principles and methods of application of video microscopy are discussed in inoue,3 and recent developments have been summarized in a conference proceedings4 (see also ref. 5,6,7).

Traditional biochemical assays of cellular messengers require grinding up thousands or millions of cells for each data point. Such destructive measurements use up large amounts of tissue, have poor time resolution, and cannot assess heterogeneity between individual cells or dynamic spatial localizations. Recent technical advances now enable important ionic signals to be continuously imaged inside individual living cells with micron spatial resolution and subsecond time resolution. This methodology relies on the molecular engineering of indicator dyes whose fluorescence is strong and highly sensitive to ions such as Ca2+, H+, or Na+. Binding of these ions shifts the fluorescence excitation spectrum of the corresponding indicator. The ratio of excitation amplitudes at two wavelengths measures the free ion concentration while canceling out intensity variations due to nonuniform cell thickness or dye content. By rapidly alternating between the two ion-sensitive excitation wavelengths, a fluorescence microscope equipped with a low-light television camera and digital image processor can produce dynamic images of intracellular messenger levels. In many populations of cells traditionally assumed to be homogeneous, we find that neighboring individual cells can differ enormously in their cytosolic Ca2+ response to agonist stimulation, some ignoring the stimulus, others raising cytosolic Ca2+ transiently, others showing oscillations. Oscillations have been speculated to be important as a basis for frequency-coding of oscillations. Oscillations have been speculated to be important as a basis for frequency-coding of graded inputs; we are investigating the mechanism of their generation using light flashes to generate pulses of intracellular messengers. Spatial gradients of cytosolic Ca t+ within single cells have been observed in embryos during fertilization and development, neurons exposed to electrical or drug stimulation and in cytotoxic T lymphocytes during killing of target cells. Variations of cytosolic pH and free Na+ are especially interesting in epithelia and in cells undergoing mitogenic or oncogenic activation.

The recent application of digital imaging technologies to the analysis of video images from a DIC (differential interference contrast) light microscope has given 1-2 nm precision for the measurement of movements of submicron particles at 30 Hz. [Gelles et al., Nature 331(1988)450]. This technique has been applied to the study of the movements of 1) motor proteins (kinesin and cytoplasmic dynein) on microtubules (MTs) and 2) membrane glycoproteins in cellular plasma membranes. In both cases the nanometer measurements have provided important new insights into the molecular mechanisms which govern those biological movements.

Our laboratory is concerned with understanding the dynamic structure of the plasma membrane with particular reference to the movement of membrane constituents during cell locomotion. We employ digitized fluorescence microscopy (DFM) alone or in combination with fluorescence recovery after photobleaching (FRAP) to investigate individual cells. DFM is really a new form of light microscopy in that the distribution of individual classes of ions, molecules, and macromolecules can be followed in single, living cells. By employing fluorescent antibodies to define antigens or fluorescent analogs of cellular constituents as well as ultra-sensitive, electronic image detectors and video image averaging to improve signal to noise, fluorescent images of living cells can be acquired over an extended period without significant fading and loss of cell viability. FRAP allows the measurement of translational mobility of membrane and cytoplasmic molecules in small regions of single, living cells.

A video rate confocal reflected light microscope with no moving parts has been developed. Return light from an acousto-optically raster scanned laser beam is imaged from the microscope stage onto the photocathode of an Image Dissector Tube (IDT). Confocal operation is achieved by appropriately raster scanning with the IDT x and y deflection coils so as to continuously "sample" that portion of the photocathode that is being instantaneously illuminated by the return image of the scanning laser spot. Optimum IDT scan parameters and geometric distortion correction parameters are determined under computer control within seconds and are then continuously applied to insure system alignment. The system is operational and reflected light images from a variety of objects have been obtained. The operating principle can be extended to fluorescence and transmission microscopy.

Confocal scanning microscopes have improved depth discrimination over conventional microscopes. The out-of-focus contributions in a through-focus series of images are greatly reduced by the confocal geometry but not completely removed. Simulations have shown that posteriori iterative constrained deconvolution can produce depth discrimination equal to the lateral diffraction limit. However, such deterministic algorithms perform poorly when the optical sections are noisy, such as occurs in ultralow-light fluorescence microscopy. In this paper we present a stochastic approach to 3-D reconstruction and discuss the relationship of the microscope's imaging properties to the postprocessing task.

The problem of microscopic image formation with optically thick, highly reflecting objects is examined. The optical system considered is a focussed beam scanning microscope with an S-polarized source. The surfaces studied, assumed perfectly conducting, consist of steps, ridges and grooves. Our approach is only two-dimensional but otherwise quite rigorous. The boundary conditions for the electromagnetic field on the surface are determined by solving numerically an integral equation and the resulting images are compared with those found using the Kirchhoff approximation.

A new microscopic technique will be presented for imaging surface topography and the locally varying surface tension of the object. With this technique it is possible to image the locally varying chemical composition of the specimen surface on a microscopic scale because the surface tension depends on the chemical composition. The imaging technique can be described as follows: By a simple preparation technique a thin (thickness several microns) liquid layer (e.g. immersion oil), is placed on the surface of the specimen. The resulting surface tension forces the boundary of the liquid layer to move. As the surface tension is a function of the location the boundary is modulated according to the magnitude of the surface tension at each place. Thus registering the shape of the moving boundary of the liquid layer at equidistant time intervals yields information on the specimen surface. The shape of the moving boundary is detected by a light microscope with differential interference contrast in combination with an image analysis system suited for real-time processing of image sequences in a threshold detection mode.

In this paper we present a new auto-microscope system controlled by microcomputer for image processing. The microstructure of objects can be observed and processed. A charge-coupled device (CCD) array is used as the photoelectric detector. This auto-microscope system can complete autofocusing, recognizing or screening out pattern in the field of view and automatic measuring for photomicrographing by the light intensity information sampling by CCD array. For this auto-microscope system we designed the scanning mechanism of microscope stage, adjustable holder of CCD in 5-Dimention, autofocusing mechanism and a auto-photomicrograph camera with large cassette. The autofocusing discrimination function constructed by means of self entropy is better suitable for low contrast image produced by microscope system. The geometrical features of the microscope image are selected for composing the discrimination function and this discrimination function is used for classifing the patterns in the microscope field of view by training samples. Automatically measuring light for photomicrograph is realized by the same CCD array and the controlling circuit, interface circuit and other circuits of this system are developed. Some experiments results: autofocusing, auto-screening the cell pattern in the microscope field of view are given.

By observing the specular reflectance of monochromatic light from the anterior and posterior boundaries of the tear film on the front surface of a contact lens, optical conditions can be met which allow the observation of interference fringe patterns. Such patterns permit the thickness distribution of the pre-lens tear layer to be seen over a sizable area of the curved lens surface. Since the resulting interference pattern is analogous to a topographical map of the tear film thickness, but which is changing rapidly with time, continuous observation and recording of the patterns allow detailed information to be obtained concerning the time course and dynamics of tear film flow, thinning, and breakup during the interblink periods associated with normal blinking. With a contact lens of a given material, evaporation of the thin anterior tear layer was thought to be the major factor causing lens drying. Use of the tear film interferometer indicates that while this is usually the case for soft, hydrophilic lenses, both hard and rigid gas permeable (RGP) lenses have much shorter tear film breakup times largely independent of evaporative effects. The prelens tear fluid can be seen to be drawn off as a continuous sheet by the fluid in the tear menisci along the upper and lower lid margin due to surface tension differences, as long as continuity exists between the prelens tear film and either meniscus. This meniscus effect acts more rapidly than thinning by evaporation, and appears to be the primary factor determining the length of time hard or RGP lens surface will remain wet during interblink periods.

This paper reports a laser-beam scanning system applicable for the imaging of the human eye fundus. The system uses an AOD for the horizontal deflector of the high scanning frequency corresponding to the NTSC standard television. The optical dispersion due to the AOD is compensated with an electronic signal processor and the colbr imaging is achieved. The color variations due to fundus layers of the human eye are observable under illumination of low light.

We describe the application of confocal laser scanning microscopy for three-dimensional measurements of the eye in vivo. Formation and analysis of three-dimensional images are discussed using the example of the optic nerve head. The accuracy of height measurements at this structure is better than 50 microns and can be improved significantly by using active optical components.

In order to diagnose certain ocular diseases such as macular edema and atrophy, and to monitor their therapy closely, an accurate measurement of the retinal thickness is essential. We have developed a method which is capable of providing a quantitative thickness profile along a 2 mm length on the retina. A preliminary evaluation of the instrument was performed in normal human subjects. With this method, the retinal thickness can be visualized in a region extending from the optic disc to the fovea and the reproducibility of the measurement was found to be 8% (32 microns).

A laser scanning confocal microscope was used to study the structure of human donor eyes and enucleated rabbit eyes. Reflected light confocal images were obtained with a Leitz water immersion objective (50X, NA 1.0). A drop of bicarbonate Ringer's was placed between the objective and the tissue to optically couple the tissue. The confocal microscope was used to image the following objects within the eye: superficial epithelial cells, super basal and basal epithelial cells, basement membrane, stromal nerve plexus, nerve fibers, nuclei and cell bodies of stromal keratocytes, cell processes of stromal keratocytes, Descemet's membrane, and the endothelial cells. In addition, the ocular lens and excised retina were imaged. The confocal microscope produces images of the eye with the following enhanced features: increased lateral resolution, decreased depth of field, and increased contrast of transparent ocular structures. It is concluded that confocal imaging systems are an improvement over traditional optical instruments, and they may develop into a new tool for basic visual science and clinical ophthalmology.

In 1986, we adapted the Tandem Scanning Reflected Light Microscope of Petran and Hadraysky to permit non-invasive, confocal imaging of the living eye in real-time. We were first to obtain stable, confocal optical sections in vivo, from human and animal eyes. Using confocal imaging systems we have now studied living, normal volunteers, rabbits, cats and primates sequentially, non-invasively, and in real-time. The continued development of real-time confocal imaging systems will unlock the door to a new field of cell biology involving for the first time the study of dynamic cellular processes in living organ systems. Towards this end we have concentrated our initial studies on three areas (1) evaluation of confocal microscope systems for real-time image acquisition, (2) studies of the living normal cornea (epithelium, stroma, endothelium) in human and other species; and (3) sequential wound-healing responses in the cornea in single animals to lamellar-keratectomy injury (cellular migration, inflammation, scarring). We believe that this instrument represents an important, new paradigm for research in cell biology and pathology and that it will fundamentally alter all experimental and clinical approaches in future years.

Confocal slits offer certain advantages over confocal pinholes for optical sectioning microscopy, particularly with non-laser illumination. One-dimensional scanning simplifies the instrumentation and does not generate scan lines in the image. Slit widths can be adjusted during an experiment, in order to optimize the optical section depth and image illuminance. The optical sectioning effect provided by the out-of-focus effect can be enhanced by separating the illuminating and imaging light paths through the objective lens and within the specimen except for the region of the focal plane. Cellular structures are observed in vivo in the cornea and in the cochlea.

Scanning laser ophthalmoscopy is an opto-electronic technique to investigate the posterior segments of the human eye. Current instrumentation, which is based on a confocal optical set-up, helps to improve image formation with the given imperfections of contrast transfer through the optical system of the eye. The optical characteristics of a prototype instrument will be discussed. Applications will be presented to demonstrate the effect of focussing and confocal filtering on retinal imaging with and without presence of pathology.

Indocyanine green dye angiography has made possible visualization of choroidal blood flow in the human eye; however, to date, its clinical utility has been limited. The complex, multilayered, vascular structure of the choroid and an overlying layer of densely pigmented tissue combine to produce angiographic images of low contrast that are difficult to interpret. Conventional image processing can enhance individual images of the blood vessels, but such an approach contributes no information about the dynamics of blood flow through them. In this paper, an image-processing approach will be discussed that focuses on the dynamic information present in an angiographic sequence. Related topics such as high-speed digital acquisition and low-light-level ophthalmic imaging will also be discussed.

The cornea is the major refractive element in the eye; even minor surface distortions can produce a significant reduction in visual acuity. Standard clinical methods used to evaluate corneal shape include keratometry, which assumes the cornea is ellipsoidal in shape, and photokeratoscopy, which images a series of concentric light rings on the corneal surface. These methods fail to document many of the corneal distortions that can degrade visual acuity. Algorithms have been developed to reconstruct the three dimensional shape of the cornea from keratoscope images, and to present these data in the clinically useful display of color-coded contour maps of corneal surface power. This approach has been implemented on a new generation video keratoscope system (Computed Anatomy, Inc.) with rapid automatic digitization of the image rings by a rule-based approach. The system has found clinical use in the early diagnosis of corneal shape anomalies such as keratoconus and contact lens-induced corneal warpage, in the evaluation of cataract and corneal transplant procedures, and in the assessment of corneal refractive surgical procedures. Currently, ray tracing techniques are being used to correlate corneal surface topography with potential visual acuity in an effort to more fully understand the tolerances of corneal shape consistent with good vision and to help determine the site of dysfunction in the visually impaired.

Laser scanning tomography can be used to quantitatively assess the morphometry of the retina of the human eye. In this paper, the confocal imaging mode of the Laser Tomographic Scanner (LTS) is applied to evaluate the thickness of the nerve fiber layer. The contrast of nerve fiber layer images can be enhanced by polarization dependent imaging modes, like Fourier - ellipsometry (F.E.) and differential interference contrast (DIC) imaging. To improve the depth resolution, an active-optical focusing system is employed.

Analysis of the optical abberrations of the cornea using Two-Wavelength Holography and Computer Generated Holographic Testing will be discussed. Examples of surfaces from living human corneas will be shown. Special instrumentation that has been developed for this measurement will be presented with data and analysis on the surfaces of corneas and test surfaces.