You are on page 1of 8

Iournal of Microscopy. Vol. 170. Pt 3 , June 1993. pp. 229-236.

Received 10 July 1992: accepted 1 December 1992

Orthogonal-plane fluorescence optical sectioning: three-


dimensional imaging of macroscopic biological specimens

A. H. VOIE,*t D. H. BURNS* & F. A. SPELMAN*?


'Center for Bioengineering and j'Regiona1 Primate Research Center, University of Washington.
Seattle. W A 98195. U.S.A.

Key words. Fluorescence, imaging, optical sectioning, three-dimensional


reconstruction, cochlea.

Summary Important structures in the cochIea are best observed


from a number of angular orientations. This is not possible
An imaging technique called orthogonal-plane fluorescence
with histological methods.
optical sectioning (OPFOS) was developed to image the
Recently, optical imaging techniques have been devel-
internal architecture of the cochlea. Expressions for the
oped for obtaining a thin section within a specimen. Optical
three-dimensional point spread function and the axial and
sectioning is the process of moving a thick specimen
lateral resolution are derived. Methodologies for tissue
through the focal plane of an imaging detector to produce
preparation and for construction, alignment, calibration
an image of a thin region of that specimen (Castleman,
and characterization of an OPFOS apparatus are presented.
1979: Agard, 1984). There are many advantages to this
The instrument described produced focused, high-resolu-
technique. Optical sectioning maintains the register
tion images of optical sections of an intact, excised guinea-
between sections. Also, the geometrical distortion inherent
pig cochlea. The lateral and axial resolutions of the images
in histological techniques is avoided. The image of an
were 1 0 and 26 pm, respectively, within a 1.5-mm field of
object viewed in this manner will contain information
view.
from structures both within and outside the focal
region, and so will appear blurred. The unfocused
Introduction
component of the image needs to be removed or mini-
Three-dimensional (3-D) reconstructions of biological mized in order to resolve the in-focus component. Simil-
specimens and systems provide an important way to arly, greater depth discrimination is achieved by removal
study, measure and analyse form and function. Most of the unfocused information. The unfocused information
often, a 3-D reconstruction is obtained by extracting may be removed optically by use of techniques such
information from a series of 2-D histological sections of as confocal microscopy (Brackenhoff et al., 1989; Carlsson
an object. Thin histological sections are mounted on & Liljeborg, 1989; Wilson, 1989), or by image pro-
microscope slides and can be imaged with either optical cessing (Castleman, 1979; Agard, 1984; Agard et al.,
or electron microscopes. The image can provide extra- 1989).
ordinarily fine detail. However, histological sections have The literature does not discount the possibility of using a
serious disadvantages for the purpose of 3-D reconstruction confocal system to section a macroscopic biological speci-
(Castleman, 1979). One is the loss of register that occurs men opticallly. The main disadvantages with this aproach
when the sections become separated during sectioning. are the high cost of such a system and the difficulty of its
Although resulting errors can be reduced by use of fiducials, alignment owing to instrumentation tolerances. Addition-
complete correction is not possible. Another problem is the ally, low-light-level conditions, as with fluorescence micro-
geometric distortion that can result from processing of the scopy, increase the image-acquisition time, and this
slices. Finally, serial sectioning results in a single set of precludes imaging in real-time.
parallel-plane slices. For some structures, this may not yield Image deblurring by digital image processing methods
a convenient means to extract data for accurate 3-D has several disadvantages. One is the relatively poor spatial
reconstruction because there is a loss of sampling resolution resolution compared with that obtained from histological
where the boundary of a structure of interest is nearly sections (Agard et al., 1989). Also, image processing
parallel to the slicing planes. An example would be the precludes the practical viewing of the sections in real-
structures in the spiral-shaped cochlea (Santos, 1987). time. Reconstruction of spiral structures remains proble-

0 1993 The Royal Microscopical Society


230 A . H. V O I E E T A L .

measure of the planar illumination thickness


along the z-axis
measure of the planar illumination length along
the x-axis
f-number of the lens
intensity distribution function of the laser illumi-
nation beam
3-D point spread function of the imaging system
2-D image formed at back focal plane of the
imaging lens
wave number of the image-forming light
refractive index of the clearing solution
3-D structure in object space
2-D pixellation function
Fig. 1. Orientation of the illumination axis, imaging axis and speci- width of a square pixel that has been calibrated to
men for OPFOS imaging. object-space dimensions
lateral resolution for a diffraction-limited optical
matic, since existing deblurring methods require parallel,
system
regularly spaced images (Agard et al., 1989).
imaging lens blurring function
Another approach to optical sectioning involves the use
illumination beam boundary function in the x-z
of a planar illumination beam (Root, 1985; White, 1987).
plane
Planar illumination has been used in other imaging modalities
coordinates of an arbitrary point in space
such as flow cytometry and flow visualization (Merzkirch,
acceptance angle of the imaging lens
1974). The illumination beam is focused into a plane with a
illumination wavelength in air
cylindrical lens, and aligned to be co-planar with the depth of
field of the imaging detector. The beam is thus orthogonal to the
imagmg axis. The region where both the imaging system and
laser illumination focus coincide is the system focal zone. Theory
We developed an imaging system that employs planar The coordinate system and notation conventions used here
illumination to excite fluorescent dye so that we could are similar to those employed by Castleman (1979) and
produce images of optical sections of excised cochleas. The Agard (1984). The orientation of the axes is shown in Fig.
motivation for developing this system was to obtain suitable 1. The origin is taken to be the centre of the focal zone. The
2-D data to enable an accurate 3-D reconstruction of the x- and y-axes of the image plane are scaled and aligned to
cochlea. As shown in Fig. 1, planar illumination passes project onto the focal plane axes. The implicit simplification
through the cleared and fluorescently labelled specimen. is that the image magnification is unity. Also implied is that
Dye on structures within the thin focal zone emits no coordinate inversion occurs between the focal plane and
omnidirectional light, forming an image on the sensing the image plane. This allows the subscripts to be dropped
element of the camera. Only dye on structures within the from the notation, simplifying the expressions.
focal zone emits light because no excitation occurs outside For spatially invariant blurring, image formation through
this region. Since no light originates from outside the focal an optical system can be expressed as
region, the resulting image contains no out-of-focus
component. This imaging technique has been dubbed
orthogonal-plane fluorescence optical sectioning (OPFOS). where I(x, y) is the image formed with the imaging lens
The aim of this article is to describe the OPFOS technique focused at z = 0 , O(x, y, z) is the 3-D structure in object
and demonstrate its usefulness. First, the theoretical space, h(x, y, z) is the 3-D point spread function (PSF) of the
framework is developed, followed by calibration and system, and * is the convolution operator. An image results
characterization of the apparatus. An example of an from a convolution of the object with the PSF of the system.
OPFOS image of an excised cochlea is shown. Finally, a 3-
D reconstruction of one of the spiral canals, produced from
the digitally recorded images, is presented. Three-dimensional PSF
The PSF of an optical system is the product of the
List of symbols illumination intensity distribution, g(x, y, z), and the
To enable readers to understand the following descriptions imaging lens blurring function, s(x, y, z) (Hiraoka et al.,
more easily, the symbols used in this paper are listed below. 1990; Nakamura & Kawata, 1990). The PSF of the system
ing

I
Collimating lens

Fig. 2. Schematic diagram of the optical components assembled for


OPFOS. The imaging lens does not contact the clearing solution.

can be expressed as

Fig. 3. The focusing characteristics of the cylindrical lens that pro-


In most optical imaging techniques, the illumination is duce the boundary function w ( ~ )The. thickness of the beam waist
is 2a, and the focal distance of the beam is 2b.
coaxial with the imaging axis and uniform upon or through
the object. In these cases the illumination intensity (Siegman, 1986). The spatial extent, or boundary, of the
distribution is constant, and thereby the PSF reduces to spot is defined as the location at which the intensity falls to
s(xly, z). However, the illumination used in OPFOS imaging l/e2 of its maximum (in this case, x-axis) value. The locus
is spatially non-uniform and is transverse to the imaging of these points in the x-z plane is the beam boundary
axis. The illumination intensity distribution is determined function, w(x). The boundary function describes the shape
by the optical components of the illumination system. of the beam and is dependent on the focusing characteristics
of the cylindrical lens. The shape of w (x) is hyperbolic (Fig.
3) and is described by
Illumination system
An optical layout of the illumination system is shown in Fig.
2. The light source is a laser whose wavelength excites the
fluorescent dye in the specimen. A cylindrical lens with a The parameters a and bin Eq. (4) describe the dimensions
suitable f-number focuses the collimated illumination beam of the nearly planar illumination region. The thickness of
to approximately planar dimensions. The f-number of the the beam waist is a function of the laser wavelength, A, and
cylindrical lens is the lens focal length divided by the the f-number of the cylindrical lens. This thickness is 2a.
diameter of the expanded and collimated beam. The where
illumination system is aligned so that the intensity
distribution is co-planar with the depth of field of the
imaging detector.
The beam thickness increases by a factor of J 2 at an x
value of b on either side of the waist. The focal region of the
Illumination intensity distribution beam is defined to be between +b and -b. The focal distance
of the beam is the length in x of the focal region, and this
The illumination beam is characterized as an intensity
length varies with the square of the waist thickness. The
distribution in a progression of planes normal to the x-axis.
focal distance is 2b, where
The shape of the spot is circular up to the cylindrical lens.
The cylindrical lens focuses the beam so as to compress the
spot in the z-direction, but not in the y-direction. The shape
of the spot approaches a line at the focal plane of the
cylindrical lens at x = 0. with n being the refractive index of the clearing solution. It
In the focal zone, the laser spot has a Gaussian intensity should be noted that Eq. (5) is independent of n. This is
distribution. In the x-z plane, the variation in the intensity because wavelength and f-number are offsetting functions
can be expressed as of the refractive index in Eq. (5). Since there is one more
factor of X than f-number in Eq. (6), n appears in its
numerator.
232 A . H. VOIE ET AL.

Imaging system Methods


Ideally, for diffraction-limited optical components, the lens
Specimen preparation
blurring function is described by
Cochleas were excised from guinea-pigs that had been
2 (kr sin a ) deeply anaesthetized and perfused with neutral buffered
s= [
krsina ] formalin. The bulla and vestibular system were trimmed
away with dental burrs and the cochlea was cemented to a
(Inoue, 1986). is a first-order Bessel function, k is the small nylon sphere. The specimen was decalcified in
wave number (equal to 27~/X,where X is the wavelength of tetrasodium salt of ethylenediaminetetraacetic acid
the image-forming light), a is the acceptance angle of the (EDTA) and dehydrated in ethyl alcohol. It was cleared in
lens and r = (x2 + y2)1/2. This is the well-known Airy a solution of five parts methyl salicylate and three parts
pattern formed in the image plane by a point source in the benzyl benzoate (Haq, 1988). The cleared cochlea was
focal plane of the lens. The radius of the central bright spot treated with fluorescent dye by immersion in a dye bath.
of s(x, y, z ) is r,, where The bath was prepared by dissolving rhodamine isothio-
cyanate (RITC) in ethyl alcohol to a concentration of 1 mg/
0.61X ml. This solution was then diluted with clearing solution to
r, = -. a final dye concentration of 5 x mg/ml. The RITC
n sin CY
absorbs at 550 nm and emits at 585 nm (Haugland, 1981).
In a diffraction-limited imaging system, the lateral resolu-
tion is r,.
However, for digital imaging systems, the effect of Specimen holder
pixellation must be taken into account. This can be done A custom-made specimen holder was designed for mount-
by convolving the continuous-plane image with a pixella- ing the cochlea. It was fabricated from a block of Delrin. The
tion function, p(x, y). The resultant image is then described chamber had clear sides to allow the laser illumination
by beam to pass through unimpeded. This created an interior
chamber that was open on top so that it could be filled with
clearing solution. The purpose of the clearing solution was
to maintain the relative transparency of the cochlea during
For simplicity, p(x, y) is taken to be a square with sides of the imaging process. The imaging system lens does not
length go. The value of p(x, y) is unity within the square contact this fluid. A rotation shaft extended into the
and zero otherwise. chamber and was orientated in the y-axis direction. The
shaft was attached to an optical shaft encoder. This
provided a digital display of the rotation angle in 0.7"
increments. A four-part chuck was built into the end of the
Image resolution
shaft to hold the specimen for imaging. To position the
The lateral resolution of the image depends on the size of po sample, the specimen holder was attached to a stage that
relative to r,,. If po is less than ro/2, then the Nyquist could be translated with 5-pm precision along the three
criterion for spatial sampling is satisfied (Inoue, 1986). In coordinate axes.
that event, Eq. (9) approximates Eq. (1). However, for low-
magnification digital imaging systems where this criterion
is not met, the lateral resolution limit is 2po rather than ro. Design criteria
A consequence of pixel-limited lateral resolution is an The primary design criterion for the OPFOS system was to
increased depth of field. This is due to the fact that depth of obtain images of the canal cross-sections of guinea-pig
field varies inversely with lateral resolution (Sheppard, cochleas. These cross-sections were estimated to be 1.5 mm
1988). across at the widest point. Therefore, the components of
In most optical sectioning techniques, the axial resolution illumination optics were chosen to provide a beam whose
limit is the imaging depth of field (Hiraoka et al., 1990). In focal distance was 1-5mm and whose thickness was
the OPFOS method, however, the illumination beam is ,< 30 pm over this length.
focused to a thickness that is less than the imaging depth of
field. This ensures that the resulting image is entirely in
focus. Consequently, the axial resolution of the image is Illumination optics
equal to the thickness of the illumination beam, given by The illumination source was a 1.25-mW green helium-
Eq. (4). The axial resolution limit is 2a, defined in Eq. (5). neon laser with a wavelength of 543.5 nm. Using Eqs. (5)
O R T H O G O N A L - P L A N E FLUORESCENCE O P T I C A L S E C T I O N I N G 233

and ( 6 ) . we calculated that an f/26.7 cylindrical lens Reflected


liaht
would meet the design criteria. Specifically, the focal
distance was calculated to be 1.53mm, with a waist
thickness of 18.5 pm and a maximum thickness of 26.2 pm.

Imaging system
The imaging lens was an f l 0 . 9 5 Navitron macro lens with
a 25-mm focal length. This lens was chosen for its high
light-gathering capacity. The image was formed on the
sensor of a Javelin CCD video camera. A 560-nm-long pass
filter was positioned on the lens to block scattered laser light
from contributing to the image. Images were digitized with
a Gateway 2000 486 computer, equipped with a
640 x 480 pixel frame grabber from Imaging Technology.
'
Image acquisition and processing was done with Optimas
(Bioscan). Fig. 4. Configuration for beam intensity profile measurement.

then translated vertically by a fixed amount. The lens and


Calibration camera were translated along the imaging axis until the rod
A rectangular calibration grid was imaged to determine the tip image was again in focus. The screen coordinates of the
pixel dimensions. This grid was placed in the sample holder tip image were again noted to ascertain the lateral shift
and immersed in clearing solution. Incident incandescent with vertical translation. Translation was performed in both
light was used to illuminate the grid. The image of the grid positive and negative directions to measure the effect of
was produced by the scattered light. In the calibration micrometer backlash. Micrometer backlash was less than
procedure, two pixels in the image were selected with a could be measured with pixels that were 5 pm across, and
mouse. The object-space distance between these two points therefore was considered negligible.
was specified and the dimensions of the pixels were then
computed. Grid vertices were used as calibration points. The
Results and discussion
pixel x- and y-dimensions were 5.08 x 5-25pm, for an
aspect ratio of 0.968. These dimensions were used to assign
Illumination beam dimensions
x- and y-coordinate values to the pixels. These screen
coordinates were then used for further measurements. The thickness of the illumination beam was measured to
determine the axial resolution of the OPFOS image. A 45"
prism was placed in the sample holder and immersed in
Alignment
clearing solution. The laser illumination was focused onto
A small, pointed stainless-steel rod was used as an the prism, as illustrated in Fig. 4. The refractive index of the
alignment tool. This rod was placed in the sample holder glass prism was similar to that of the surrounding clearing
with its point orientated toward the camera. With the laser solution. Most of the illumination intensity, therefore, was
beam just grazing this rod tip, a small image of the spot was transmitted through the prism. A small fraction of the
formed from the scattered light. The pointed rod was illumination intensity was reflected by the prism in the
translated in the x- and y-directions; the illumination plane direction of the imaging optics. The prism was positioned so
was considered aligned with the camera focal plane and the that its inclined face transected the waist of the illumination
horizontal stage translation plane when the radius and beam. The resulting image appeared as a thin stripe
intensity of the spot remained constant within the camera's orientated in the y-direction. The stage was translated in
field of view. the x-direction and the waist of the beam was taken to be
Alignment of the translation stages with the axes of the the setting for which the width of the stripe was a
image-processing coordinates was also done with the minimum. The intensity profile was then measured across
pointed rod. The point was translated by fixed amounts in the stripe. The beam thickness was determined by
the x- and y-directions and the change in screen measuring the width of the profile at which its value was
coordinates was noted. Alignment in the z-direction was l/e2 of the maximum intensity. A key advantage of using a
determined as follows. The camera was positioned to prism to measure beam thickness was that an intensity
produce an in-focus image of the rod tip. The screen profile could be determined from a single image. This
coordinates of the tip image were recorded. The rod was substantially simplified the process of locating the beam waist.
Measured values

0.8 -

u
Q,
.N 0.4-

0 5 10 15 20 25 30 35 40 45
I.lm
Fig. 5. Normalized intensity profiles of the laser spot measured
along the z-axis at the focus of the cylindrical lens. The solid curve
represents theoretical values predicted from Eq. (2). The dashed
line indicates l/e2, which defines the beam boundary. Beam thick-
ness is determined by measuring the distance over which the pro- Fig. 7. An OPFOS image of a guinea-pig cochlea showing modiolus
6le exceeds this boundary value. The predicted thickness was (A), auditory nerve (B), basilar membrane (C), organ of Corti (D)
18.5 pm and the measured thickness was 20.0 pm. and Reissner's membrane (E). Dashed lines border the laser focal
Figure 5 shows the measured intensity profile of the zone. The dimensions of this region are 1.5 mm in the horizontal
and vertical directions. Lateral resolution in this region is 10pm:
illumination beam at the focus of the cylindrical lens. Also
axial resolution is 30 pm.
shown is the profile estimated from Eq. (3); the agreement is
reasonable. The measured beam thickness was 20.0pm, boundaries of the illumination beam in the x-z plane (Fig.
compared with an estimated thickness of 18.5 pm. 6). Also shown in Fig. 6 is the estimated beam boundary;
Beam width measurement was repeated after the prism very good agreement was achieved.
was translated 750pm in the x-direction on either side of
the cylindrical lens focus. On the side toward the laser, the
measured beam thickness was 28.3 pm; on the other side of
the focal point, the beam thickness was 27.9pm. The
estimated beam thickness at both these distances was
26.1 pm. These measurements were used to depict the

I='
Fig. 8. Normalized intensity profiles of the illumination beam mea-
I sured at the focus of the cylindrical lens. The solid curve (profile A)
-20 4 represents values from a beam that had not passed through any
-750 -500 -250 0 250 500 750
specimen material. A cleared cochlea was placed in the beam path
Distance (prn) along X axis
to assess its effect on the beam intensity profile. The dashed curve
Fig. 6. Boundaries of the laser illumination beam in the x-z plane (profile B) represents the beam profile at the cylindrical lens focus
of the focal zone. Solid lines represent measured values: dashed with the cochlea in the light path. The width of profile B was
lines represent theoretical values. 21.7pm, compared with 20.0pm for profile A.
causing the refractive index of the specimen to approach
that of the surrounding clearing solution. However, slight
differences in refractive indices may exist between the tissue and
the solution, and within the tissue itself. These differences can
result in geometric distortion of the OPFOS image.
The effect of the specimen on the imaging process was
examined by repeating the beam intensity profile measure-
ment at the beam focus with a cleared and stained cochlea
in the laser light path. The assumption was made that if the
presence of the cochlea resulted in little distortion to the
illumination-focusing characteristics, then there would be
little distortion introduced to the light from fluorescent
structures as well. Figure 8 shows the results of these
measurements. The normalized intensity profile at the beam
focus with no cochlea in the light path is labelled profile A.
Measurements made after placing the cochlea in the light
path resulted in profile B. The width of profile B is 2 1.7pm,
compared with 20.0 pm for profile A. This demonstrates that
tissue absorbance and scattering of the illumination beam
have only a slight effect on the intensity profile of the beam.

Three-dimensional reconstruction
The lower of the three canals in the OPFOS image is the
scala tympani. Its basal terminus (not shown in Fig. 7) is the
round window membrane. The cochlea was rotated about
the y-axis to image the cross-section of the scala tympani
from the round window membrane to its apical terminus.
Fig. 9. Three-dimensional reconstruction of the scala tympani of a The outline of the canal was traced and digitized for each
guinea-pig cochlea. image. The coordinates of these digitized points were
corrected for stage translation and shaft rotation. This
OPFOS image of intact cochlea correction restored the true spatial relation between the
Figure 7 is an OPFOS image of a mid-modiolar section digitized points in the successive images. Points were
through a guinea-pig cochlea. The dashed lines border a connected between the successive images to complete the
square region of interest that measures 1.5 x 1 . 5 mm. The 3-D reconstruction of the scala tympani (Fig. 9).
region of interest corresponds with the location where the
beam is focused to the dimensions described above. The Quantitative measurements
direction of the laser light is from left to right. Thin, dark
Several physical parameters may be measured directly or
horizontal bands can be seen in the image. These artefacts
calculated from the 3-D reconstruction data. These
are due to the illumination beam passing through regions of
quantitative measurements can provide important informa-
higher dye concentration, causing a shadow to appear.
tion for understanding cochlear architecture and function.
These shadows do not interfere with image interpretation
For example, the scala tympani centroid and cross-sectional
because the specimen may be rotated.
area were computed from the reconstruction data. Figure
Many fine cochlear structures are visible in the region of
1 0 shows the canal cross-sectional area plotted against
interest. The central core of the cochlea is the modiolus, in
centroid arc length. It is clear that cross-sectional area
which the auditory nerve can be seen. The basilar
decreases as a function of centroid arc length, and that the
membrane, the organ of Corti and the thin Reissner's
rate of change of this parameter is greatest in the first turn
membrane can be seen in the canal cross-sections.
of the canal. Viewing the data in this manner can provide
Reissner's membrane is two cell layers thick, or about
insights in such areas as fluid dynamics and cochlear
20 pm across.
implant design.

Effect of local variations in refractive index Conclusions


The clearing solution permeates the tissue of the specimen, We have described an imaging technique that is con-
236 A . H . V O I E ET A L .

References
Agard, D.A. (1984) Optical sectioning microscopy: cellular
architecture in three dimensions. Ann. Rev. Biophys. Bioeng.
13, 191-219.
Agard, D.A., Hiraoka, Y., Shaw, P. & Sedat, J.W. (1989)
Fluorescence microscopy in three dimensions. Meth. Cell Biol.
30, 353-376.
Brackenhoff, G.J., van der Voort, H.T.M. & van Spronson, E.A.
(1989) Three-dimensional imaging in fluorescence by confocal
Arc length (mm) scanning microscopy. 1. Microsc. 153, 1 51- 159.
Carlsson, K. & Liljeborg, A. (1989) A confocal laser microscope
Fig. 10. Cross-sectional area of the scala tympani as a function of scanner for digital recording of optical serial sections. 1. Microsc.
the canal centroid arc length. 153, 171-180.
Castleman. K.R. (1979) Digital lmage Processing. Prentice-Hall.
ceptually uncomplicated and relatively inexpensive to Inc., Englewood Cliis, NJ.
construct. The OPFOS technique allows convenient Haq. N. (1988) Application and evaluation of an imaging technique for
imaging of a much larger specimen than is generally thiee dimensional cochlear reconstruction. Master's thesis. Uni-
possible by other optical sectioning methods. It provides a versity of Washington.
long working distance and a wide field of view. Addition- Haugland, R.P. (1981) Handbook of Fluorescent Compounds.
ally, the specimen may be translated and rotated during the Molecular Probes. Inc., Junction City. OR.
imaging process. Because of this there is little loss of Hiraoka, Y..Sedat, J.W. & Agard. D.A. (1990) Determination of
three-dimensional imaging properties of a light microscope
sampling resolution for a complex structure between
system. Biophys. 1. 57, 325-333.
images. This is particuarly useful for making quantitative Inoue, S. (1986) Video Microscopy. Plenum Press, New York.
measurements and 3-D reconstructions of complex biologi- Merzkich, W. (1974) Flow Visualization. Academic Press. New
cal structures. York.
The depth range of this imaging technique is limited Nakarnura, 0 . & Kawata, S. (1990) Three-dimensional transfer-
primarily by the extent to which the clearing solution can function analysis of the tomographic capability of a confocal
render a specimen transparent. The success of the clearing fluorescence microscope. J. Opt. Soc. Am. A7, 522-526.
process is in turn dependent on the decalcification step. Root, N. (1985) Light scanning photomacrography-a brief
Large bony structures (e.g. vertebrae) could be expected to history and its current status. 1. Biol Photog. 53, 69-77.
be more difficult to decalcify than smaller bony structures Santos, M.P. (1987) A computer assisted morphometric method for 3-
such as the cochlea. Specimens that contain little calcium dimensional morphological reconstruction of the cochlea. Master's
thesis. University of Washington.
could be expected to clear even more easily, so it may be
Sheppard. C.J.R. (1988) Depth of field in optical microscopy. 1.
possible to image larger dimensions. Microsc. 149, 73-75.
Siegman, A.E. (1986) Lasers. University Science Books. Mill Valley.
Acknowledgments CA.
White, W. (1987) Photomacrography: An Introduction. Butterworth
This research was supported by National Institutes of Publishers, Boston, MA.
Health grants RR00166 and DC002 74 and by a grant from Wilson, T. (1989) Three-dimensional imaging confocal systems. 1.
the Bloedel Hearing Research Center. Microsc. 153, 161-169.

You might also like