METHODS 18, 447 458 (1999)

Article ID meth.1999.0812, available online at http://www.idealibrary.com on

Practical Confocal Microscopy and the Evaluation

of System Performance 1
Robert M. Zucker 2 and Owen T. Price
Reproductive Toxicology Division, National Health and Environmental Effects Research Laboratory,
U.S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711

The laser scanning confocal microscope has enormous potential

in many fields of biology. Currently there is a subjective nature in the
assessment of a confocal microscopes performance by primarily
evaluating the system with a specific test slide provided by the users
laboratory. To achieve better performance from the equipment, it is
necessary to run a series of tests to ensure that the optical machine
is functioning properly. We have devised these methods on the Leica
TCS-SP and TCS-4D systems. Tests measuring field illumination, lens
clarity, laser power output, dichroic functioning, spectral alignment,
axial resolution, laser power stability, machine performance, and
system noise were derived to test the Leica laser scanning confocal
microscopy system. These tests should be applicable to other manufacturers systems as well. The relationship between photomultiplier tube (PMT) voltage, laser power, and averaging using a 10-mmdiameter test bead has shown that the noise (coefficient of variation
of bead intensity, CV) in an image increases as the PMT increases.
Therefore increasing the PMT setting results in increased noise. For
ideal image quality, it appears that it is better to decrease the PMT
setting and increase laser power, as noise generated by high PMT
settings will reduce the image quality far more than the bleaching
caused by higher laser power. Averaging can be used to improve the
image at high PMT values, provided the sample is not bleached by
repeated passes of the laser.

Laser scanning confocal microscopy (LSCM) has

emerged as a very useful technique to visualize bio1
The research described in this article has been reviewed and
approved for publication as an EPA document. Approval does not
necessarily signify that the contents reflects the views and policies of
the Agency, nor does mention of trade names or commercial products
constitute endorsement or recommendation for use.
2
To whom all correspondence and reprint requests should be addressed: U.S. Environmental Protection Agency, Reproductive Toxicology Division (MD-72), National Health and Environmental Effects
Research Laboratory, Research Triangle Park, NC 27711. Fax: (919)
541 4017. E-mail: zucker.robert@epamail.epa.gov.

1046-2023/99

logical structures. In LSCM, small apertures (pinholes) are used to eliminate the fluorescent light
from above and below the plane of focus, resulting in
increased resolution compared with that obtained
from conventional fluorescence microscopy. The addition of a powerful laser also allows the investigator
to image deeper inside a tissue than is possible with
a conventional mercury arc lamp. To achieve the
maximum performance from these sophisticated optical machines it is necessary that they are aligned
properly and tuned for efficiency (15). Although the
lasers are powerful and the optical systems are efficiently designed, photons should not be wasted.
From both scientific and quality assurance standpoints, optical performance delivered by the system
should be optimized (1). Without sufficient light being delivered by the optical system, it will be necessary to raise the PMT voltage to visualize the specimen. As the PMT is increased, the noise (CV) also
increases greatly. At higher PMT settings, it is necessary to use more averaging to reduce the noise
(CV) in the image. The need to raise the PMT settings should be minimized by operating the optical
system at maximum efficiency.
Although the human visual system is exquisitely
sensitive to contrast and resolution, it is less adept
at judging gradual luminance variations. Therefore,
it is beneficial to use beads or test slides to statistically quantify system performance. This review describes various methods that have been successfully
used in our laboratory to evaluate the functioning of
the Leica TCS-SP and TCS-4D confocal microscopes.
These tests should be valid for other point scanning
systems and should not only be confined to the
Leica equipment. Additional guidance on optimizing
LSCM performance has been published recently (1,
2).

447

448

ZUCKER AND PRICE

MATERIALS
Fluorescent Slides
The field illumination test slides were three fluorescent plastic slides with excitation peak wavelengths of
408, 488, and 590 nm and emission peak wavelengths
of 440, 519, and 650 nm, respectively. The slides have
a chemical embedded within them and become fluorescent at the desired excitation wavelength. They were
obtained from Applied Precision Inc (Issaquah, WA).
Power Meter
The power meter used to measure light on the microscope stage was a Lasermate/Q (Coherent, Auburn,
CA) with visible (LN36) and UV (L818) detectors. A
second power meter (No. 0163-662-00, Coherent, Santa
Clara, CA) was used to measure and control the light
emitted from UV Enterprise laser (Coherent, Santa
Clara, CA).
Beads
The following beads used in this study were obtained
from Spherotech (Libertyville, Illinois): rainbow fluorescent
particles
(10
mm,
RFSP-100),
ex
365,488,568,647; yellow beads (4.2 mm, fps-4052), ex
488; Nile red beads (2.9 mm, fps-3056), ex 568; UV
beads (3.1 mm, fps-3040), ex 365; blue beads (3.2 mm,
fps-306,810), ex 647. Ten-micrometer Fluorospheres
(Fullbright green II, Coulter, Hialeah, FL, ex 488) and
1-mm 1 Tetraspeck beads (T7284 Molecular Probes,
Eugene, OR, ex 365,488,568,647) were used for standardization and spectral registration, respectively.
Biological Test Slides
Two slides of FluoCells (F-14781, and F-14780, Molecular Probes) stained with three fluorochomes were
used as biological test slides. Slide 1 (F-14780) was
stained with the following dyes: Mitotracker red
CMXRos, BODIPY FL phallacidin, DAPI.
Slide 2 (F-14781) was stained with the following
dyes: BODIPY Fl goat anti-mouse IgG, Texas Red-X
phalloidin, DAPI.
Additional slides were made with cells growing on
coverslips, fixed with paraformaldehyde or glutaraldehyde, and stained with DAPI and other suitable fluorochromes.
Axial (z) Resolution Test
To test the axial resolution of the laser scanning
confocal microscope it is necessary to use a reflecting
mirror. A single reflecting mirror, 21 mm square (No.
31008, Edmonds Scientific, Philadelphia, PA) was
glued onto a microscope slide. A coverglass 1.5 (Fischer, Pittsburgh, PA) was placed on top of the slide
with a drop of immersion oil (Leica immersion oil, n 5

1.518) The coverslip is placed firmly onto the mirror to

remove all excessive oil. A standard test slide can also
be obtained from the confocal manufacturer or Spherotech. It is important to compare the laboratory-derived
slide with the standard bought slide so a known standard for axial resolution in the laboratory is obtained.
The Leica standard is specified at 350 nm for a TCS-SP
(6).
Confocal Microscope
The Leica TCS-SP and Leica TCS4D (Heidelberg,
Germany) confocal microscope systems with an Omnichrome argon krypton laser emitting 488-, 568-, and
647-nm lines and a Coherent Enterprise laser emitting
351-, and 365-nm lines were used for these calibration
tests. The results should be applicable to all point
scanning systems although some of the software may
not be available on the different LSCM systems. TIFF
images were acquired with TCS-SP software and imported into Image Pro Plus (Media Cybernetics Silver
Springs, MD) for subsequent measurement and analysis.

METHODS AND RESULTS

To evaluate the performance of the LSCM system it
was necessary to apply various types of tests to measure and evaluate the following possible errors in the
microspectrophotometer, homogeneity of illumination,
instability of light source, reflectivity and transmission
of laser light, chromatic aberration in illumination and
observation optics, and system noise (1 6). The tests to
evaluate these are outlined below.
Field Illumination
To ensure that there is homogeneous field illumination, the intensity of illumination across the observation field can be measured with a number of test specimens. These include a concentrated fluorescent dye
suspended in a hanging drop well slide, small (13 mm)
or large (10 mm) spatially concentrated fluorescent
beads (Spherotech), fluorescent specimens, uranyl
glass slides, or plastic fluorescent slides (Applied Precision, Spherotech). Other tests for field illumination
include a piece of tissue paper stained with fluorescent
dye or fluorescent dye solution fluorescein (F-7505 or
rhodamine B, R-6626 Sigma, St Louis, MO) mixed with
immersion oil (Leica Immersion oil, n 5 1.518) that can
be easily made in the laboratory (1). Uranyl glass was
previously used to check field illumination but it is
currently difficult to obtain and we have found that
plastic slides have higher efficiency than uranyl glass
at all wavelengths. It is useful to have a few different
samples to test the field illumination, as there can be
problems with specific test slides. A field of small or

LSCM OPTIMIZATION

large beads suspended in a slide can also be used to

test field illumination. We have found the plastic slides
to be the most consistent sample for this test. It is
important to measure the field illumination at a specific depth, as the intensity distribution will change as
one goes deeper into the slide. The surface of the slide
is the brighest point emitting the most intense fluores-

449

cence in the z axial direction and we routinely measure

the intensity 50 150 mm beneath the surface dependent on the objective.
Figure 1 shows the field illumination test patterns of
six different objectives on our microscope. Using
Image-Pro Plus, the original images were contoured so
that each drop in brightness corresponds to a 10%

FIG. 1. Field illumination of the following lenses on the Leica DMIRBE inverted microscope: (A) 6.23 (Zeiss 53 Fluotar, N.A. 5 0.25); (B)
103 (Plan Fluotar, N.A. 5 0.3); (C) 203 (Plan-Apo, N.A. 5 0.6); (D) 403 (Plan Fluotar, N.A. 5 0.51.0); (E) 633 (Plan Apo NA 1.2); (F) 1003
(Plan-Apo, N.A. 5 1.4). The Zeiss 53 lens yielded an effective magnification of 6.23 on the Leica system. The illumination pattern was
derived from a plastic fluorescent slide imaged 50 150 mm below the surface.

450

ZUCKER AND PRICE

reduction in intensity. The lower-magnification objectives show highly concentrated illumination in the center of the field (bulls-eye). This problem is partially the
result of using the low-magnification objectives on a
machine designed for high-magnification work. The 53
(Fig. 1A) and 103 (Fig. 1B) objectives clearly illustrate
the problems with nonuniform illumination and bullseye patterns. One solution to this problem is to increase the zoom factor. This enlarges the illumination
center and pushes the lower intensities off the field of
view. However, increasing zoom also increases the
magnification and bleaching rate of the sample and
may defeat the purpose of using a low-magnification
objective to observe a large field of illumination. The
nonuniform pattern clearly illustrates a field illumination problem.
Some of the objectives demonstrate illumination intensities that are 40% less than the maximum illumination intensity (53 and 103). Usually the accepted

values obtained with high-magnification objectives are

less than 10% from the maximum center values. The
1003 objective appears to fulfill this criterion. This
field illumination test allows for the evaluation of both
the lens and the alignment of the microscope system.
It is important to allow the laser to warm up for 30
min and stabilize prior to taking measurements. It is
also necessary to test every lens in the system, as many
of the problems appear to be in the compatibility of the
specific lens with the alignment of the confocal microscope, which is normally always done at highmagnification power.
Figure 2 illustrates the problem of a dirty lens. A
nonuniform pattern varying greatly between the left
and right sides of the image is shown in Fig. 2B. The
intensity of the field varied approximately 70% from
the left to the right side of the field. Cleaning the lens
to remove oil and other particles gave an acceptable
illumination pattern (Fig. 2A). A comparison of the

FIG. 2. Field illumination pattern of dirty (A) and clean (B) 203 (Plan Apo, N.A. 5 0.6) Leica lenses. The variation in intensity from the
left to right side of the field for both images is illustrated by the intensity profile (C). Note that the dirty lens yields a reduction in intensity
of 70% across the field.

LSCM OPTIMIZATION

clean and dirty x-direction intensity profiles of the images in Figs. 2A and 2B is shown in Fig. 2C. A similar
illustration of a dirty lens can be found in Ref. (1).
Power Meter Readings
In order not to waste photons (1) it is necessary to
evaluate the performance of the laser when it first
arrives in the laboratory and periodically thereafter.
Lasers usually need periodic care and both horizontal
and vertical controls have been installed to adjust the
mirrors. Since some of the manufacturers do not allow
users access to these adjustments, it is necessary to
make sure they are functioning at maximum performance. They do need alignment during their lifetime
and it is not always sufficient to turn the power up to
obtain additional power at the microscope stage. Running lasers at excessive power is not good for the life of
the laser and it wastes photons.
A power meter can be placed on the stage to evaluate
the power and stability of the different wavelength
lines. The Bio-Rad 1024 has an open configuration
allowing the laser light to be measured prior to arrival
at the scan head and at the microscope stage to determine laser attenuation in the LSCM system. We have
devised a special holder for a probe detector (Coherent)
to allow it to fit securely on the microscope stage for the
measurement of both UV and visible intensities.
The laser power should be measured with a lowmagnification lens. The test is done the in the following
manner: the lens is lowered to a minimum specified
height, the detector head of the power meter is placed
on the stage, the detector is centered using mercury arc
fluorescence light and laser light, and field is zoomed to
maximum (32 times). Initially the center of the detector can be located by using the mercury arc UV fluorescence light. The LSCM zoom factor is set to maximum to reduce the beam scan and to focus it into the
sweet spot of the detector. The scanner is set at slow
speed bidirectional. The detector position is slightly
adjusted to achieve maximum signal by the microscopes x/y joystick or positioning controls. Once this is
achieved the power can be evaluated with the laser
meter.
The values obtained over time will determine the
stability of the laser and the maximum power output of
the laser. Customarily, it is necessary to allow the laser
to run at the desired power for 30 min prior to testing
its stability. The stability test consists of repeated measurements of the power over time. In our experiments
the argon krypton laser varied less than 10% over 3 to
4 h, based on measurements made approximately every 15 min. The laser can be turned up briefly for a
short period to determine the maximum output at all
three lines. These two controls allow the user to determine the maximum laser output delivered and the
laser stability at the desired power used. It should be
emphasized that this test is performed at the micro-

451

scope stage prior to the emission barrier filters. A new

argon krypton mixed gas laser delivered the following
power outputs: 488 nm, 1.10 mW; 568 nm, 2.68 mW;
647 nm, 1.60 mW.
The maximum laser power should be measured at
each available wavelength (365, 488, 568, and 647 nm).
These measurements can be used as a standard to
compare with future laser output measurements. Bad
alignment or an aging laser could be detected by this
test. Also, because of the instability of the laser at high
power, similar measurements can be made when the
lasers power control lies at a user-defined position (i.e.,
half-power, 12 oclock, or 9 oclock position). The laser
should be warm prior to making this measurement
(usually 30 min is sufficient).
A power meter (No. 0163-662-00, Coherent) was
used to measure and control the light emitted from UV
Enterprise laser. This power meter was also useful in
comparing the amount of light emitted from the laser
with what was delivered onto the stage measured by
the Coherent Lasermate/Q with a UV detector (L818).
It was found that as the power on the laser head
increased above 40 mW, additional laser light was not
delivered onto the stage. In our system, a Coherent UV
60-mW, 3-year-old Enterprise laser delivered 40 mW of
laser power by a meter reading, but only 432-mW maximum power through the objective using a 53 Zeiss
lens (Fluar 0.25NA). The laser light coming out of the
laser was in excess of 30 mW. This indicated that there
was an attenuation of the laser light through the optics
of the system. However, to perform this test properly it
is essential to have an UV-transmissible dry objective
between 53 and 103 in our system. It may be possible
to use higher-magnification objectives with different
confocal systems.
The newer systems are designed with various optical
and merge components that attenuate some of the light
and redirect the light with optical fibers. These optical
configurations have a tendency to reduce the power
and it is important to measure this power output in the
system.
Laser Care
Argon, UV, and krypton lasers usually need to be
adjusted and aligned regularly. Small adjustments are
made to the mirrors with the horizontal and vertical
adjustment knobs on the back of these lasers. Since
most confocal companies do not allow the user to adjust
these controls, it is the responsibility of the manufacturer on a service contract to ensure that the laser is
functioning properly. Most lasers will go out of alignment over time. The user usually does not notice this
and will turn the power of the system to higher values
to compensate for this lack of laser power. This will
shorten the life of the laser. Some applications need
maximum laser light and minimum PMT voltage for
optimal resolution. Loss of laser power or bad align-

452

ZUCKER AND PRICE

ment will reduce the power in the system, necessitating an increase in PMT to compensate for lack of laser
light. When there is insufficient light coming though
the system, careful realignment of the laser beam is
required (a separate procedure done by qualified personnel) to increase the laser output. If this cannot be
done it may be time to replace the laser.
Reflectivity of Dichroic and Barrier Filters
Dichroic filters are made to reflect specific wavelengths of light and pass other wavelengths of light.
The efficiency of the dichroic filter in reflecting light
can be evaluated easily by placing the meter on the
stage and inserting various filters to bounce the light.
For example, with a 488-nm line, placing either a single, double, or triple dichroic in the light path should
reflect successively less light, if they were designed
properly. Photons should not be wasted in a LSCM
system (1). In our Leica TCS4D system, the triple
dichroic (TD) reflected more light than the double dichroic (DD) due to a possible ineffective DD coating.
Although this test can be done with a specific sample,
it is easier and more conclusive to do this with a laser
power meter positioned on the microscope stage.
Laser output can aid the user in determining the
relative efficiency of the dichroic filters. The maximum
power and half-maximum power should be measured
at each laser wavelength and each available dichroic
setting. For example, the 568-nm line output should be
quantified with both the DD 488/568 and TD 488/568/
647 dichroic filters. The relative differences in efficiency between dichroics will allow the user to choose
which filter setting to apply with a specific dye in a
given experiment.

value will vary with different test slides so it is very

important to compare the user-determined values with
the service technicians slide. Once an acceptable value
is obtained it should remain constant and the scan
head can be configured with either an upright or an
inverted microscope. However, if alterations are made
in the scan head it may be necessary to realign which
can affect the axial resolution. Although it is not the
only criterion for a good image, the axial resolution of
the system is very important and must be maintained
while the instrument is functional (6). If lasers are
changed or adjusted it will be necessary to realign the
system to ensure that the axial resolution is meeting
manufacturers standards. As an added bonus, it seems
that the resolution among various manufacturers systems and among units of the same manufacturer can
be compared with this axial resolution test.
The value of 370 nm shown in Fig. 3 was above the
acceptable Leica specifications of 350 nm. Although the
resolution of 370 nm approaches the manufacturers
acceptable value, it failed this test by not meeting or
exceeding this 350-nm value. The major significance of
this observation was that the unit did not meet the
manufacturers specification, and by a simple test that
can be performed by the user without opening the scan
head we were able to identify a problem that needed
the attention of service personnel. A nonsymmetrical
diffraction ring pattern was observed by examining the
light pattern derived from laser light traversing
through the excitation pinhole. This results in laser
light attenuation and laser light corruption. Further
examination by the Leica service personnel revealed
that the sheet of metal that contains the pinhole was
slightly warped. By changing the excitation pinhole

Axial (z) Resolution

The axial resolution test is considered the gold standard of resolution in confocal microscopy (1, 6). To
measure axial resolution it is necessary to have a front
surface, single reflected mirror (No. 31008, Edmonds
Scientific). The mirror is glued onto a glass slide and a
1.5 coverslip (17 mm) is placed on top of it with a drop
of the manufacturers immersion oil (Leica, N 5
1.5180). Initially, axial resolution is tested in reflection
mode with the 1003 objective (1.4-N.A. Plan-Apo lens),
zoom of 243, large pinhole diameter opening, and minimum laser power. After the reflected surface is found
by scanning in xz mode, the pinhole aperture is reduced to a minimum size (0 value in Leica TCS-SP).
The image is obtained with averaging and the axial
intensity profile across the reflected surface is determined as shown in Fig. 3. The half-maximum intensity
value of the profile is obtained to determine the full
width half-maximum distance (FWHM). This value
represents the axial resolution of the system. It should
be below 350 nm with the Leica TCS-SP system. The

FIG. 3. Axial resolution. Histograms of a front-surface, single reflecting mirror under a coverslip (0.170 mm). The test was done in xz
scanning mode, reflection mode, and a zoom magnification over 10
coupled with a 1003 lens. The resolution is given by the FWHM at
370 nm. Note that the peak intensity is 240 and the half-maximum
intensity is at 120.

LSCM OPTIMIZATION

and getting the axial resolution below 350 nm, system

performance was increased.
Biological Test Slides
Most LSCMs are assessed for proper functioning
with the users slides. Although this is very subjective,
it does work in a crude sense. We have used the following test slides: HE-stained specimen (Goblet cells H215
Carolina Biological, Burlington, NC), mixed pollen
grains (B690 Carolina Biological) and tissue culture
cells stained with 496-Diamidino-2-phenylindole (DAPI),
fluorescein isothiocyanate (FITC), and tetramethylrhodamine isothiocyanate (TRITC) (FL 1280, FL 1281, Molecular Probes), chicken red blood cells (fixed with glutaraldehyde and stained with various probes), or user-made
slides stained with DAPI and miscellaneous other fluorescent probes.
The most useful in our laboratory appears to be the
FluoCells slide (Molecular Probes F-14780). This slide
allows for proper evaluation of resolution, cross-talk
between detector channels, and excitation/emission
with UV, FITC, and TRITC. Resolution of the biological structures mitochondria (mitotracker), nucleis
(DAPI), and tubules (FITC) can be assessed with the
slide. The spore slide can be used to show fine structure
of an object at various magnifications with different
excitation wavelengths. By using subjective criteria,
most sales and service personnel use their favorite
slide as their gold standard to determine if the system
is functioning properly. Although the human visual
system is exquisitely sensitive to contrast and resolution, it is beneficial to use beads or test images that will
evaluate the system performance with numbers and
statistics whenever possible. This allows for standardization and reproducibility. For this reason, in place of
a subjective biological slide, we use beads and test
slides to determine the system perfomance of the laser
scanning confocal microscope.
Beads
Beads have been essential to the flow cytometry field
in its assessment of the system for proper alignment
and operation (7). They can be used to assess many
features of alignment, field illumination, registration,
laser beam line, and laser power. Similar types of applications are possible when using confocal microscopes. The choice of beads is dependent on the users
preference and the type of test performed. We use
rainbow beads (Spherotech) which are excited by each
of the laser lines (365, 488, 514, 543, 568, 633, and 647
nm) available on our system. The visualization of these
excitation wavelengths ensures that the laser is emitting the desired excitation wavelength and the emitted
light is traversing through the proper bandpass filter.
It is also useful to have uniform beads that emit a
specific wavelength of light (365, 488, 568, 647 nm,

453

Spherotech). In many systems, the 647-nm line, emitted by the argon krypton laser, begins to fail and the
visualization of the 647-nm excitation bead is a simple
test to determine the existence of this line. Beads may
also be moved to various regions of the field to test field
illumination. If the beads are spatially concentrated,
they can be used to test field illumination. The problem
with this test is that it is possible that the beads in a
given field are centered at different depths and thus
may produce different intensities. To ensure that the
beads are all centered in the same plane of focus it is
necessary to take a stack of images and perform a 3D
reconstruction (maximum projection).
Spectral Registration
To evaluate the registration of the 365-, 488-, 568-,
and 647-nm lines, we use 1-mm multiple-wavelength
fluorescent beads (Tetraspec, Molecular Probes; Rainbow, Spherotech). Figure 4 shows the visible spectral
registration of a lens (1003 Plan-Apo, N.A. 5 1.4)
using a 1-mm multicolored bead (Tetra Spec, excitation
and emission peaks). The acoustical optical transmission filter (AOTF) was used to minimize laser crossover
between detection channels. The image of the bead
showed that the emission of the 488-, 568-, and 647-nm
lines is not totally coherent (Fig. 4). The bead was
imaged (xy and xz scans) with a 243 zoom and a slow
scanning rate, and was averaged eight times. The xy
and xz outlines of the bead (Figs. 4A, 4C) were produced with Image-Pro Plus. A mask, eliminating intensities below half of the maximum intensity in each
color channel, was created for each channel. The resulting images were median filtered (7 3 7); edge detected (Roberts), and combined into a single three-color
image. These procedures result in the smoothed halfmaximum outlines in Figs. 4A and 4C. In addition to
this method of visualization, the conventional intensity
profiles through the center of the bead in the x direction (Fig. 4B) and the z direction (Fig. 4D) are also
shown. Only visible wavelengths were analyzed with
these tests.
One, however, can also compare UV with one or more
of the visible wavelengths (usually 568 nm) in a similar
manner. It should be noted that there could be significant registration misalignments between the UV and
visible wavelengths. The system should be checked in
this manner for each objective used in UV.
Daily Bead Tests
To test the system for stability, the power of a laser line
is set to a specific and constant level using the power
meter with either the 488- or 568-nm line. A large bead
(610 mm) from a population of uniform size and intensity is positioned in the center of the field. This large bead
eliminates the need for a zoom factor and reduces bleaching of the bead. Next, the machine variables (pinhole size,

454

ZUCKER AND PRICE

PMT voltage, averaging, etc.) are set to reproducible values that will be used for all future studies. The image of
the bead should be obtained at a depth corresponding to
the maximum lateral diameter of the bead and with
maximum intensity slightly below saturation (5255).
This image is imported into Image-Pro Plus and its intensity values, lying inside a uniform region within the
circumference of the bead, are measured. These intensity
values should be relatively stable over a period. The CV of
the population was defined as CV 5 s/m (s 5 standard
deviation of intensity, m 5 mean intensity). This measure
can be thought of as a noise-to-signal ratio, where m
represents the true signal in the image and s represents
the noise in the image. In this case, the CV that we are
measuring is the variation of intensity within the bead,
as opposed to the variation of intensity among beads. A
reduction in the mean intensity of the population or an
increase in CV implies that there is either a laser power
or system alignment problem.

Figure 5 shows the intensity values of a 10-mm bead

that was measured at constant laser power and over
various PMT voltages. This test appears to be useful to
measure the reproducibility of the system over weeks
and months. It may also be useful to assess PMT stability and linearity. The intensity values increase in a
nonlinear, approximately exponential fashion, as the
PMT voltage is increased in value. This property may
be due to the inherent PMT nonlinearity and possible
instability as the voltage (gain) of the PMT is raised to
high values (3, 79).
The increase in intensity values is accompanied by
increased noise at higher PMT values. This relationship has significant effects on image quality as shown
in Figs. 6 and 7.
Bead Noise
The noise of the system was evaluated with a 1003
objective (Plan-Apo, N.A. 5 1.4). The test sample was

FIG. 4. The visible spectral registration of a lens (1003 Plan Apo, N.A. 5 1.4) was evaluated with a 1-mm multiple-wavelength fluorescent
bead (Tetra Spec). Image processing procedures were applied to the original, three-channel images to visualize spatial differences in spectral
emission among the channels. These procedures result in the smoothed half-maximum outlines shown in (A) and (C). In addition to this
method of visualization, the conventional intensity profiles through the center of the bead in the x direction (B) and the z direction (D) are
also shown. Each pair of figures demonstrates the lack of complete registration among the three channels.

LSCM OPTIMIZATION

a large 10-mm bead (Coulter) of nearly uniform intensity (small CV). The large bead permitted repeated measurements without bleaching the sample.
By varying the PMT settings, the number of averaged frames (exposure time), and laser power, the
noise associated with various settings was evaluated. The system was set up with a bandpass filter of
505555 nm (BP-FITC), a single dichroic, and laser
excitation wavelength of 488 nm. A fixed region of
interest (ROI) was determined to cover about half of
the beads area and the Leica analysis software permitted the mean and standard deviation (SD) of the
intensities in the ROI to be determined. Image-Pro
Plus can also be used to measure the intensities over
a ROI. The PMT was set at approximately 50-unit
intervals starting at a setting of 449 and ending at
1000. Images corresponding to averaged frame
amounts of 1, 4, and 32 scans are produced at each
PMT setting. The laser power was varied using a
combination of the AOTF and laser power knob. The
laser power was set at a value that allowed the mean
intensity level of the bead to be approximately 150
(out of 255) for each PMT setting. For each PMT
setting between 449 and 1000, the mean intensity
and standard deviation of intensity were measured
on the same 10-mm bead. Thus, the noise associated
with each PMT and averaging amount combination
was evaluated for a uniform particle at a constant
intensity level (5150). The relationship between
PMT value, number of frames averaged, and calculated CV is shown in Fig. 6.

FIG. 5. The intensity of a 10-mm bead was determined at a constant laser power and various PMT settings. The intensity was
measured at low PMT values followed by higher PMT values to
minimize any effects caused by bleaching. Note that the intensity
increases in a nonlinear, almost exponential, fashion as the PMT is
increased in value. Note that some systems may not be set up to yield
a linear relationship between PMT setting (user controlled) and
actual electronic PMT voltage. This test can be used to standardize
the system and evaluate stability over a long period. The intensity
and PMT values are measured in arbitrary units (a.u.).

455

The averaging amount curves in Fig. 6 show that the

as the PMT value increases, the noise (CV) of the
system also increases. The noise at a specific PMT
setting can be reduced if the averaging is increased.
The major limitation is the increased amount of bleaching at lower CV values. It is useful to generate such a
curve to establish the relationship between PMT, laser
exposure time, and noise.
Noise in Biological Specimens
The information derived in Fig. 6 was applied to a
biological sample (FluoCells, F-14780 Molecular
Probes) to determine if a similar type of relationship
existed. The laser power was adjusted to yield images
at the same intensity level with PMT settings of 552
and 799. At the PMT setting of 552 the image was not
averaged (Fig. 7A). At the PMT setting of 799, images
were obtained with no averaging (Fig. 7B), 4 frames
averaged (Fig. 7C), and 32 frames averaged (Fig. 7D).
The CVs of intensities within the same ROI in the cell
nuclei (lower right) were (A) 49%, (B) 212%, (C) 109%,
and (D) 47%. The CVs from the equivalent PMT and
frame averaging settings on a 10-mm bead (Fig. 6) was
11, 45, 23, and 8%, respectively. The relative percentage changes in noise between the nuclei and beads are

FIG. 6. Noise evaluation. The noise of the system was evaluated

with a 10-mm bead (Coulter) and a 1003 Plan-Apo (1.4 N.A.) objective. By varying the PMT settings, the number of averaged frames
(exposure time), and laser power the noise associated with various
settings was evaluated. The excitation laser wavelength was 488 nm
and emission was a 50-nm bandpass filter (505555 nm). The PMT
setting was varied and the laser power was adjusted to counter the
PMT gain and yield images with a constant mean intensity. Images
corresponding to averaged frame amounts of 1, 2, 4, 8, 16, and 32
scans are produced at each PMT setting. The CV (s/m) of the intensity within the bead was measured at each combination of PMT
setting and frame averaging amount. The noise at a specific PMT
setting can be reduced if the averaging is increased. Note that each
constant averaging line implies that noise (CV) increases as PMT
voltage is raised.

456

ZUCKER AND PRICE

similar (approximately fivefold increase in CV) at the

same user settings. This suggests that the bead data
are scalable to other specimens and can be used as a
standard for setting up the machine and minimizing
noise.

It can be clearly shown that, at a given PMT setting, higher averaging values produce a greater reduction of CV (decreased noise). The use of lower
PMT settings allows for less averaging due to less
noise in the system. Thus increased scan exposure

FIG. 7. FluoCells (F-14780, Molecular Probes) were excited, with a 568-nm laser line and detected with a 50-nm bandpass filter (581 631
nm). The resolution was measured at 552 PMT using 1 and 4 averaging. The laser power was adjusted to yield images at the same intensity
level with PMT settings of 552 and 799. At the PMT setting of 552 the image was not averaged (A). At the PMT setting of 799, images were
obtained with no averaging (B), 4 frames averaged (C), and 32 frames averaged (D). The CVs of intensities within the same ROI in the cell
nuclei (lower right) were (A) 49%, (B) 212%, (C) 109%, and (D) 47%, respectively. A linear relationship between bead noise and noise in other
samples is implied by these data.

LSCM OPTIMIZATION

time decreases the noise in the system, lowers the

CV of the image, and produces a higher quality image. The challenge for the investigator is to create
this image without bleaching the sample by excessive laser power or exposure time. One way is to
reduce the size of the sampled image field (e.g., from
512 3 512 to 128 3 128). This will result in shorter
exposure times with less bleaching while keeping the
PMT and laser power settings constant.

DISCUSSION
The functioning of a LSCM system is dependent on
a number of variables that have to be controlled and
checked periodically to ensure system performance.
We have described the following tests: field illumination for individual lenses, laser power indicators,
dichroic efficiency tests, lens chromatic aberration,
axial resolution, system stability using beads, PMT
stability, and noise analysis. It is important to conduct these tests when the system arrives and routinely thereafter to assess the functionality of the
machine.
To obtain an image, it is necessary to balance laser
power, PMT setting, frame averaging amount, pinhole aperture size, and zoom magnification. Normally, if there is insufficient light to visualize the
sample, it is best to increase the laser power, not the
PMT setting. We have demonstrated that the PMT
should be minimized and averaging should be used
to reduce noise (CV) and increase image quality.
Although the pinhole diameter determines the confocality of the system and should ideally be set at the
size of the first Airy disk, the diameter can be increased if insufficient light is obtained to visualize
the sample. If available, the detection barrier filter
can also be replaced with a Schott long-pass filter to
increase detected light. It is almost always better to
increase the laser power and lower the PMT voltage
for ideal image quality.
We demonstrated the existence of a relationship between PMT, laser power, and frame averaging using a
10-mm test bead. The noise (CV) in an image increases
in a quasi-linear fashion as the PMT voltage increases.
Optimum resolution of an image appears to be determined by choosing a low PMT setting, high frame averaging, and high laser power. These factors yield a low
CV, increasing image quality. The data in Figs. 5, 6,
and 7 are interrelated. Figure 6 can be used as a rough
guideline for generating different image qualities at
different PMT settings. It can also be used to determine how much averaging is necessary at a specific
PMT setting for adequate image quality. Frame averaging is useful to reduce noise, but at the increased
amount of averaging necessary to reduce noise, bleach-

457

ing of the sample may occur. It is important to use high

laser power coupled with slow scan rates and low PMT
settings if frame averaging is not acceptable in an
experiment.
The noise introduced by operating the system with
high PMT values is very detrimental to image quality.
The way a PMT functions is known to electronic engineers but may be unknown to the end user biologist
who uses the system to answer scientific questions (7,
9). Often the LSCM user will set the PMT at high
values to see an image. This generates noise and poor
image quality that has to be eliminated with excessive
averaging. It is best to minimize the PMT setting to
ensure that the system is at optimum peak performance.
To obtain good image quality a number of variables
have to be controlled during image acquisition. One
must be aware of pinhole aperture size, PMT gain,
zoom magnification, scan speed, averaging of scans,
and laser power (13, 10). It is extremely difficult to
know how to balance these factors to produce the
ideal image. Although all of these factors are essential, we have found that the zoom greatly bleaches
samples and high PMT settings introduce excessive
noise which can be eliminated only though averaging, which in turn can bleach the sample. If possible,
magnification should be obtained with objectives, not
with zoom. The PMT values should be kept as low as
possible, even to the extent of increasing the pinhole
diameter and the laser power. These factors should
be regarded as starting points for using the LSCM
and not as absolute rules, as every sample has its
own unique qualities.
The lenses are essential for maximal image quality.
Some of the factors to consider are the following: high
numerical aperture relative to the specific magnification, flat field objectives, long working distance relative
to N.A., good transmission, and good achromatic correction at desired wavelengths. Usually these lenses
are very expensive, but they are necessary to obtain
the highest-quality images. They are the motors that
drive confocal imaging. This study has derived a number of tests (field illumination, spectral registration,
etc.) to ensure that the lenses are functioning properly
and yield the performance necessary for optimization
of the LSCM system.
To achieve maximum performance from a laser
scanning confocal microscope, it is necessary to run a
series of tests to ensure that the machine is functioning properly. Tests measuring field illumination,
lens clarity, laser power and stability, dichroic functionality, spectral registration, axial resolution,
overall machine stability, and system noise were
derived. These tests have been used by our laboratory but should not be considered the only useful
tests. Many other tests can be made to assess the

458

ZUCKER AND PRICE

performance of the machine (1, 2). We have developed some new tests in this study and perfected some
standard tests to assess the functionality of the laser
scanning confocal microscope. The major aim of this
study is to eliminate the subjective nature of assessing the confocal microscope by evaluating only a
user-defined sample. The ability to apply estimates
of noise (CV) and image resolution allows for a better
way of assessing the functionality of the confocal
system (11). It is anticipated that additional quality
assurance tests will be derived to assess this important scientific machine and ensure that it functions
at an optimum performance level.

ACKNOWLEDGMENTS
We thank the Leica Service EngineersJohn Ehli, Dennis Gil,
Paul Lilagan and Ulrich Fuchsfor their assistance. Without their
impute this investigation of LSCM system optimization would never
have been undertaken or accomplished.

REFERENCES
1. Centroze, V., and Pawley, J. (1995) in Handbook of Biological
Confocal Microscopy (Pawley, J., Ed.), 2nd ed., pp. 559 567,
Plenum, New York.
2. Centroze, V., and Pawley, J. (1998) in Cell Biology (Celis, J., Ed.),
2nd ed., Vol. 3, pp. 149 169, Academic Press, New York.
3. Sheppard, C. J. R., and Shotton, D. M. (1997) Confocal Laser
Scanning Microscopy, Bios Scientific, New York.
4. Pawley, J. (1995) in Handbook of Biological Confocal Microscopy,
(Pawley, J., Ed.), 2nd ed., pp. 19 36, Plenum, New York.
5. Roost, F. W. D. (1991) Quantitative Fluorescence Microscopy,
Cambridge Univ. Press, New York.
6. Cogwheel, C. J., and Lark, K. G. (1995) in Handbook of Biological
Confocal Microscopy (Pawley, J., Ed.), pp. 127137, Plenum,
New York.
7. Shapiro, H. (1995) Practical Flow Cytometry, 3rd ed., Wiley
Liss, New York.
8. Russ, J. C. (1995) Image Processing Handbook, 2nd ed., CRC
Press, Boca Raton, FL.
9. Art, J. (1995) in Handbook of Biological Confocal Microscopy
(Pawley, J., Ed.), 2nd ed., pp. 183195, Plenum, New York.
10. Teraskim, M., and Dailey, M. E. (1995) in Handbook of Biological
Confocal Microscopy (Pawley, J., Ed.), 2nd ed., pp. 327344,
Plenum, New York.
11. Watson, J. V. (1992) Flow Cytometry Data AnalysisBasic Concepts and Statistics, Cambridge Univ. Press, New York.