Microscopic investigations of thinly cut fixed tissue sections and living cells adhered to glass substrates routinely produce superb high-resolution images when employing plan apochromat or fluorite objectives having a high numerical aperture. However, a significant amount of current biological research involves the investigation of cellular dynamics inside living tissue where important events can occur deep within the specimen, far away from the cover glass. Attempts to image cellular details and activities at a micrometer distances from the specimen cover glass with conventional oil immersion techniques often suffer from artifacts, including severe optical (spherical) aberration. The use of water in place of oil, as the immersion medium, is an effective approach to overcoming the aberration problems, and highly corrected water immersion objectives have been introduced by several manufacturers for applications involving living cells and tissues.
Recent technological advances in instrumentation and software systems, coupled to the development of new fluorochrome probes, have combined to significantly advance the frontiers of knowledge in living cell and tissue studies. The primary optical and quantitative imaging techniques utilized for these investigations include confocal and multiphoton microscopy, differential interference contrast (DIC), and traditional widefield epi-fluorescence methods. A fundamental issue in living cell studies is that proper maintenance requires the cells be surrounded in a chamber or vessel with an appropriate nutrient physiological solution, and regions or events of interest are often located 50 to more than 200 micrometers away from the cover glass. Several investigators have discussed the limitations of employing high numerical-aperture oil immersion objectives for imaging focal planes that are not immediately adjacent to the cover glass. The most severe deficiencies identified are reduced resolution and image intensity, and these artifacts become significant at distances from the cover glass of more than approximately 15 micrometers. Spherical aberration caused by the mismatch of refractive indices in the optical path is the primary cause of the image deterioration, and this aberration increases proportionally with imaging depth.
The primary rational for utilization of immersion fluids is to realize the maximum numerical aperture of the objective, providing improved diffraction-limited spatial resolution. However, a more practical advantage is that the occurrence of spherical aberration is made less likely. Because of the typical refractive index of optical glass employed for microscope objectives, the highest ultimate optical performance is currently provided by oil immersion objectives. Under ideal imaging conditions, the best optical performance is achieved by use of immersion oil that exactly matches the refractive index of the objective front lens element and cover glass. Substitution of water or another immersion medium having a higher or lower refractive index degrades this performance. However, in situations that are non-ideal, in which spherical aberration becomes the limiting factor for image quality, the use of lower-index immersion fluids can often be advantageous. Introduction of aqueous media into the optical path of an oil immersion system increases spherical aberration, and the benefits achieved by employing a water immersion objective result simply from reduction of the most significant and limiting aberration under the prevailing imaging conditions.
Theoretical Optical Considerations
One of the most widely employed methodologies in living cell and tissue research is conventional widefield fluorescence microscopy, in which high numerical-aperture oil immersion objectives are used to observe relatively flat specimens immediately adjacent to the cover glass. Fixed tissue sections or cells are embedded in an aqueous medium having a different (lower) refractive index than that of the cover glass. Light exiting the objective traverses the immersion oil and the cover glass (which by design have the same refractive index), and is refracted upon encountering the interface with the embedding medium, having a different refractive index. The net effect of the light deviation at the refractive-index interface depends upon the depth of the observation plane. At focal planes near the cover glass, the objective optical design requirements are met, and the system performs quite well.
The introduction and development of techniques such as confocal and multiphoton microscopy has altered imaging requirements from the conventional situation by enabling study of much thicker specimens, and the three-dimensional visualization of large specimen volumes, often involving some form of image deconvolution or reconstruction. The ability to image optical sections at large depths within a specimen, far removed from contact with the cover glass, changes the optical properties of the system in that the light traverses a medium that was not intended by design. Optical correction is optimized for the light distribution in a homogeneous medium adjacent to the cover glass, although in practice, the observation volume may be some distance away, and the distribution of light is dramatically affected by refractive index differences and by the distance of the focal plane from the glass. The resulting deviations cause a loss of resolution and signal intensity, and a focus shift. The reduction in resolution and image brightness that occurs in many instances when the plane of observation is moved from the specimen-glass interface to deeper regions in the specimen was noted in the mid-1980s by Stefan W. Hell and others, and became the subject of much research aimed at interpretation of the confocal fluorescence image.
Illustrated in Figure 2 are the comparative optical situations for which the oil immersion objective functions ideally (Figure 2(a)), and for which a refractive index mismatch between the immersion oil and an aqueous medium results in serious image aberrations (Figure 2(b)). Imaging a specimen plane that is separated from the cover glass by a region of aqueous medium is representative of the optical conditions that prevail in thick biological specimen studies, and the image deterioration encountered is a primary incentive for the employment of water immersion objectives.
The ideal optical situation for employment of a plan apochromat oil immersion objective is established when the specimen is in direct contact with the cover glass, ensuring a homogeneous refractive index (nominally 1.515) throughout the light path from the focal point at the surface of the specimen, through the cover glass, into the immersion oil, and continuing into the front element of the objective. In this configuration (Figure 2(a)), no refraction of the light waves occurs and the full numerical aperture of the objective is utilized. Furthermore, lens aberrations are well controlled and the resulting images exhibit maximum resolution and contrast.
As the focal plane is adjusted toward deeper regions within the cell or tissue section, or if the specimen lies beneath a layer of physiological medium, the light path traverses refractive index interfaces or gradients between the specimen (n = 1.35), the aqueous physiological saline solution (n = 1.33), and the cover glass-immersion oil-objective combination (n = 1.515). Refraction of the light rays occurs at each refractive index interface, with the result that the full numerical aperture of the objective is not realized, and optical aberrations are introduced as the objective departs from its design criteria. Refraction causes bending of the light rays toward the optical axis as they pass from the aqueous medium into the higher-refractive-index glass, effectively limiting the maximum numerical aperture of the objective (see Figure 2(b)). Introduction of water into the light path of an oil immersion system utilizing a 1.4 numerical-aperture objective, for example, reduces the effective numerical aperture to a maximum value of 1.33. The oil immersion objective cannot meet its design performance when a specimen is viewed through a layer of water or physiological medium, and the spherical aberration introduced results in severe adverse effects in the image. These effects increase proportionally with depth in the specimen, as the distance of the focal point becomes farther from the lower surface of the cover glass.
In studies involving the imaging of living cells utilizing oil immersion, spherical aberration becomes a limiting factor in image quality. The proportional increase of the aberration with imaging depth in the cellular material or the aqueous media surrounding the cells manifests itself as diminished intensity and contrast that prevents the resolution of smaller specimen details. It has been experimentally demonstrated that the effects of this distortion are sufficient to cause misinterpretation of specimen structures such as cilia in marine organisms. The application of deconvolution methods to mathematically compensate for distortions of this nature is one possible solution, although accurate measurements of the point spread function (PSF) are necessary, and accomplishing this becomes problematic when the point spread function is distorted axially as well as transversely.
In all of the methods of three-dimensional microscopy, which are increasingly being applied in investigations of living cells and other non-embedded biological specimens, aberrations in the point spread function are significant because of the relatively low refractive index of the specimens. Point spread function distortions can have particular significance in confocal microscopy since the spherical aberration that is induced impairs the very abilities that are the primary advantages of confocal microscopy: elimination of out-of-focus information to increase contrast and effective resolution in the x-y plane, and the creation of high resolution x-y-z optical sections. To the extent that refractive index mismatch is the cause of image aberrations, the employment of water immersion objectives should greatly enhance high-resolution depth dependent imaging of low refractive index specimens.
In microscopy of a three-dimensional specimen, the image data may be considered a representation of the specimen that contains aberrations, or has been blurred, by the three-dimensional point spread function. If deconvolution methods are to be utilized to reconstruct the object from the aberrant image data, an accurate determination of the point spread function is required. Both direct measurement and computational approaches are employed to describe the point spread function, and each technique has advantages and disadvantages. When experimental comparison is made of data acquired with an oil immersion system and with water immersion, using either measured or computed three-dimensional point spread functions, focal plane-dependent spherical aberration is markedly reduced in the water immersion system.
A comparison of imaging ray traces that occur between the specimen and immersion objectives (both oil and water) is presented in Figure 3, along with a geometric model prediction of an x-z image for a fluorescent bead 100 micrometers below the interface in aqueous media. In Figure 3(a), the sphere on the left represents the actual shape of the fluorescent bead, while the elongated shape on the right is the impression provided by imaging with an oil immersion objective. Ray traces (yellow arrows in Figure 3) from the bead into the front lens of an oil immersion objective (Figure 3(b)) and water immersion objective (Figure 3(c)) reveal details about how the refractive index mismatch obscures actual specimen geometry. In aqueous media, the bead is distorted into an apparent elongated oval when imaged with an oil immersion objective as illustrated in Figure 3(b), but remains spherical when using a water immersion objective (Figure 3(c)). The actual specimens are represented by a blue sphere and the apparent images are indicated as a red oval or red sphere.
In order for three-dimensional image data is to be a reliable representation of the true specimen (convoluted by the point spread function), it must be known that the point spread function does not vary with axial or transverse focus shifts. In practice, this may not be the case if the conditions for which the objective is designed are not met. The objective's design criteria are a function of the refractive indices of specimen, the immersion medium, and the thickness and refractive index of the cover glass. When the refractive index of the specimen and immersion medium match, aberrations are minimized for any specimen thickness, because focusing to deeper planes in the specimen medium combines an increased optical path length in that medium with a compensating reduced path length in the immersion medium. The two effects of the stage displacement offset one another so that the object and image conjugate surfaces of the lens system coincide, as required for image formation without aberrations. The oil immersion objective, therefore, meets the design requirements when utilized to image embedded specimens, but exhibits significant deviation in the point spread function along the optical axis when employed to image low-index biological specimens. In a system in which the point spread function varies axially, the point spread function would have to be computed for each image plane in a three-dimensional stack, requiring a complex model for computing the variance, as well as an image-processing algorithm more powerful than deconvolution. These considerations provide strong justification to utilize water immersion objectives for investigation of three-dimensional biological specimens, in an effort to minimize the axial variance of the point spread function, and the resulting image aberrations.
Design and Performance of Water Immersion Optics
One of the basic functions of any immersion objective, one that requires a fluid other than air between its front lens element and the specimen, is to increase the numerical aperture of the system. In utilizing an oil immersion objective, it initially appears that the thickness of the cover glass would be of little importance, since its refractive index approximately matches that of the immersion fluid. This is essentially true as long as the specimen is mounted in Canada balsam or another medium with refractive index similar to that of the cover glass. When a specimen is mounted in an aqueous medium such as physiological saline, having a refractive index significantly different from that of the glass and immersion oil, the optical performance changes considerably. Consequently, focusing through an aqueous layer even 10 micrometers thick can introduce severe image aberrations, due to asymmetry in the point spread function with respect to the focal plane (see Figures 2 and 3). Unless the specimen region under observation is in direct contact with the cover glass, the optical assumptions utilized to correct the lens aberrations in oil immersion objectives are not valid.
As this behavior of oil immersion objectives became more apparent to investigators, and the limitations imposed on the emerging techniques of three-dimensional imaging in the study of living cells and tissues were recognized, several microscope manufacturers began introducing well-corrected high numerical-aperture water immersion objectives in the mid-1990s. With plan apochromatic correction, and numerical apertures of approximately 1.2, the water immersion objectives have somewhat lower numerical aperture values than the comparable oil immersion lenses, but add the critical capability of allowing high-resolution imaging through aqueous layers on the order of 200-micrometers thickness. Although the principal advantage of water immersion objectives is improved imaging capabilities in thick preparations of low refractive index biological specimens, other practical benefits are derived from the use of water as the immersion fluid. Water has no inherent fluorescence to complicate image interpretation, there is little risk of contaminating physiological solutions, aqueous solutions do not require special cleanup methods, and the cost is negligible.
A highly corrected 60x plan apochromat water immersion objective produced by Nikon was developed with consideration of specifications first suggested by Shinya Inoue, and is representative of similar objectives introduced by other manufacturers. The objective features a 1.2 numerical aperture and a working distance of 220 micrometers, affording it the potential to image focal planes at this depth within a water-borne specimen (see Figure 1). A correction collar allows adjustment to accommodate cover glass thickness ranging from 0.15 to 0.18 millimeters, an essential feature for elimination of spherical aberration. Additionally, the objective exhibits high transmission and chromatic aberration correction from the near ultraviolet through the red visible spectral regions, and therefore can be utilized both for confocal microscopy and for conventional fluorescence and differential interference contrast (DIC) techniques.
As previously discussed, homogeneous immersion would ensure that light rays are not deflected on their path through the specimen and immersion media until reaching the rear surface of the objective's first lens element. If refractive index interfaces are eliminated, an objective can be designed to achieve diffraction-limited performance throughout its entire focusing range. Application of water immersion with low-index specimens eliminates the problem of the higher refractive index with immersion oils, but if the water immersion objective is used with a cover glass, a difference in refractive index between the glass and water is introduced (Figure 2(c)). The exact refractive index and thickness of the cover glass becomes crucial in achieving maximum resolution, and is the reason that many water immersion objectives include a correction collar for compensation of varying cover glass properties. Optical plastics may also be useful in reducing refractive index mismatches between the mounting medium and the immersion fluid for water objectives. Plastic cover "glasses" having a refractive index in the range between 1.35 and 1.4 should significantly reduce the refraction angle of imaging light rays that traverse from the specimen to the objective front lens element through an aqueous medium.
Practical evaluation of water immersion objectives has been carried out by a number of investigators, utilizing different means of assessing the benefits of this type of system in certain applications. Based on theoretical considerations on the impact of spherical aberration on point spread function, clear advantages of water immersion over oil immersion techniques for biological studies is predicted, particularly for specimen planes located some distance away from the cover glass. The experimental results have largely supported theoretical predictions, demonstrating significant improvements in the ability to image deep into aqueous specimens, compared to results with oil immersion objectives.
One experimental evaluation of the performance of a 60x water immersion plan apochromat objective of 1.2 numerical aperture, as described above, in comparison to a 60x (1.4 numerical aperture) plan apochromat oil immersion objective, was conducted by imaging a test target and a highly detailed diatom at various distances below the cover glass (see Figure 4). The results were consistent with theoretical predictions: the oil immersion objective produced images with excellent resolution and high contrast only when the test specimens were situated in direct contact with the cover glass, and exhibited severe contrast degradation when the targets were imaged through an 84-micrometer water layer. The water immersion objective produced slightly lower resolution and contrast, compared to the oil immersion objective, when the target specimens were in contact with the cover glass, but the image quality was maintained with essentially no degradation when the water layer was added to the imaging path.
Presented in Figure 4 are central portions of a star test target imaged with a standard 60x apochromat 1.4 numerical aperture oil immersion objective (Figures 4(a) and 4(b)) compared to the results with a 60x apochromat 1.2 numerical aperture water immersion objective (Figures 4(c) and 4(d)). The specimen was prepared either with water (Figures 4(b) and 4(d)) or without water (Figures 4(a) and 4(c)) between the cover glass (170 micrometers thickness) and the test target. In Figure 4(a), the target image was captured with the oil immersion objective and no water between the test target and cover glass. The contrast vanishes at a radius of 2.3 micrometers from the center, which corresponds to a spacing of approximately 0.2 micrometers. The central black disk has a diameter of 1.2 micrometers to provide a size reference. When the same objective is utilized to image the target with 84 micrometers of water between the test target and cover glass, the contrast is seriously degraded and spacings below 0.4 micrometer become invisible (Figure 4(b)). In comparison, when the test target is imaged by the water immersion objective with no water between the target and cover glass, the contrast disappears for spacings below 0.24 micrometers (Figure 4(c)). When an 84-micrometer layer of water is placed between the target and the cover glass (similar to the oil immersion objective discussed above), the contrast remains high (Figure 4(d)) and the performance of the water immersion objective is not compromised by the additional water layer between the cover glass and test target.
Quantitative evaluation of the performance of the two objectives is illustrated by graphs of the contrast transfer function for test targets consisting of equally spaced light and dark bars of various spatial frequencies (Figure 5). The contrast transfer graphs illustrate the amount of target contrast (as a percentage) that the optical system is capable of transferring from the target (specimen) to the image. An image that maintains the full contrast of the target for a given spatial frequency would be plotted as 100 percent on the graph, representing perfect contrast transfer by the system. As contrast deteriorates at higher spatial frequencies, it eventually is reduced to zero at a particular line spacing, which can be taken as the absolute limit of resolution for the optical system being evaluated. Each graph illustrates contrast transfer functions produced under several conditions: with no water layer between the cover glass and the test grating, and with different water-layer thickness added. In addition, theoretically calculated contrast transfer functions are plotted for aberration-free objectives of corresponding numerical aperture. Data for up to 153 micrometers of water is presented for the water immersion objective, while 50 micrometers is the maximum water thickness illustrated for the oil immersion objective.
As illustrated in Figure 5, the water immersion objective delivers contrast and resolution values nearly equivalent to the theoretical limits, and maintains its performance when water layers of 80 and 153 micrometers are added between the target specimen and coverslip, a simulation of the situation encountered in imaging deep within aqueous material such as living cells or tissue. In contrast, the oil immersion objective exhibited a 50 percent reduction in resolution limit, and severe degradation in contrast when tested with only 50 micrometers of water overlying the target. The performance declines steeply with increasing spatial frequency.
Additional evaluation has demonstrated the ability of the high numerical-aperture plan apochromatic water immersion objective to obtain high quality images at a depth of 220 micrometers in water, a feat that would simply not be possible using an oil immersion objective. Other investigations have performed measurements of the water immersion objective point spread function, which support the test target performance reported, and further illustrate the benefits of improved symmetry of the function above and below the plane of focus. The measurements demonstrate that depth-dependent distortion can be modeled and corrected, enabling the lens to be used for accurate measurements along the z-axis for determination of vertical resolution. The fact that the point spread function of the objective is symmetrical above and below the focal plane (indicating minimal spherical aberration) allows it to match the theoretical axial resolution calculated for an objective of its numerical aperture. One overall benefit of this optical performance is a significant improvement in image deconvolution methods applied to three-dimensional specimens, as compared to the same techniques utilizing oil immersion objectives. Furthermore, the essential elimination of spherical aberration in the water immersion objective results in improved signal collection and image brightness when imaging at depths of more than approximately 20 micrometers in aqueous media or tissue.
Special Aspects of Confocal Microscopy
Principal benefits of confocal methods include controlled restriction of the focal plane thickness to allow optical sectioning, and improved resolution and contrast by elimination of flare from signals arising outside the image plane. These two factors combine to permit x-zscan images that provide three-dimensional representations of thick specimens. Spherical aberration limits these capabilities, and increases proportionally with depth in the specimen when the refractive index of the specimen differs from that of the immersion fluid. If an oil immersion objective is used with an aqueous specimen, approximately one-third-wave spherical aberration is added for every micrometer focus depth below the cover glass. Small amounts of spherical aberration cause an expansion of the point spread function and an equivalent loss in axial resolution (see Figure 6). The large degree of aberration that accumulates if focusing beyond about 10 micrometers into a low-index specimen produce considerable blurring of the point spread function and loss of contrast in the image. If spherical aberration is not eliminated, sharpness and contrast losses override any benefit in the confocal approach when imaging at depths of more than approximately 15 micrometers from the cover glass. The utilization of water immersion objectives offers substantial benefits in elimination of these problems when imaging aqueous specimens such as live cells.
Spherical aberration resulting from mismatched refractive indices can distort optical data to the extent that morphological misinterpretation and errors in dimension measurement can occur. The well-known distortion of specimens in three-dimensional microscopy manifests itself as an elongation of features along the optical axis (z-axis). A number of techniques have been employed to measure and to computationally simulate the effect, but contradictions exist over both the magnitude and exact cause. The artifact has been experimentally verified, however, and is known to produce an axial elongation of an object causing it to appear to be up to three times its actual size (Figure 3). The anomaly depends upon the immersion conditions, and is thought to be caused by the fact that axial stage movements do not result in a direct equivalent displacement of the focal position. Errors in the estimation of distances and volumes occur, having major implications in all forms of three-dimensional quantitative microscopy. Among the factors that have been demonstrated to play a role in the distortion effect are the refractive index mismatch between the embedding or surrounding medium and the immersion fluid, the specimen size, the distance from the cover glass, and the numerical aperture of the objective. Utilization of water immersion when imaging low-index specimens such as biological material will lessen the effect, although under some conditions it will not be completely eliminated. Cellular material typically varies in refractive index between about 1.33 and 1.39, and consequently some refractive index mismatch may still exist, even when water is used as the immersion medium.
In addition to dimensional scaling errors when a refractive index mismatch occurs, a significant effect on signal intensity can be induced by the same distortions of the point spread function. In many confocal system configurations, the illuminating pinhole that is scanned across the specimen is utilized in the detection path as well, being scanned by the same mechanism with the purpose of excluding all out-of-focus light from the detector. When imaging deep within a specimen utilizing oil immersion objectives, the severity of spherical aberration can cause sufficient focus shift that much of the light emitted by the fluorophores in the specimen cannot pass through the pinhole to the detector. Therefore, most of the emitted signal from regions of the specimen removed from the cover glass is lost before reaching the confocal detector. The focus shift induced by the spherical aberration is accompanied by a loss of intensity in the acquired image, and the decrease continues geometrically with distance into the specimen.
Published theoretical and experimental analyses confirm that when a 1.3-numerical aperture oil immersion objective is used, imaging a fluorescent plane 20-micrometers deep within an aqueous medium results in a detected peak intensity that is 40 percent less than that from a plane at 10-micrometer depth. This concept is illustrated in Figure 6, which displays contour plots of the confocal point spread functions of a high numerical aperture oil immersion objective, and their respective axial responses for several imaging depths in water. The ideal point spread function (no spherical aberration) is presented in Figure 6(a), while those for imaging depths of 5, 10, 15, and 20 micrometers into aqueous media are illustrated in Figures 6(b)-6(e), respectively. Reduction or elimination of spherical aberration through application of a high numerical-aperture water immersion objective is an effective approach to maintaining adequate signal level in high-resolution fluorescence microscopy.
One often-overlooked advantage of using water immersion objectives in confocal techniques is that water is much less viscous than most immersion oils, and consequently exerts less force (surface tension) on the cover glass during focusing, when the objective and specimen preparation are displaced relative to each other. The cover glass is, therefore, less likely to flex and possibly displace the specimen when focus is changed during acquisition of a confocal z-series. Minimizing specimen movement during the repeated refocusing that is required during optical sectioning can result in sharper and more meaningful three-dimensional reconstructions from the image stack.
Recently, water immersion objectives have been experimentally demonstrated to be suitable for multiple-objective techniques such as 4Pi confocal microscopy and theta microscopy. The axial resolution achieved in 4Pi confocal microscopy is similar to that of near-field optical techniques, and is accomplished by the combination of coherent focused spherical wave fronts from two opposing high-aperture objectives. The coherent addition of two spherical wave fronts results in increased aperture along the optical axis, and a narrower point spread function minimum. This technique has produced the highest three-dimensional far-field resolution obtained to date, which is on the order of 100 nanometers in combination with image reconstruction.
Prior to the development of high-numerical aperture water immersion objectives, the reliance on oil immersion had limited 4Pi confocal microscopy to glycerol-mounted specimens. The refractive index of glycerol (1.47) is sufficiently close to that of immersion oil (1.51), so that minimal compensation is required for the phase shift during axial scanning. A large portion of cellular studies involve glycerol-based mounting media, and at least one manufacturer has developed a high numerical-aperture glycerol immersion objective to minimize the degradation of data resulting from the oil-glycerol refractive index mismatch. Designed for use with a quartz cover glass (refractive index 1.46), the lens incorporates an aberration correction collar that accommodates glycerol concentrations between 72 and 88 percent. This objective has been successfully applied in three-dimensional fluorescence microscopy, and should simplify 4Pi microscopy of glycerol-mounted specimens.
In the case of imaging at depth into water or physiological solution, however, severe spherical aberration and phase shifts do not allow 4Pi microscopy to be carried out with oil immersion or glycerol immersion objectives. Consequently, oil-immersion 4Pi methods are not applicable to live cell imaging. The high-aperture water immersion objectives developed to minimize the spherical aberration induced by refractive-index-mismatch distortions in conventional confocal and multiphoton imaging offer the same advantages in 4Pi methods applied to live cell studies. Although the water immersion objectives have lower numerical aperture than comparable oil immersion lenses, several studies have demonstrated that they produce favorable point spread function characteristics, which allow a fundamental improvement of axial resolution in three-dimensional imaging of living specimens utilizing 4Pi microscopy.
Mel Brenner and Stanley Schwartz - Bioscience Department, Nikon Instruments, Inc., 1300 Walt Whitman Road, Melville, New York 11747.
Thomas J. Fellers and Michael W. Davidson - National High Magnetic Field Laboratory, 1800 East Paul Dirac Dr., The Florida State University, Tallahassee, Florida, 32310.