The computed tomography imaging spectrometer (CTIS) is a
non-scanning instrument capable of simultaneously acquiring full spectral information (450-750 nm) from every position element within
its field of view (75 µm × 75 µm). The current spatial and
spectral sampling intervals of the spectrometer are 1.0 µm and 10 nm,
respectively. This level of resolution is adequate to resolve signal
responses from multiple fluorescence probes located within individual
cells or different locations within the same cell. Spectral imaging
results are presented from the CTIS combined with a commercial inverted
fluorescence microscope. Results demonstrate the capability of the CTIS
to monitor the spatiotemporal evolution of pH in rat insulinoma cells
loaded with SNARF-1. The ability to analyze full spectral information for two-dimensional (x, y) images allows precise evaluation
of heterogeneous physiological responses within cell populations. Due
to low signal levels, integration times up to 2 s were required. However, reasonable modifications to the instrument design will provide
higher system transmission efficiency with increased temporal and
spatial resolution. Specifically, a custom optical design including the
use of a larger format detector array is under development for a
second-generation system.
 |
INTRODUCTION |
Fluorescence microscopy has developed into a
powerful analytical tool in such areas as biology, cell physiology, and
medicine (Wang and Herman, 1996
). The basis for this
power is the development of stable and sensitive analysis devices and
fluorescent and luminescent probes (Haugland, 1999;
Mason, 1999). Although current technology affords the
opportunity to view many biological processes, the full advantage of
fluorescence microscopy with simultaneous high temporal, spectral, and
spatial resolution is just now fully emerging.
The objective of imaging spectrometry is to measure the proportions of
colors associated with narrow and contiguous spectral bands at every
position in an imaging instrument's field of view (FOV). The resultant
data set is referred to by several names: an object cube, an image
cube, a hypercube, or a data cube. Instrumental techniques currently
available to acquire an object cube are 1) a filtered camera or
multiple cameras, each with its own spectral filter (Wiegmann et
al., 1993; Garini et al., 1999); 2) a
monochromator configured for imaging, that is, equipped with an imaging
array in the exit-slit plane (Eng et al., 1989;
Richmond et al., 1997); 3) a Fourier-transform
spectrometer equipped with an imaging array (Garini et al.,
1999); 4) a tunable illumination source (Zangaro et al.,
1996; Zeng et al., 1999); 5) a rotating Risley
prism chromatomograph (Mooney et al., 1997); and 6) an
acousto-optic tunable filter (AOTF) or liquid-crystal tunable filter
(LCTF) combined with a CCD camera (Morris et al., 1994;
Wachman et al., 1997; Ornberg et al.,
1999). These approaches all require scanning in position or in
wavelength in order to record the entire (x, y, 
) object cube. Alternatively, rather than acquiring full spectral information, data can be recorded over a limited region of the spectrum
(Wiegmann et al., 1993).
In this paper a computed-tomography imaging spectrometer (CTIS) is
described. The CTIS takes advantage of spatial and spectral multiplexing to avoid the need for scanning while maintaining high
spectral and spatial resolution over the visible spectrum. The ability
of the CTIS to simultaneously acquire data over a broad spectral range
at precise spatial positions offers several benefits over existing
systems. These benefits include spatial co-registration of spectral
images and the possibility of high temporal resolution when the CTIS is
coupled to a camera with a fast read-out rate. The current temporal
sampling rate of the CTIS falls within the required range for many
physiological experiments (0.01-2 s) and the potential temporal
resolution is limited only by the detector read-out rate and the
signal-to-noise ratio on the array.
The CTIS acquires a continuous fluorescence spectrum (e.g., 450-750
nm) at every pixel so that two-dimensional (2D) images composed of
signals from one or multiple fluorescence reporter probes can be
acquired simultaneously. This may be of particular importance because
analysis of contiguously sampled spectra provides a sensitive approach
to correct for probe-independent artifacts, dye-dye interactions, and
the cross-sensitivity of probes for multiple factors
(Martinez-Zaguilán et al., 1996a, b). In the case
of analyzing a single probe, shifts in a probe's spectrum often occur
when it interacts with cellular constituents or is trapped within
subcellular compartments (Martinez-Zaguilán et al.,
1996b). For example, the peak fluorescence emission wavelength may shift corresponding to the ion bound and unbound forms of the
probe. Therefore, a ratio of the intensities at two peak emission wavelengths provides precise information about the relative
concentration of the ions (Mason, 1999). The accuracy of
such ratiometric calculations is improved by removing the need for
post-registration of spectral images. In addition, background
fluorescence due to endogenous fluorophores can be significant, and
identification of spectral overlap between the background- and
probe-specific signals enhances the accuracy of the measurement in such
cases (Eng, 1989).
Furthermore, during the course of many physical responses (e.g.,
hormone secretion, muscle contraction), several cellular parameters may
change either simultaneously or in sequence (e.g., Richmond et
al., 1997). Because many tissues are heterogeneous with
respect to the time course of activation at the cellular level, it is
critical to analyze multiple parameters in individual cells with
reasonable spatial and temporal resolution. A test of the CTIS will be
in applications where cells are loaded with multiple probes for
simultaneous analysis of several physiological processes. In this case,
spectral data provide precise determination of probe function that is
required to rule out probe-probe interactions that can severely
influence probe sensitivity and selectivity (Martinez-Zaguilán et al., 1996a, b). Moreover, in
the case where probe spectra overlap to a significant degree, spectral
line analysis protocols can be utilized to precisely evaluate the
signal from the individual probes, which would be impossible if signal
responses were monitored over only a few selective wavelengths
(Schröck et al., 1996). Thus, with continued
development of the CTIS system, the ability to obtain full spectral
data will enhance the accuracy and reliability of physiological measurements.
| |
INSTRUMENT DESCRIPTION |
The CTIS microscope consists of two optical subsystems: an
interchangeable fore-optics subsystem and the CTIS subsystem. Within the context of this paper, the fore-optics subsystem consists of a
standard inverted fluorescence microscope (Olympus IMT-2) equipped with
a 100 W mercury lamp as the illumination source. Fig.
1 schematically illustrates the two
subsystems of a complete CTIS microscope. For the imaging experiments
described herein, an Olympus 60× 1.4 NA objective was used to collect
light from the sample, and one of two filter cubes was utilized to
reflect the excitation light to the sample and transmit emission light to the CTIS. During the SNARF-1 (free acid) calibration and the microsphere imaging experiments, a single excitation wavelength was
selected with a 10-nm bandpass filter centered at 490 nm. The dichroic
mirror transmitted >50% of incident emission light above 505 nm. For
the SNARF-1 (in situ) calibration and successive pH experiments, the
excitation light was centered at 520 nm with a 15-nm bandpass filter to
provide more optimal excitation of the SNARF-1. The dichroic mirror in
this case passed 50% of the incident emission above 550 nm. An
eyepiece (6.7×) at the side photo port of the microscope forms the
intermediate image at the field stop of the CTIS subsystem. Thus, a
single imaging eyepiece is the only optical element required for
adapting the CTIS subsystem to a standard microscope.
View larger version (26K):
[in this window]
[in a new window]
|
FIGURE 1
Schematic layout of the CTIS microscope combination. In
fluorescence imaging experiments, the bare-necessities fore-optics
subsystem consists of a light guide from an excitation source, a
dichroic beam splitter, and a microscope objective. Distances are not
to scale. See text for details.
|
|
The CTIS subsystem includes a field stop, collimator and re-imaging
lenses, a computer-generated-hologram (CGH) disperser, and a CCD
detector array. The CTIS is constructed with commercial optics, with
the exception of the CGH disperser. The disperser is designed to
produce a 7 × 7 array of diffraction orders. In designing the
CGH, the sum of diffraction efficiencies
(
tot) associated with the 49 diffraction orders is maximized such that a maximum fraction of the
light collected by the microscope objective is detected. Measurements
of tot have been as high as 78%
(Descour et al., 1997a). In order to maintain the same
signal and thus the same signal-to-noise ratio across the CCD array,
the disperser was designed so that the diffraction efficiency
() increases in higher diffraction orders to
compensate for the increasing dispersion.
In modeling the function of the CTIS, the (x, y, ) object
cube is interpreted as a collection of smaller volume elements, voxels
(
x, y, ). The CTIS maps the signal from each
voxel to a distinct diffraction pattern on a CCD array by means of the CGH disperser. A set of diffraction patterns, such as those shown in Fig. 2, is recorded for all
voxels in the object cube during a single integration time. A shift
in the center wavelength of a voxel results in the expansion or
contraction of the diffraction pattern within the focal plane. A change
in the voxel's spatial position results in a corresponding translation
of the diffraction pattern across the focal plane. The ensemble of the
diffraction patterns associated with each voxel describes the mapping
from the three-dimensional (3D) (x, y, ) object space to
the 2D image space (detector array) effected by the CTIS. This mapping
can be mathematically inverted to reconstruct an object cube from a raw
image.
View larger version (25K):
[in this window]
[in a new window]
|
FIGURE 2
Mapping of signal from voxel to imaging array. The
signal on the imaging array is distributed in a diffraction pattern
that depends on a voxel's (x, y, ) coordinates in the
object cube. Part (a) shows a voxel at a center wavelength
of 450 nm and the distribution of that voxel's signal on the imaging
array. Part (b) shows a voxel at a center wavelength of 710 nm and the distribution of that voxel's signal on the imaging array.
|
|
The zeroth diffraction order image is the result of direct imaging
through the CGH disperser without any dispersion (see center field
image in Fig. 3). This image therefore
represents an undispersed, broadband view of the specimen.
Consequently, the zeroth order image can be used to aim and focus the
microscope without the need for any data processing. Thus, sample
selection and initial positioning information are easily obtained. The
higher order images are associated with an increased dispersion, which
manifests itself as a radial blur in the higher diffraction orders.
View larger version (33K):
[in this window]
[in a new window]
|
FIGURE 3
A raw image of 15-µm-diameter microspheres with
excitation and emission peaks in nanometers at 505/535, 540/560, and
580/605. The image was taken at a magnification of ~320×, using a
40×, NA = 1.3 objective and an 8× imaging eyepiece.
|
|
| |
INSTRUMENT MODELING |
The CTIS instrument is modeled as a linear imaging system and
therefore can be described in terms of linear algebra. The 2D image
(Fig. 3) and the 3D object cube are reorganized as vectors g
and f, respectively. These vectors are related to each other
by means of a system matrix, denoted by H. The matrix H can be acquired experimentally by recording calibration
images, such as those shown in Fig. 2, for every voxel within the
object cube (Descour and Dereniak, 1995; Descour
et al., 1997b). As a result of the shift-invariance of the CTIS
subsystem, only a single calibration image per wavelength band needs to
be acquired and recorded (Volin et al., 1998). As
unrecorded calibration images are needed during reconstruction of the
object cube, the stored calibration images are recalled and shifted to
the desired spatial position.
The reconstruction of the 3D object cube is performed using the
multiplicative algebraic reconstruction technique (MART). The iterative
progression from the kth estimated object cube, 
k, to the (k + 1)st occurs according to the equation
|
(1)
|
where T indicates the matrix transpose and
HTg and
HTHk
form the backprojection of the collected raw image and the current image estimate, respectively. The multiplications and divisions are
taken to be point-by-point operations (Lent, 1976).
The CTIS belongs to the category of restricted view angles instruments.
As such, each voxel of the object cube is viewed through only a limited
set of angles corresponding to the limited number of projections on the
2D detector array. As is the case for all instruments within this
class, there exist two missing cones of data within the 3D Fourier
transform of the object cube (Barrett and Swindell,
1981). This translates into a limitation on the reconstruction
of objects whose frequency space representation falls within these
missing cones. Such objects include those with little spatial contrast
and sharp spectral transitions (Descour and Dereniak,
1995). Placing additional constraints on the reconstruction algorithm can reduce the reconstruction artifacts attributed to objects
of this type. These constraints will become essential as we shift from
imaging isolated cell populations to imaging confluent monolayers of
cells or whole tissue slices.
Reconstruction tests performed in conjunction with a non-imaging
reference spectrometer were used to determine the number of iterations
yielding the greatest accuracy in the reconstructed spectra. Typically,
seven or eight iterations proved optimal. Therefore, the results
presented in the next section were obtained after eight (8) iterations
of the reconstruction algorithm. Iterations required ~22 s to
complete on a 450 MHz Pentium II personal computer for 75 × 75 spatial resolution elements and 30 spectral bands. In our current work,
the initial estimate of the object cube, 0, corresponds to a spectrally
and spatially uniform field of view. Use of an initial guess that more
closely resembles the object would provide a more accurate
reconstruction using fewer iterations. This can be accomplished by
using an estimate, which is spatially identical to the zeroth order
image and spectrally uniform.
In the above model, we have assumed that the light incident on the
detector array arises from a single axial slice through the 3D sample
(x, y, z). For the types of samples presented here (monolayers of cells and sparse microspheres), the out-of-focus contributions can be reasonably neglected. Extending the applicability of the CTIS to the spectral analysis of multilayer samples will require
the development of multispectral deconvolution techniques. The design
of a second CTIS system dedicated to 3D (x, y, z) spectral imaging is currently underway.
| |
RESULTS |
Resolution of spectral signatures using fluorescent microspheres
A mixture of three 15-µm-diameter microspheres with different
fluorescent characteristics was imaged through the CTIS microscope. The
raw image shown in Fig. 3 was recorded using a digital scientific camera operating at a frame rate of 15 frames/s and with an integration time of 40 ms. Fig. 4 a shows
the reconstructed image with pixel locations corresponding to the
reconstructed spectra in Fig. 4 b. These spectral curves show the
clear spectral separation of the three reconstructed microspheres. The
full spectral and spatial information for this image was obtained
during a single integration time. The estimated spatial sampling
distance between adjacent pixels is 0.7 µm. A raster display of 18 reconstructed spectral images of this sample is shown in Fig.
5. Due to the inherent co-registration of
these spectral images, post-processing for correcting pixel alignment
is unnecessary. Due to the relatively large fluorescence signal
associated with the microspheres, a high degree of temporal resolution
also was realized. These results demonstrate the instrument's
potential to simultaneously capture the spectral signatures of multiple
fluorophores from different regions within the same sample with a
relatively high temporal resolution.
View larger version (19K):
[in this window]
[in a new window]
|
FIGURE 4
(a) Reconstructed image of multispectral
microspheres. Microspheres are 15 µm in diameter. The dimensions of
the reconstructed object cube were 83 × 83 × 31. The
spatial sampling between adjacent pixels is 0.7 µm for this image.
(b) Reconstructed spectra from three different locations
within the combination microsphere sample shown in a. Eight
iterations of the reconstruction algorithm were performed. Each
iteration required 17 s on a Pentium II 450 MHz personal
computer.
|
|
View larger version (66K):
[in this window]
[in a new window]
|
FIGURE 5
Raster display of a subset of spectral images of the
combination microsphere sample. Wavelength increases in raster fashion
from left to right, bottom to top, in steps of 10 nm. The numbers
provided in the above image are wavelengths in nanometers. Note the
clear spectral separation of the YG, orange, and red microspheres.
|
|
Response of SNARF-1 free acid to changes in media pH
SNARF-1 is a molecule whose fluorescence spectrum shifts in
response to changes in ambient pH. A calibration of SNARF-1 free acid
in an aqueous solution was performed to determine the sensitivity and
stability of the CTIS. Reconstructed spectra from images acquired at
different pH values and the responses at specific wavelengths are shown
in Fig. 6, a and b,
respectively. Calibration curves based on the ratio of fluorescence
emission at the ion-sensitive wavelengths (650 nm/590 nm) and the ratio
of the SNARF-1 basic peak relative to the isoemissive wavelength (650 nm/620 nm) are shown in Fig. 7. From Fig.
6 b it is clear that the fluorescence emission at the
isoemissive wavelength is stable, while the emissions at the
H+-sensitive wavelengths change as expected. However, based
on the ratio of ion-sensitive wavelengths, the pKa appears
to be shifted to the right (accepted pKa ~ 7.6;
Bassnett et al., 1990) with the dye ratio not reaching a
maximal value until after a pH of 9.0. This apparent shift in
pKa can be corrected if the signal from an ion-sensitive
wavelength is normalized to the fluorescence intensity at the
isoemissive point (Fig. 7). The basis for the spurious calibration
using the ratio of ion-sensitive wavelengths is the low signal
intensity at the acidic peak (590 nm), which can lead to errors in the
ratio at high pH values.
View larger version (17K):
[in this window]
[in a new window]
|
FIGURE 6
(a) Reconstructed spectra acquired during a
single trial of the free acid calibration of SNARF-1. Each curve
corresponds to the average spectrum for different pH solutions.
(b) Average response of the SNARF-1 at the pH-sensitive
wavelengths, 590 nm (squares) and 650 nm
(triangles), and the isoemissive point, 620 nm
(crosses) during a single calibration trial.
|
|
View larger version (17K):
[in this window]
[in a new window]
|
FIGURE 7
Free acid calibration of SNARF-1 generated from ratios
of the pH-sensitive wavelengths (650/590). Squares, data are
means ± standard error of five independent experiments. Ratio of
the pH-sensitive wavelength, 650 nm, to the isoemissive wavelength, 620 nm, for a single calibration trial (triangles).
|
|
In situ calibration of SNARF-1 loaded into rat insulinoma (RIN)
cells
The applicability of the CTIS microscope to standard physiological
experiments was demonstrated by recording the response of SNARF-1 to
changes in pH in an insulin-secreting rat insulinoma cell line RIN-38
(Clark et al., 1990). SNARF-1 exhibits a distinct shift
in its emission spectrum in response to changes in pH within the
physiological range of pH values observed within the RIN cells. Raw
images were collected using a 12-bit cooled CCD camera with integration
times of up to 2 s. A post-reconstruction image of several
SNARF-1-loaded RIN cells is shown in Fig.
8. By eliminating the necessity for
spectral or spatial scanning, full spectral information is obtained
from all cells during a single integration time. This allows for
comparison of probe response between individual cells and potentially
individual subcellular compartments.
View larger version (28K):
[in this window]
[in a new window]
|
FIGURE 8
(a) Post-reconstruction image of several RIN
cells loaded with SNARF-1. Eight iterations of the reconstruction
algorithm were performed to reconstruct the object cubes (75 × 75 × 30). Each iteration required 22 s on a Pentium II 450 MHz personal computer. (b) Reconstructed spectra acquired
during a single trial of the in situ calibration of SNARF-1. Each curve
corresponds to the average spectrum across an RIN cell in the presence
of different pH solutions.
|
|
Reconstructed spectra shown in Fig. 9
a were obtained by plotting the average fluorescence
emission over a 10 µm × 10 µm area of a single cell during
the in situ calibration. The spectral shift of the SNARF-1 fluorescence
is clearly observed with changing pH. Fig. 9 a shows the
response at the individual ion-sensitive wavelengths and the
isoemissive wavelength. Because the data at all wavelengths are
collected simultaneously, issues of co-registration are removed,
improving the measurement accuracy when compared to approaches that
require filter switching. Ratio data collected during an in situ
calibration and averaged for five cells is shown in Fig. 9. As seen
with the calibration of the free acid, the pKa derived from
the plot of ion-sensitive wavelengths appears shifted to higher pH
values (Fig. 9 b). In this case, there was a substantial
decrease in dye density between pH 7.5 and 8 shown by the intensity at
the isoemissive point. Again, when normalizing the intensity at the
basic peak to the response at the isoemissive wavelength, a reasonable
calibration curve is obtained. These observations demonstrate the
utility of acquiring full spectral data for accurate analysis of probe
function.
View larger version (14K):
[in this window]
[in a new window]
|
FIGURE 9
(a) Average response of the SNARF-1 loaded
cells at the pH sensitive wavelengths, 590 nm (squares) and
640 nm (triangles), and the isoemissive point, 610 nm
(crosses) during a single calibration trial. (b)
In situ calibration of SNARF-1 loaded in RIN cells generated from
ratios of the pH-sensitive wavelengths (640/590). Squares,
data are means ± standard error of five independent experiments.
Ratio of the pH-sensitive wavelength, 640 nm, to the isoemissive
wavelength, 610 nm, for a single calibration trial
(triangles).
|
|
The functional response of SNARF-1 was analyzed by treating
SNARF-1-loaded cells with the weak base NH4Cl. The
responses of four individual cells are shown in Fig.
10. The 640/590 calibration curve
shown in Fig. 9 was used to compute the pH values. Addition of
NH4Cl elicits a sudden alkalization of the cytosol followed by recovery to near resting pH. Upon removal of NH4Cl from
the media, NH4 rapidly exits the cell, causing sudden
acidification. Cytosolic pH recovers from this acid load due to the
activity of H+ extrusion mechanisms resident in the
cell-limiting membrane. Measured recovery rates are similar to those
observed in these cells using a standard spectral imaging microscope
(Martinez-Zaguilán et al., 1996a). Notice that
resting pH differs between individual cells, and this influences the
rate of recovery from the acid load.
View larger version (24K):
[in this window]
[in a new window]
|
FIGURE 10
The functional responses of four cells loaded with
SNARF-1 treated with the weak base NH4Cl. Plotting symbols
denote individual cellular responses within the sample.
|
|
| |
CONCLUSIONS |
We have presented proof-of-concept imaging results obtained with
the CTIS in combination with a commercial inverted fluorescence microscope. The CTIS microscope can simultaneously collect spatial and
spectral image data with relatively high spatial and spectral resolution. The current spatial and spectral sampling intervals of the
instrument are 1.0 µm and 10 nm, respectively. This level of
resolution is adequate to resolve signal responses from multiple fluorescence probes located within individual cells. Two aspects of the
current system need further development for standard use in biological
imaging. First, the field of view must be increased such that sampling
of signal over a wider tissue area can be accomplished. This
improvement will require an increase in the number of spatial resolution elements beyond the maximum of 83 × 83 reported in this paper. Additionally, increasing the signal-to-noise ratio of the
CTIS microscope is required because many functional probes either have
relatively low quantum yield or cannot be loaded to a high level
without affecting cell function. During preliminary spectral imaging
experiments, we determined that the transmission of the combined
microscope and spectrometer system (currently ~40%) could be
significantly increased with relatively modest modifications. An
increase in system transmission will result in an increased
signal-to-noise ratio while providing the capability for faster data
acquisition. We estimate the total system transmission can be increased
by 5-10% by simplifying the optical system, taking advantage of
high-efficiency anti-reflection coatings, and redesigning the CGH
disperser to maximize its efficiency. These improvements will be
addressed in a second generation of the CTIS microscope. The new
configuration will use a large-format detector array (2048 × 2048) and revised disperser design, resulting in an increased number of
spatial resolution elements to 200 × 200. Therefore, as a
consequence of reasonable hardware modifications, simultaneous 2D
analysis of multiple probes (cell functions) throughout a
heterogeneous sample will be attainable.
The simultaneous spectral and spatial imaging capabilities of the CTIS
have the potential to provide an important technological advancement in
a range of research areas. For example, the CTIS is an optimal tool for
the study of physiological changes in function from populations of
cells, which respond heterogeneously. Examples include individual cells
within a tissue, such as a liver slice, or a population of isolated
cells, such as neuronal cultures. To understand how individual cells
respond and how interactions between unique cell types occur, it is
critical to simultaneously monitor several functional parameters from
many cells. The use of fluorescent reporter compounds provides a tool
to evaluate changes in specific functional characteristics, such as ion
movements and metabolism. The reporters, however, may not be completely selective, or their signals may be difficult to calibrate. Analysis of
contiguously sampled spectra provides a sensitive approach to correct
for probe-independent artifacts, dye-dye interactions, and
cross-sensitivity between probes (Martinez-Zaguilán et
al., 1996b). The CTIS microscope also can be an important tool
to study the effects of drugs and toxic agents on the physiology of
tissue such as the liver. Because these types of effectors often elicit changes in many cell functions, the ability to monitor more than one
probe simultaneously from unique tissue regions will provide an
important technical advance, providing detailed information regarding
changes in ion movements (H+, Ca2+,
K+) and metabolism (O2, NADH) from cells
throughout a functioning hepatic unit, for example.
The CTIS also may be beneficial in clinical diagnosis and disease
staging where optical biopsy is providing a less invasive alternative
to more traditional approaches. Optical biopsy can be used in
conjunction with imaging of either tissue autofluorescence or emission
from a fluorescent drug to induce an optical contrast between tumors
and the surrounding normal tissue (Sabharwal et al.,
1999). Analysis of the fluorophore distribution for normal and
tumor tissue provides additional contrast within the spatial FOV,
allowing a determination of tumor infiltration. As optical biopsy is
often an endoscopic procedure, rapid data acquisition is essential in
reducing motion artifacts and for patient comfort. In this respect, the
potential high-speed imaging capabilities of the CTIS may prove
advantageous (Ford et al., 1999).
The authors thank E. L. Dereniak of the Optical Sciences
Center and G. H. Bearman of the Jet Propulsion Laboratory for
their support of and assistance with the work presented in this paper. We are also grateful to D. Wilson and P. D. Maker of the Jet
Propulsion Laboratory for providing the CGH disperser.
The work presented in this paper was funded in part by the National
Science Foundation (Grant DBI-9876717, B.K.F.), the American Diabetes
Association (R.M.L.), and the National Institutes of Health (Grant
RR14239, to M.R.D. and R.M.L.).
Address reprint requests to Bridget K. Ford, Optical Sciences Center,
University of Arizona, 1630 E. University Blvd., Tucson, AZ 85721. Tel.: 520-626-7212; Fax: 520-621-3389; E-mail:
bford{at}u.arizona.edu.
TOP
ABSTRACT
INTRODUCTION
-cell line.
Endocrinology.
127:2779-2788
acoustooptic and liquid crystal tunable filters.
Appl. Spectrosc.
48:857-866
