Three Dimensional Microanatomy of Cells and Tissue Surfaces: Proceedings of the Symposium on Three Dimensional Microanatomy held in Mexico City, Mexico, August 17-23, 1980

Arthur

TECHNICAL ASPECTS OF STEREOPROJECTION FOR ELECTRON MICROSCOPY

FRANK N. LOW, GREGG E. OLSON, BRUCE PERSKY and JOHN J. VAN RYBROEK,     Department of Anatomy, University of North Dakota, Grand Forks, North, Dakota, USA

Publisher Summary

Stereo techniques that create an illusion of depth in conventional micrographs are being developed for use in electron microscopy. The underlying principles of stereoscopy are simple, but their application to electron microscopy could be complex. This chapter discusses certain workable approaches that should be useful to the electron microscopists who are interested in developing skill in stereo techniques. These techniques include obtaining, mounting, and presenting stereo pairs for stereo projection or journal publication. Three dimensional perception of stereo pairs of slightly disparate images is familiar in various forms: 3-D movies, stereo-atlases, children’s toys, and the old fashioned stereopticon of Victorian days. The stereopticon is the most familiar method and can serve as a straightforward and understandable expression of the basic principles of stereo techniques. Comparable stereo perception can be obtained from the preparations that are routinely used in electron microscopy, provided three dimensions are amenable to visualization.

INTRODUCTION

Stereo techniques that create an illusion of depth in conventional micrographs are now being developed extensively for use in electron microscopy. While the underlying principles of stereoscopy are simple their application to electron microscopy can be complex. The purpose of this chapter is to explain certain workable approaches that should be useful to electron microscopists interested in developing skill in stereo techniques. These techniques include obtaining, mounting, and presenting stereo pairs for stereo projection or journal publication.

Three dimensional perception of stereo pairs of slightly disparate images is familiar to most of us in various forms: 3-D movies, stereo-atlases, children’s toys (Viewmasters) and the old fashioned stereopticon of Victorian days. The stereopticon is perhaps the most familiar method and can therefore serve as a straightforward and understandable expression of the basic principles of stereo techniques. The object chosen to be presented in stereo, let us say a cube (fig. 1A) is photographed twice from slightly different angles of incidence. These two angles must always converge toward the object along a horizontal plane. This is of great importance, as we shall see later, since their convergence must correspond to the horizontally related lines of vision extending from the eyes of the observer to the eventual stereo pair of photographs (fig. 1B). The two photographs are viewed through a simple device held in the hand of the observer. This viewer is so constructed that it presents the photograph taken from the right to the right eye of the observer and the left photograph to the observer’s left eye. Consequently the observer’s two eyes see the two slightly disparate images of the original scene exactly as photographed. When so viewed these images are relayed to the observer’s brain where they undergo the essential and critical process known as cortical fusion. It is in the brain of the observer that perception of the third dimension occurs. It must be remembered that, without proper cortical fusion, stereo presentation cannot be successful. Subjectively the stereopticon observer should not be able to distinguish the photographs from the real scene. It is as if the observer’s two eyes were located precisely at the two points from which the photographs were taken (fig. 2).

Fig. 1 Photography for stereopticon viewing. A. The chosen object, a cube, observed from above is photographed from two different angles of incidence along a horizontal plane. B. The two photographs are observed through a viewer so that only one eye of the observer perceives each photograph, still on a horizontal plane, as extrapolated along the dotted lines in the figure. The cube appears, through cortical fusion, to be a three dimensional object.

Fig. 2 Disparity of an object seen with both eyes. Each eye relays a slightly disparate image of the cube to the brain. Cortical fusion in the brain causes perception of the third dimension, but only if the disparity is horizontal with reference to the eyes of the observer.

Comparable stereo perception can be obtained from preparations routinely used in electron microscopy, provided three dimensions are amenable to visualization. This possibility prevails in scanning electron microscopy (SEM),¹ high voltage electron microscopy (HVEM),², ³ the replicas of freeze-fracturing (FF)⁴, ⁵ and, to a limited extent, in conventional transmission electron microscopy (TEM)⁶, ⁷, ⁸ at 100 kv if semithick sections or whole cells are used.

Most electron microscopes in use today produce the disparate imagery of stereo pairs by mechanical tilting of the specimen stage. After taking the first micrograph of a pair the stage is tilted a controlled number of degrees (fig. 3A). Alternatively, the disparity can be produced by deflection of the electron beam (fig. 3B). After any necessary readjustment of the field, the second micrograph is taken. The resultant disparity in the two images is essentially comparable to that obtained when photographing an external object from horizontally differing angles of incidence. But one must remember, and this is the most critical point of all, the axis of tilt around which the specimen is turned must be rendered vertical when the micrographs are mounted for stereo presentation. Effectively this renders the disparate incidence of viewing horizontal with respect to the eyes of the observer. Without this condition stereo presentation cannot occur because cortical fusion is possible only when image disparity exists along a horizontal line. Positional correction of the negatives of electron microscopic stereo pairs is often necessary because the tilting mechanisms of most instruments do not produce tilting movement around an axis vertical with respect to the eyes of the observer. A conceptual understanding of these principles is essential in obtaining successful stereo pairs.

Fig. 3 Stereo imaging in electron microscopy. A. The specimen stage can be tilted in different directions for the taking of two micrographs, each at a different angle of incidence. B. Essentially the same effect can be created by electronically altering the angle of indicence of the electron beam.

The three dimensions so frequently referred to in this chapter may most conveniently be identified with the X, Y and Z axes of solid geometry (fig. 4). These geometric axes should be associated with the finished micrographs, as they appear on a projection screen. The X axis represents the horizontal plane along which disparity of the two micrographs occurs. The Y axis is vertical and represents the axis of tilt that originally produced the disparity of the images. The Z axis extends from the projection screen toward the observer and is perceived as depth in the micrographs. Electron micrographs possess two dimensions, X and Y as above. The third, or Z dimension, becomes appreciable only through use of properly presented stereo pairs.

Fig. 4 The three planes of space as they appear on a silver lenticular screen. The X axis is the axis of horizontal disparity between the two micrographs. The vertical Y axis is the axis of tilt used to produce two disparate micrographs. The Z axis is the axis of the depth perception that is produced by cortical fusion in the brain of the observer.

Stereo projection is currently practiced in electron microscopy by two different approaches; (1) the anaglyph (or red-green) method¹, ⁹ and (2) the polarized light method¹⁰, ¹¹ Both use the basic principles described above. The anaglyph method utilizes two disparate images, one produced in red and the other in green. Three dimensional perception is accomplished by viewing through color filters, red for one eye and green for the other. The polarized light method utilizes two disparate transparencies each projected through separate polarized light systems. The images are superimposed on a silver lenticular screen and viewed through polarizing glasses. This combination of polarized projection and viewing with polarized glasses allows each eye to perceive a single image so that the third dimension is subjectively perceived. The optical system for polarized stereo projection was developed in the Department of Molecular, Cellular and Developmental Biology at the University of Colorado (Boulder) by Drs. Keith R. Porter, Lee D. Peachey, Douglas Mohr and Mr. George Wray. The purpose of this chapter is concerned with the acquisition of stereo pairs from various modes of electron microscopy and the way in which they may be applied to the polarized light method of stereo projection.

SCANNING ELECTRON MICROSCOPY

The extraordinary depth of field possible in a scanning electron microscope makes this instrument well suited to the techniques of stereo presentation (fig. 5). In our laboratory we use a Cambridge S4 without goniometer stage. Specimen control is possible in all three axes (X, Y and Z) as well as tilt (0° or horizontal to 90° or vertical) and rotation (unlimited; in either direction). Our experiences with this instrument may be adopted as typical for the acquisition of stereo pairs. To this end our procedures are described in some detail. This account assumes, of course, that the reader is familiar with routine instrument operation in scanning electron microscopy.

Fig. 5 Scanning electron microscopy of the canine subarachnoid space, STEREO PAIR. Discernable are the dura (DM), arachnoid (AM), pia (PM), left vertebral artery (LVA), pyramids (P), intracranial vessels (white arrow), anterior median fissure (black arrow), and basilar artery (*).

The operator chooses a desired field in the usual manner. There are no special requirements for position of the stage controls, but it may be convenient for the tilt control to be somewhere in the middle rather than at the ends of its traverse (0° or horizontal; 90° or vertical). The chosen field is carefully trimmed for desired pictorial quality. It is carefully focused by means of the final condenser controls and the first micrograph of the stereo pair is taken. The specimen is then tilted a chosen number of degrees, usually 5° to 9°. This can be done by increasing the reading of the tilt control (e.g. 20° to 27°). This produces upward displacement of the image, usually moving it off the viewing screen (video cathode ray tube). The original field must then be returned to the screen with the Y control and refocused with the Z control. The setting of the final condenser is not disturbed before taking the second micrograph. Use of the Y control for recentering does not correct for any change in working distance created by the tilt. This is, however, corrected by focusing the tilted image with the Z control. This method assures that the working distance and magnification remain the same for both micrographs. The procedure for recovering the original field after tilting may cause some difficulty. Recognition of the original field may be facilitated by observing the desired field at higher (or lower) magnifications before the tilting process. Simultaneous manipulation of the Y control during the tilting process should maintain the desired field on the viewing screen. Several resettings of the Y and Z controls are usually necessary to accomplish the final focus. In an alternate procedure, if tilt is produced in the opposite direction, by reducing the angular reading, then the image will move downward off the screen. The recentering and focusing are the same except that recovery of the original image requires upward (rather than the previous downward) movement of the displaced image.

Certain electron microscopes are provided with electronic beam deflection devices. Here the electron beam is deflected and the stage remains in the same position (fig. 3B). The net effect is essentially the same, in that the beam impinges on the specimen by controlled angles of incidence, each of which is separately photographed. The remainder of the techniques does not differ from the procedure followed after mechanical tilting.

The negatives of a stereo pair, after proper development and drying, are best protected by transparent sleeves. Assessment of their value depends on a considerable amount of shifting and rotation of both negatives. This procedure will otherwise cause soiling or scratching of the emulsion before the final transparencies (or prints) are made. Judgements are made through a mirror stereoscope with the negatives lying flat on the horizontal surface of a transilluminator. When making this assessment the emulsion side of the negatives should always be toward the eyes of the observer.

Stereo pair negatives obtained as described above should be assessed as follows. Remembering that the axis of tilt must be rendered vertical with respect to the eyes of the observer it is necessary to rotate the developed negatives while observing them on a mirror stereoscope until the axis of tilt is in the proper position.¹¹ Since the chosen field in our scanning electron microscope is displaced vertically on the screen, a 90° rotation of each negative (arbitrarily counterclockwise) will accomplish this. When stereo appears it is sudden and dramatic to the observer. Failure for it to appear on rotation may be due to inadvertent transposition of the negatives. If this is the case, reversal of the negatives, left to right and vice versa, brings stereo perception into proper relief.

Returning to the chosen field on the viewing screen of our microscope, it follows from the above that the top of the field (as it will eventually be seen in stereo) is the right hand side of the field on the viewing screen. With this in mind the operator should be aware of the need for repositioning the field by rotation. Some specimens require a conventional position in order to be more easily interpreted. The necessary rotation must therefore be performed before photography.

The magnification at which stereo pairs are photographed can have considerable effect on the outcome, even when all other procedures remain constant. Magnifications in the lower to middle ranges (20X to 2000X) are usually easiest to obtain. But care should be taken that no significant structure in the field is presented to the observer in foreshortening; that is, viewed at a very acute angle. If this is the case, a structure may disappear from the line of vision when tilted for the second micrograph. This can cause ineffective stereo. As higher magnifications (5000X and above) are used certain other subtle imperfections can appear in the negatives obtained. They are usually attributed to small but damaging inaccuracies in the movement of the specimen stage which are reflected in the disparity between the negatives of the stereo pair. Another difficulty that occurs at higher magnifications is the large displacement of the chosen field from the viewing screen at customary angular tilts. The higher the magnification the greater the displacement. Relocation of the same field becomes more difficult to recognize through exaggerated out-of-focus images. But successful stereo pairs at 20X to 5000X can routinely be obtained even in the presence of these difficulties.

Goniometer stages are designed so that both tilt and rotational movement occur around the center of the field on the viewing screen. The axis of tilt may be more difficult to determine. In such cases it may be determined experimentally by rotating the transilluminated negatives when observing them through a mirror stereoscope. The negatives may need to be rotated through 360°. If stereo does not appear, the negatives should be transposed and rotated again. The best stereo perception appears when the axis of tilt is vertical and the disparity is horizontal with reference to the eyes of the observer.

Tilting displacement on the viewing screen may not always take place in a purely vertical direction. Occasionally stage mechanics and microscope electronics cause other directional displacements to occur. Whatever may be the direction of image displacement the axis of tilt is always at right angles to it. It is this axis that must be rendered vertical to the eyes of the observer to produce proper stereo. On the other hand, if the image displacement is exactly horizontal the negatives do not need to be rotated. If the axis of displacement is diagonal then the rotation necessary to produce stereo will cause the edges of the negatives to be off-horizontal and off-vertical. New margins of the useable field must be cut in true horizontal and vertical directions during mounting for stereo projection. It follows from this that taking the original micrographs at lower magnification will enable trimming without sacrifice to the boundaries of the desired field. As to rotational components combined with tilting movement, these can sometimes be corrected in a stereo pair by selective rotation of one of the negatives. The ability of the observer to correct rotational movement by means of cortical fusion is very limited.

Two prime requirements govern the acquisition and use of successful stereo pairs of electron micrographs: (1) the tilting motion must be pure (unidirectional) and (2) the axis of tilt must be rendered vertical with respect to the eyes of the observer. These requirements are readily transferable to situations prevailing in HVEM and FF.

HIGH VOLTAGE ELECTRON MICROSCOPY

Transmission electron microscopes (TEM), of which the HVEM is a special example, possess a depth of field that is much greater than the thickness of the sections used in them. This circumstance makes the HVEM well suited to stereo microscopy since the thick sections used remain well within the depth of field in both positions of tilt (fig. 6). The practical sections of HVEM (0.25 μm to 2 μm or more thick) are readily penetrated by the highly accelerated beam (1000 kV in the JEM-1000) without substantial change in the resolution of micrographs. Since the resolution is much finer than section thickness (by a factor as great as 2 log units) tissue components may be analyzed with confidence in the stereo pairs of HVEM.¹²

Fig. 6 High voltage electron microscopy of the chick aorta, STEREO PAIR. Longitudinally oriented elastic fibers tend to run parallel to the long axis of the aorta.

Disparity between the two micrographs of stereo pairs is caused by a controlled angle of tilt. This is produced either mechanically or electronically, depending on the instrument used. Ideal degrees of tilt are determined empirically, usually varying from 1° to 40°. In HVEM the rules for making stereo pairs are the same as in SEM. The tilting motion must be rectilinear and its axis vertical with respect to the observer when projected on a viewing screen.

Conventional electron microscopes may also be used for stereo analysis of sectioned tissue components. Semithick sections (1/2 μm) can be used effectively in TEM at 100 kv accelerating voltage.

A technique recently developed in the laboratory of Dr. K. R. Porter involves aldehyde fixation and critical point drying of cultured cells on gold grids. The cells are neither embedded nor sectioned before viewing and remain essentially whole mounts. Study of these preparations has yielded excellent results in both HVEM³ and conventional TEM at 100 kV⁷ and are ideally suited to