Direct-View High-Speed Confocal Scanner: The CSU-10

Shinya Inoué*; Ted Inoué    Marine Biological Laboratory, Woods Hole, Massachusetts
Universal Imaging Corporation, West Chester, Pennsylvania
*This author is a consultant to Yokogawa Electric Corporation.

I Introduction

Over the past decade, confocal microscopes have dramatically improved our ability to examine the structural and functional detail of biological tissues and cells. The same characteristics that made these advances possible, namely the ability of confocal microscopes to provide exceptionally clean serial optical sections that are free of out-of-focus flare, have also made it possible to directly generate tilted sections, as well as to generate striking three-dimensional (3-D) images.

Alternative methods, such as the application of computationally intensive deconvolution algorithms to serial sections (obtained with nonconfocal, or wide-field, fluorescence microscopes) can also yield clean optical sections and reconstructed 3-D images (e.g., Agard, 1984; Holmes et al., 1995).

More recently, two- (and multi-) photon optics have been used to further improve optical sectioning, especially for deep, living tissues. With only the fluorophores lying in the focal plane being exposed to the coincident action of the double- (multiple-) wavelength excitation wave, fluorescence is strictly limited to that plane. Meanwhile, the longer wavelength excitation wave dramatically reduces light scattering in general and photobleaching of the fluorophore that lie outside of the focal plane.

Although many of these confocal and deconvolution approaches yield exquisite optical sections and their composites, it may take from many seconds to considerably longer for each two-dimensional image covering a reasonable area to become available.

The CSU-10 described in this chapter is a disk-scanning, direct-view confocal scanning unit.1 As detailed in the next section, the CSU-10 incorporates, in addition to the main Nipkow disk, a second disk with some 20,000 microlenses that are each aligned with a corresponding pinhole on the main Nipkow disk, thus substantially improving the transmission of the confocal illuminating beam. Thus, one can view confocal fluorescence images in real time, i.e., at video rate or faster, through the eyepiece or captured through a video camera.

The compact unit, attached to an upright or inverted microscope, transforms a research microscope into an exceptionally easy-to-use and effective, direct-view epifluorescence confocal microscope (Fig. 1). Thus, optical sections of fluorescent specimens show with high resolution, and in true color, directly through the ocular as one focuses through the specimen. Sparsely distributed, weakly fluorescent objects are readily brought to view.

In addition to viewing the image through the ocular in real time, the confocal image generated by the CSU-10 can be directly captured by a photographic camera or a video or CCD camera attached to the C-mount on top of the unit.

Since images of the microscope field are scanned at 360 frames/s by the multiple arrays of pinholes on its Nipkow disk, the CSU-10 not only provides very clean, full-frame video-rate images but, by use of high-speed intensified cameras, can even capture frames in as short an interval as 3 ms or less.

In this paper prepared in 1999, we describe the basic design of the CSU-10; some sample applications including video-rate and higher speed confocal imaging; the low rate of fluorescence bleaching observed using the CSU-10; rapid digital processing for further haze removal and image sharpening and for generation of dynamic stereo images; mechanical and optical performance of the CSU-10; and potential future applications. Addendum A provides some recent updates.

II Overview of Confocal Microscopes

In a point-scanning confocal microscope, the specimen is illuminated through a well-corrected objective lens by an intense, reduced image of a point source that is located in the image plane (or its conjugate) of the objective lens. Light emitted by the focused point on the specimen traces the light path back through the objective lens and (after deviation through a dichromatic mirror) progresses to an exit pinhole. The small exit pinhole, which is also placed in the image plane of the objective lens, effectively excludes light emanating from planes in the specimen other than those in focus. Thus, confocal imaging selectively collects signals from the focused spot in the specimen with dramatic reduction of signals from out-of-focus planes.

To achieve an image of the specimen with a point-scanning confocal system, either the specimen itself is moved in a raster pattern or the confocal spot and exit pinhole together are made to scan the specimen. In other words, either the specimen is precisely raster scanned in the x–y plane, or the illuminating and return imaging beam are made to raster scan a stationary specimen by tilting these beams at the aperture plane of the objective lens (commonly with galvanometer-driven mirrors) (see, e.g., this volume, Chapter 1, and Pawley, 1995). The very rapidly changing intensity of light passing the exit pinhole is detected by a photomultiplier tube and captured in a digital frame buffer. The output of the frame buffer, the confocal image, is displayed on a computer monitor.

For a number of reasons, it generally requires a few seconds to generate a low-noise, full-screen confocal fluorescent image with a point-scanning confocal system (see, e.g., Amos and White, 1995). One way of overcoming this speed limitation is to use many points to scan the specimen in parallel, such as by the use of a Nipkow disk.

In Nipkow-disk-type confocal microscopes, the disk with multiple sets of spirally arranged pinholes is placed in the image plane of the objective lens. The pinholes are illuminated from the rear, and their highly reduced images are focused by the objective lens onto the specimen. As the Nipkow disk spins, the specimen is thus raster scanned by successive sets of reduced images of the pinholes.

The light emitted by each illuminated point on the specimen is focused back, by the same objective lens, onto a corresponding pinhole on the Nipkow disk. This exit pinhole may be the same pinhole that provided the scanning spot, as in the Kino-type confocal system (Kino, 1995), or may be one located on the diametrically opposite side of the Nipkow disk as in the Petráň-type confocal microscope (Petráň et al., 1968; see this volume, Chapter 1, Figs. 8 and 9).

With any Nipkow-type confocal system, one needs to maintain a moderately large separation between adjacent pinholes relative to their diameters in order to minimize crosstalk (i.e., leakage) of the return beam through neighboring pinholes. On the other hand, with the pinholes separated by, say, 10 times their diameter, only 1% of the incoming beam is transmitted by the Nipkow disk since the pinholes occupy only that fraction of the disk area. Nipkow disk systems, therefore, generally tend to suffer from low levels of transmission of the beam that illuminates the specimen.

In addition, a major fraction of the illuminating beam can be backscattered by the Nipkow disk and contribute to unwanted background light in the Kino-type arrangement. The Kino system thus includes several design features to minimize this source of unwanted light (see, e.g., Kino, 1995). The Petráň, tandem-scanning-type does not suffer from backscatter of the illuminating beam but does require very high precision of pin-hole placement on the Nipkow disk as well as stability of rotational axis since the entrance and exit pinholes are located on opposite sides of the axis of rotation of the disk.

III Design of the CSU-10

In order to overcome these difficulties encountered with the past Nipkow disk systems, the Yokogawa CSU-10 uses the same pinholes for the entrance and exit beams but is equipped with a second, coaxially aligned Nipkow disk that contains some 20,000 microlenses. Each microlens is precisely aligned with its corresponding pinhole on the main Nipkow disk onto which the illuminating beam is focused. Thus, instead of the 1% or so found with conventional Nipkow disk systems, some 40% to 60% of the light impinging on the disk containing the microlenses becomes transmitted through the pinholes to illuminate the specimen (Fig. 2).