Optical Image Formation

Optical Image Formation
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 2
Microscopy from Carl Zeiss
Principles
Confocal
Laser Scanning Microscopy
Optical Image Formation
Electronic Signal Processing
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 1
Highlights of
Laser Scanning Microscopy
1982
The first Laser Scanning Microscope from Carl Zeiss.
The prototype of the LSM 44 series is now on display in the
Deutsches Museum in Munich.
1988
The LSM 10 – a confocal system
with two fluorescence channels.
1991
The LSM 310 combines confocal laser
scanning microscopy with
state-of-the-art computer technology.
1992
The LSM 410 is the first inverted microscope of the LSM family.
1997
The LSM 510 – the first system of the
LSM 5 family and a major breakthrough
in confocal imaging and analysis.
1998
The LSM 510 NLO is ready for multiphoton microscopy.
1999
The LSM 5 PASCAL –
the personal confocal microscope.
2000
The LSM is combined with
the ConfoCor 2 Fluorescence
Correlation Spectroscope.
2001
The LSM 510 META –
featuring multispectral analysis.
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 3
Confocal Laser Scanning Microscopy
In recent years, the confocal Laser Scanning Microscope (LSM) has become widely established as a
research instrument.
The present brochure aims at giving a scientifically
sound survey of the special nature of image formation in a confocal LSM.
LSM applications in biology and medicine predominantly employ fluorescence, but it is also possible to
use the transmission mode with conventional contrasting methods, such as differential interference
contrast (DIC), as well as to overlay the transmission
and confocal fluorescence images of the same specimen area.
Another important field of application is materials
science, where the LSM is used mostly in the reflection mode and with such methods as polarization.
Confocal microscopes are even used in routine quality inspection in industry. Here, confocal images
provide an efficient way to detect defects in semiconductor circuits.
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 4
Contents
Introduction
Part 1
Part 2
2
Optical Image Formation
Point Spread Function
6
Resolution and Confocality
8
Resolution
9
Geometric optic confocality
10
Wave-optical confocality
12
Overview
15
Signal Processing
Sampling and Digitization
16
Types of A/D conversion
17
Nyquist theorem
18
Pixel size
19
Noise
20
Resolution and shot noise –
resolution probability
21
Possibilities to improve SNR
23
Summary
25
Glossary
26
Details
Pupil Illumination
I
Optical Coordinates
II
Fluorescence
III
Sources of Noise
V
Literature
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 5
Following a description of the fundamental diffe-
Image generation
rences between a conventional and a confocal
The complete generation of the two-dimensional
microscope, this monograph will set out the
object information from the focal plane (object
special features of the confocal LSM and the capa-
plane) of a confocal LSM essentially comprises
bilities resulting from them.
three process steps:
The conditions in fluorescence applications will be
1. Line-by-line scanning of the specimen with a
given priority treatment throughout.
focused laser beam deflected in the X and Y
directions by means of two galvanometric
scanners.
2. Pixel-by-pixel detection of the fluorescence
emitted by the scanned specimen details, by
means of a photomultiplier tube (PMT).
3. Digitization of the object information contained
in the electrical signal provided by the PMT (for
Fig.1 The quality of the image generated in a confocal LSM is not only
influenced by the optics (as in a conventional microscope), but also, e.g.,
by the confocal aperture (pinhole) and
by the digitization of the object information (pixel size). Another important
factor is noise (laser noise, or the shot
noise of the fluorescent light).
To minimize noise, signal-processing
as well as optoelectronic and electronic
devices need to be optimized.
presentation, the image data are displayed,
pixel by pixel, from a digital matrix memory to a
monitor screen).
Digitization
Pixel size
Noise
Detector, laser, electronics,
photons (light; quantum noise)
Object
Resolution
Ideal optical theory
Pupil
Illumination
Resudial optical
aberations
Confocal aperture
2
Image
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 6
Introduction
Scanning process
Pinhole
In a conventional light microscope, object-to-
Depending on the diameter of the pinhole, light
image transformation takes place simultaneously
coming from object points outside the focal plane
and parallel for all object points. By contrast, the
is more or less obstructed and thus excluded from
specimen in a confocal LSM is irradiated in a point-
detection. As the corresponding object areas are
wise fashion, i.e. serially, and the physical inter-
invisible in the image, the confocal microscope can
action between the laser light and the specimen
be understood as an inherently depth-discriminat-
detail irradiated (e.g. fluorescence) is measured
ing optical system.
point by point. To obtain information about the
By varying the pinhole diameter, the degree of
entire specimen, it is necessary to guide the laser
confocality can be adapted to practical require-
beam across the specimen, or to move the speci-
ments. With the aperture fully open, the image is
men relative to the laser beam, a process known
nonconfocal. As an added advantage, the pinhole
as scanning. Accordingly, confocal systems are
suppresses stray light, which improves image con-
also known as point-probing scanners.
trast.
To obtain images of microscopic resolution from a
confocal LSM, a computer and dedicated software
are indispensable.
The descriptions below exclusively cover the point
scanner principle as implemented, for example, in
Carl Zeiss laser scanning microscopes. Configurations in which several object points are irradiated
simultaneously are not considered.
Fig. 2 Beam path in a confocal LSM. A microscope objective is used to
focus a laser beam onto the specimen, where it excites fluorescence, for
example. The fluorescent radiation is collected by the objective and efficiently directed onto the detector via a dichroic beamsplitter. The interesting wavelength range of the fluorescence spectrum is selected by an emission filter, which also acts as a barrier blocking the excitation laser line.
The pinhole is arranged in front of the detector, on a plane conjugate to
the focal plane of the objective. Light coming from planes above or below
the focal plane is out of focus when it hits the pinhole, so most of it cannot
pass the pinhole and therefore does not contribute to forming the image.
Confocal beam path
Detector (PMT)
Emission filter
The decisive design feature of a confocal LSM
Pinhole
compared with a conventional microscope is the
confocal aperture (usually called pinhole) arranged
Dichroic mirror
Beam expander
in a plane conjugate to the intermediate image
plane and, thus, to the object plane of the microscope. As a result, the detector (PMT) can only
Laser
detect light that has passed the pinhole. The pinhole diameter is variable; ideally, it is infinitely
Microscope objective
Z
small, and thus the detector looks at a point (point
detection).
X
As the laser beam is focused to a diffraction-limited
spot, which illuminates only a point of the object
at a time, the point illuminated and the point
Focal plane
Background
observed (i.e. image and object points) are situated
in conjugate planes, i.e. they are focused onto
each other. The result is what is called a confocal
Detection volume
beam path (see figure 2).
3
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 7
Optical slices
With a confocal LSM it is therefore possible to
A confocal LSM can therefore be used to advan-
exclusively image a thin optical slice out of a thick
tage especially where thick specimens (such as
specimen (typically, up to 100 µm), a method
biological cells in tissue) have to be examined by
known as optical sectioning. Under suitable condi-
fluorescence. The possibility of optical sectioning
tions, the thickness (Z dimension) of such a slice
eliminates the drawbacks attached to the obser-
may be less than 500 nm.
vation of such specimens by conventional fluores-
The fundamental advantage of the confocal
cence microscopy. With multicolor fluorescence,
LSM over a conventional microscope is obvious:
the various channels are satisfactorily separated
In conventional fluorescence microscopy, the
and can be recorded simultaneously.
image of a thick biological specimen will only be in
With regard to reflective specimens, the main
focus if its Z dimension is not greater than the
application is the investigation of the topography
wave-optical depth of focus specified for the
of 3D surface textures.
respective objective.
Figure 3 demonstrates the capability of a confocal
Unless this condition is satisfied, the in-focus
Laser Scanning Microscope.
image information from the object plane of interest is mixed with out-of focus image information
from planes outside the focal plane. This reduces
image contrast and increases the share of stray
light detected. If multiple fluorescences are
observed, there will in addition be a color mix of
the image information obtained from the channels
involved (figure 3, left).
4
Fig. 3 Non-confocal (left) and confocal (right) image of a triple-labeled
cell aggregate (mouse intestine section). In the non-confocal image,
specimen planes outside the focal plane degrade the information of
interest from the focal plane, and differently stained specimen details
appear in mixed color. In the confocal image (right), specimen details
blurred in non-confocal imaging become distinctly visible, and the image
throughout is greatly improved in contrast.
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 8
Introduction
3 rd dimension
Time series
In addition to the possibility to observe a single
A field of growing importance is the investigation
plane (or slice) of a thick specimen in good con-
of living specimens that show dynamic changes
trast, optical sectioning allows a great number of
even in the range of microseconds. Here, the
slices to be cut and recorded at different planes of
acquisition of time-resolved confocal image series
the specimen, with the specimen being moved
(known as time series) provides a possibility of
along the optical axis (Z) by controlled increments.
visualizing and quantifying the changes.
The result is a 3D data set, which provides infor-
The following section (Part 1, page 6 ff) deals with
mation about the spatial structure of the object.
the purely optical conditions in a confocal LSM
The quality and accuracy of this information
and the influence of the pinhole on image forma-
depend on the thickness of the slice and on the
tion. From this, ideal values for resolution and
spacing between successive slices (optimum scan-
optical slice thickness are derived.
ning rate in Z direction = 0.5x the slice thickness).
Part 2, page 16 ff limits the idealized view, looking
By computation, various aspects of the object can
at the digitizing process and the noise introduced
be generated from the 3D data set (3D reconstruc-
by the light as well as by the optoelectronic com-
tion, sections of any spatial orientation, stereo
ponents of the system.
pairs etc.). Figure 4 shows a 3D reconstruction
computed from a 3D data set.
The table on page 15 provides a summary of the
essential results of Part 1. A schematic overview of
the entire content and its practical relevance is
given on the poster inside this brochure.
Fig. 4 3D projection reconstructed from 108 optical slices of a
three-dimensional data set of epithelium cells of a lacrimal
gland. Actin filaments of myoepithelial cells marked with
BODIPY-FL phallacidin (green), cytoplasm and nuclei of acinar cells with ethidium homodimer-1 (red).
Fig. 5 Gallery of a time series experiment with Kaede-transfected cells. By repeated activation of the Kaede marker (greento-red color change) in a small cell region, the entire green fluorescence is converted step by step into the red fluorescence.
0.00 s
28.87 s
64.14 s
72.54 s
108.81 s
145.08 s
181.35 s
253.90 s
290.17 s
5
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 9
Point Spread Function
In order to understand the optical performance
x
characteristics of a confocal LSM in detail, it is necessary to have a closer look at the fundamental
optical phenomena resulting from the geometry of
z
the confocal beam path. As mentioned before,
what is most essential about a confocal LSM is
that both illumination and observation (detection)
are limited to a point.
Not even an optical system of diffraction-limited
design can image a truly point-like object as a
point. The image of an ideal point object will
always be somewhat blurred, or “spread” corresponding to the imaging properties of the optical
system. The image of a point can be described in
quantitative terms by the point spread function
(PSF), which maps the intensity distribution in the
image space.
x
Where the three-dimensional imaging properties
of a confocal LSM are concerned, it is necessary to
consider the 3D image or the 3D-PSF.
y
In the ideal, diffraction-limited case (no optical
aberrations, homogeneous illumination of the
pupil – see Details “Pupil Illumination”), the 3DPSF is of comet-like, rotationally symmetrical
shape.
For illustration, Figure 6 shows two-dimensional
sections (XZ and XY ) through an ideal 3D-PSF.
From the illustration it is evident that the central
maximum of the 3D-PSF, in which 86.5 % of the
total energy available in the pupil are concentrated, can be described as an ellipsoid of rotation.
For considerations of resolution and optical slice
thickness it is useful to define the half-maximum
area of the ellipsoid of rotation, i.e. the welldefined area in which the intensity of the 3D point
image in axial and lateral directions has dropped
to half of the central maximum.
6
Fig. 6 Section through the 3D-PSF in
Z direction – top, and in XY-direction
– bottom (computed; dimensionless
representation); the central, elliptical
maximum is distinctly visible.
The central maximum in the bottom
illustration is called Airy disk and is
contained in the 3D-PSF as the greatest
core diameter in lateral direction.
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 10
Optical Image Formation
Part 1
Any reference to the PSF in the following discus-
PSFdet is also influenced by all these factors and,
sion exclusively refers to the half-maximum area.
additionally, by the pinhole size. For reasons of
Quantitatively the half-maximum area is described
beam path efficiency (see Part 2), the pinhole is
in terms of the full width at half maximum
never truly a point of infinitely small size and thus
(FWHM), a lateral or axial distance corresponding
PSFdet is never smaller in dimension than PSFill. It is
to a 50% drop in intensity.
evident that the imaging properties of a confocal
The total PSF (PSFtot) of a confocal microscope
LSM are determined by the interaction between
behind the pinhole is composed of the PSFs of the
PSFill and PSFdet. As a consequence of the interac-
illuminating beam path (PSFill ; point illumination)
tion process, PSFtot ≤ PSFill.
and the detection beam path (PSFdet ; point detec-
With the pinhole diameter being variable, the
tion). Accordingly, the confocal LSM system as a
effects obtained with small and big pinhole diam-
whole generates two point images: one by pro-
eters must be expected to differ.
jecting a point light source into the object space,
In the following sections, various system states are
the other by projecting a point detail of the object
treated in quantitative terms.
into the image space. Mathematically, this rela-
From the explanations made so far, it can also be
tionship can be described as follows:
derived that the optical slice is not a sharply delimited body. It does not start abruptly at a certain Z
PSFtot(x,y,z) = PSFill(x,y,z) . PSFdet(x,y,z)
(1)
position, nor does it end abruptly at another.
Because of the intensity distribution along the
optical axis, there is a continuous transition from
PSFill corresponds to the light distribution of the
object information suppressed and such made
laser spot that scans the object. Its size is mainly a
visible.
function of the laser wavelength and the numeri-
Accordingly, the out-of-focus object information
cal aperture of the microscope objective. It is also
actually suppressed by the pinhole also depends
influenced by diffraction at the objective pupil (as
on the correct setting of the image processing
a function of pupil illumination) and the aberra-
parameters (PMT high voltage, contrast setting).
tions of all optical components integrated in the
Signal overdrive or excessive offset should be
system. [Note: In general, these aberrations are
avoided.
low, having been minimized during system
design].
Moreover, PSF ill may get deformed if the
laser focus enters thick and light-scattering
specimens, especially if the refractive indices
of immersion liquid and mounting medium are
not matched and/or if the laser focus is at
a great depth below the specimen surface
(see Hell, S., et al., [9]).
7
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 11
Resolution and Confocality
Wherever quantitative data on the resolving
The smaller the pinhole diameter, the more PSFdet
power and depth discrimination of a confocal LSM
approaches the order of magnitude of PSFill. In the
are specified, it is necessary to distinguish clearly
limit case (PH < 0.25 AU), both PSFs are approxi-
whether the objects they refer to are point-like or
mately equal in size, and wave-optical image
extended, and whether they are reflective or fluo-
formation laws clearly dominate (wave-optical
rescent. These differences involve distinctly varying
confocality).
imaging properties. Fine structures in real
Figure 7 illustrates these concepts. It is a schematic
biological specimens are mainly of a filiform or
representation of the half-intensity areas of PSFill
point-like fluorescent type, so that the explana-
and PSFdet at selected pinhole diameters.
tions below are limited to point-like fluorescent
objects. The statements made for this case are well
Depending on which kind of confocality domi-
applicable to practical assignments.
nates, the data and computation methods for
As already mentioned, the pinhole diameter plays
resolution and depth discrimination differ. A com-
a decisive role in resolution and depth discrimina-
parison with image formation in conventional
tion. With a pinhole diameter greater than 1 AU
microscopes is interesting as well. The following
(AU = Airy unit – see Details “Optical Coordi-
sections deal with this in detail.
nates”), the depth discriminating properties under
consideration are essentially based on the law of
geometric optics (geometric-optical confocality).
Fig. 7 Geometric-optical (a) and wave-optical confocality (c) [XZ view].
The pinhole diameter decreases from (a) to (c).
Accordingly, PSFdet shrinks until it approaches the order of magnitude of PSFill (c).
a)
PH~3.0 AU
b)
PH~1 AU
c)
PH~0,25 AU
FWHMill, axial
FWHMdet,axial
FWHMill, lateral
FWHMdet, lateral
PSFdet >> PSFill
Geometric-optical
confocality
8
PSFdet > PSFill
PSFdet >= PSFill
Wave-optical
confocality
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 12
Optical Image Formation
Part 1
Resolution
Resolution, in case of large pinhole diameters
Axial:
FWHMill,axial =
(PH >1 AU), is meant to express the separate visibility, both laterally and axially, of points during
the scanning process. Imagine an object consisting
of individual points: all points spaced closer than
the extension of PSFill are blurred (spread), i.e. they
are not resolved.
0.88 . exc
(n- n2-NA2)
(2)
n = refractive index of immersion liquid,
NA = numerical aperture of the microscope objective,
λexc = wavelength of the excitation light
If NA < 0.5, equation (2) can be approximated by:
≈
Quantitatively, resolution results from the axial and
1.77 . n . exc
NA2
(2a)
lateral extension of the scanning laser spot, or the
elliptical half-intensity area of PSF ill . On the
assumption of homogeneous pupil illumination,
the following equations apply:
Lateral:
FWHMill,lateral = 0.51
exc
NA
(3)
At first glance, equations (2a) and (3) are not different from those known for conventional imaging
(see Beyer, H., [3]). It is striking, however, that the
resolving power in the confocal microscope
depends only on the wavelength of the illuminating light, rather than exclusively on the emission
wavelength as in the conventional case.
Compared to the conventional fluorescence
microscope, confocal fluorescence with large pinhole diameters leads to a gain in resolution by the
factor (λem/λexc) via the Stokes shift.
9
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 13
Let the statements made on PSF so far be further
Optical axis
illustrated by the figure on the left. It shows a secrounding the focus on the illumination side
0,005
0,005
0,002
0,005
0,01
tion through the resulting diffraction pattern sur(PSFill). The lines include areas of equal brightness
0,005
0,01
ized intensity of 1. The real relationships result by
rotation of the section about the vertical (Z) axis.
0,003
Symmetry exists relative to the focal plane as well
0,015
as to the optical axis. Local intensity maxima and
minima are conspicuous. The dashed lines mark
the range covered by the aperture angle of the
microscope objective used.
For the considerations in this chapter, only the
0,003
area inside the red line, i.e. the area at half maximum, is of interest.
max.
Focal plane
0,002
0,001
min.
min.
max.
0,9
0,7 0,5
0,3
0,2
0,1
0,05
0,01
min.
0,03
0,02
0,02
0,03
min.
max.
0,01
max.
min.
min.
(isophote presentation). The center has a normal-
Fig. 8 Isophote diagram of the intensity distribution
around the illumination-side focus (PSFill).
The intensity at the focus is normalized as 1.
(Born & Wolf, Priniples of Optics, 6th edition 1988,
Pergamon Press)
10
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 14
Optical Image Formation
Part 1
Geometric optical confocality
Above a pinhole diameter of 1 AU, the influence
of diffraction effects is nearly constant and equa-
Optical slice thickness (depth discrimination) and
tion (4) is a good approximation to describe the
stray light suppression (contrast improvement) are
depth discrimination. The interaction between
basic properties of a confocal LSM, even if the
PSFill and PSFdet becomes manifest only with pin-
pinhole diameter is not an ideal point (i.e. not infi-
hole diameters smaller than 1 AU.
nitely small). In this case, both depth discrimina-
Let it be emphasized that in case of geometric
tion and stray light suppression are determined
optical confocality the diameters of the half-inten-
exclusively by PSFdet. This alone brings an improve-
sity area of PSFdet allow no statement about the
ment in the separate visibility of object details over
separate visibility of object details in axial and
the conventional microscope.
lateral direction.
In the region of the optical section (FWHMdet,axial),
Hence, the diameter of the corresponding half-
object details are resolved (imaged separately) only
intensity area and thus the optical slice thickness
unless they are spaced not closer than described
is given by:
by equations (2) / (2a) / (3).
FWHMdet,axial =
λem
PH
n
NA
=
=
=
=
0.88 . em
2
2
2
2
+
n- n -NA
2 . n . PH(4) (4)
NA
emission wavelength
object-side pinhole diameter [µm]
refractive index of immersion liquid
numerical aperture of the objective
Fig.9 Optical slice thickness as a function of the pinhole diameter
(red line). Parameters: NA = 0.6; n = 1; λ = 520 nm.
The X axis is dimensioned in Airy units, the Y axis (slice thickness)
in Rayleigh units (see also: Details “Optical Coordinates”).
In addition, the geometric-optical term in equation 4 is shown
separately (blue line).
Equation (4) shows that the optical slice thickness
comprises a geometric-optical and a wave-optical
term. The wave-optical term (first term under the
root) is of constant value for a given objective and
a given emission wavelength. The geometric-opti-
7.0
cal term (second term under the root) is dominant;
6.3
for a given objective it is influenced exclusively by
5.6
the pinhole diameter.
4.9
cality, there is a linear relationship between depth
discrimination and pinhole diameter. As the pinhole diameter is constricted, depth discrimination
FWHM [RU]
Likewise, in the case of geometric-optical confo-
4.2
3.5
2.8
improves (i.e. the optical slice thickness decreases).
2.1
A graphical representation of equation (4) is illus-
1.4
trated in figure 9. The graph shows the geometric-
0.7
optical term alone (blue line) and the curve resul-
0
ting from eq. 4 (red line). The difference between
the two curves is a consequence of the wave-
1.2
1.48 1.76 2.04 2.32 2.6
2.88 3.16 3.44 3.72
4.0
Pinhole diameter [AU]
optical term.
11
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 15
Wave-optical confocality
Thus, equations (2) and (3) for the widths of the
axial and lateral half-intensity areas are trans-
If the pinhole is closed down to a diameter of
formed into:
< 0.25 AU (virtually “infinitely small”), the character of the image changes. Additional diffraction
Axial:
effects at the pinhole have to be taken into
account, and PSFdet (optical slice thickness) shrinks
FWHMtot,axial =
to the order of magnitude of PSFill (Z resolution)
0.64 . (n- n2-NA2)
(7)
(see also figure 7c).
If NA < 0.5, equation (7) can be approximated by
In order to achieve simple formulae for the range
≈
of smallest pinhole diameters, it is practical to
regard the limit of PH = 0 at first, even though it is
of no practical use. In this case, PSFdet and PSFill
are identical.
1.28 . n . NA2
Lateral:
The total PSF can be written as
FWHMtot,lateral = 0.37
2
PSFtot(x,y,z) = (PSFill(x,y,z))
(7a)
NA
(8)
(5)
In fluorescence applications it is furthermore
necessary to consider both the excitation wavelength λexc and the emission wavelength λem. This
is done by specifying a mean wavelength1:
≈ 2
em . exc
2exc + 2em
(6) (6)
Note:
With the object being a mirror, the factor in
equation 7 is 0.45 (instead of 0.64), and 0.88
(instead of 1.28) in equation 7a. For a fluorescent plane of finite thickness, a factor of 0.7
can be used in equation 7. This underlines
that apart from the factors influencing the
optical slice thickness, the type of specimen
1
12
For rough estimates, the expression λ ≈ √λem·λexc suffices.
also affects the measurement result.
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 16
Optical Image Formation
Part 1
From equations (7) and (7a) it is evident that depth
It must also be noted that with PH <1 AU, a dis-
resolution varies linearly with the refractive index n
tinction between optical slice thickness and resolu-
of the immersion liquid and with the square of the
tion can no longer be made. The thickness of the
inverse value of the numerical aperture of the
optical slice at the same time specifies the resolu-
objective {NA = n · sin(α)}.
tion properties of the system. That is why in the
To achieve high depth discrimination, it is impor-
literature the term of depth resolution is frequently
tant, above all, to use objectives with the highest
used as a synonym for depth discrimination or
possible numerical aperture.
optical slice thickness. However, this is only correct
As an NA > 1 can only be obtained with an immer-
for pinhole diameters smaller than 1 AU.
sion liquid, confocal fluorescence microscopy is
usually performed with immersion objectives (see
also figure 11).
0.85
A comparison of the results stated before shows
0.80
that axial and lateral resolution in the limit of
0.75
PH = 0 can be improved by a factor of 1.4. Further-
0.70
more it should be noted that, because of the
performance of a confocal LSM cannot be
0.65
Factor
wave-optical relationships discussed, the optical
0.60
enhanced infinitely. Equations (7) and (8) supply
0.55
the minimum possible slice thickness and the best
0.50
possible resolution, respectively.
From the applications point of view, the case of
strictly wave-optical confocality (PH = 0) is irrelevant (see also Part 2).
0.45
0.40
0.35
0.30
0
0.1
0.2
0.3
By merely changing the factors in equations (7)
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Pinhole diameter [AU]
and (8) it is possible, though, to transfer the equations derived for PH = 0 to the pinhole diameter
axial
lateral
range up to 1 AU, to a good approximation. The
factors applicable to particular pinhole diameters
can be taken from figure 10.
Fig. 10 Theoretical factors for equations (7) and (8),
with pinhole diameters between 0 and 1 AU.
To conclude the observations about resolution and
depth discrimination (or depth resolution), the
table on page 15 provides an overview of the formulary relationships developed in Part 1. In addition, figure 11a shows the overall curve of optical
slice thickness for a microscope objective of
NA = 1.3 and n = 1.52 ( λ = 496 nm).
In figure 11b-d, equation (7) is plotted for
different objects and varied parameters (NA, λ, n).
13
337_Zeiss_Grundlagen_e
25.09.2003
Optical slice
(NA = 1.3; n = 1.52;
= 496 nm)
16:16 Uhr
Seite 17
4.5
Fig. 11
4.0
a) Variation of
pinhole diameter
FWHM [µm]
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Pinhole diameter [AU]
1000
Depth resolution
(PH = 0; n = 1.52;
= 496 nm)
b) Variation of
numerical aperture
920
FWHM [nm]
840
760
680
600
520
440
360
280
200
1
1.1
1.2
1.3
1.4
Numerical aperture
600
Depth resolution
(PH = 0; NA = 1.3; n = 1.52)
c) Variation of
wavelength ( )
560
520
FWHM [nm]
480
440
400
360
320
280
240
200
488
504
520
536
552
568
584
600
Wavelength [nm]
1600
1520
Depth resolution
(PH = 0; NA = 0.8;
= 496 nm)
d) Variation of
refractive index
FWHM [nm]
1440
1360
1280
1200
1120
1040
960
880
800
1.33
1.36
1.38
1.41
1.44
1.47
Refractive index of immersion liquid
1.49
1.52
fluorescent plane
fluorescent point
reflecting plane (mirror)
14
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 18
Optical Image Formation
Part 1
Overview
Conventional microscopy
1. Optical slice thickness not definable
With a conventional microscope, unlike in confocal microscopy, sharply defined images of
“thick” biological specimens can only be
obtained if their Z dimension is not greater than
the wave-optical depth of field specified for the
objective used. Depending on specimen thickness, object information from the focal plane is
mixed with blurred information from out-offocus object zones.
Optical sectioning is not possible; consequently,
no formula for optical slice thickness can be
given.
2. Axial resolution
(wave-optical depth of field)
n . em
2
NA
Confocal microscopy 1 AU < PH < ∞
Confocal microscopy PH < 0.25 AU
1. Optical slice thickness1)
1. Optical slice thickness
2
2
0.88 . em
2 . n . PH
NA
+
n- n2-NA2
3. For comparison: FWHM of PSF in the intermediate image (Z direction) – referred to the object
plane.
1.77 . n . em
The term results as the FWHM of the total PSF –
the pinhole acts according to wave optics.
λ
stands for a mean wavelength – see the text
body above for the exact definition.
The factor 0.64 applies only to a fluorescent
point object.
2. Axial resolution
2. Axial resolution
0.88 . exc
0.64 . 2
(n- n2-NA2)
2
FWHM of PSF ill (intensity distribution at the
focus of the microscope objective) in Z direction.
As optical slice thickness and resolution are
identical in this case, depth resolution is often
used as a synonym.
3. Approximation to 2. for NA < 0.5
3. Approximation to 2. for NA < 0.5
1.77 . n . exc
FWHM of the diffraction pattern in the intermediate image – referred to the object plane) in
X/Y direction.
1.28 . n . 2
NA2
NA
NA
0.51 . em
NA
FWHM of total PSF in Z direction
No influence by the pinhole.
2
4. Lateral resolution
(n- n2-NA2)
Corresponds to the FWHM of the intensity distribution behind the pinhole (PSFdet). The FWHM
results from the emission-side diffraction pattern
and the geometric-optical effect of the pinhole.
Here, PH is the variable object-side pinhole diameter in µm.
(n- n -NA )
Corresponds to the width of the emission-side
diffraction pattern at 80% of the maximum
intensity, referred to the object plane. In the
literature, the wave-optical depth of field in a
conventional microscope is sometimes termed
depth resolution. However, a clear distinction
should be made between the terms resolution
and depth resolution.
0.64 . 4. Lateral resolution
4. Lateral resolution
0.51 . em
NA
FWHM of PSFill (intensity distribution at the focus
of the microscope objective) in X/Y direction
plus contrast-enhancing effect of the pinhole
because of stray light suppression.
0,37 . NA
FWHM of total PSF in X/Y direction plus
contrast-enhancing effect of the pinhole because of stray light suppression.
All data in the table refer to quantities in the object space and apply to a fluorescent point object.
1) PH < ∞ is meant to express a pinhole diameter of < 4–5 AU.
15
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 19
Part 2
Sampling and
Digitization
After the optical phenomena have been discussed
in Part 1, Part 2 takes a closer look at how the digitizing process and system-inherent sources of
noise limit the performance of the system .
As stated in Part 1, a confocal LSM scans the specimen surface point by point. This means that an
image of the total specimen is not formed simultaneously, with all points imaged in parallel (as, for
example, in a CCD camera), but consecutively as a
series of point images. The resolution obtainable
depends on the number of points probed in a feature to be resolved.
Confocal microscopy, especially in the fluorescence mode, is affected by noise of light. In many
applications, the number of light quanta (photons)
contributing to image formation is extremely
small. This is due to the efficiency of the system as
a whole and the influencing factors involved, such
as quantum yield, bleaching and saturation of
fluorochromes, the transmittance of optical elements etc. (see Details “Fluorescence”). An additional influence factor is the energy loss connected
with the reduction of the pinhole diameter.
In the following passages, the influences of scanning and noise on resolution are illustrated by
practical examples and with the help of a twopoint object. This is meant to be an object consisting of two self-luminous points spaced at 0.5 AU
(see Details “Optical Coordinates”). The diffraction
patterns generated of the two points are superimposed in the image space, with the maximum of
one pattern coinciding with the first minimum of
the other. The separate visibility of the points (resolution) depends on the existence of a dip
between the two maxima (see figure 12).
16
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 20
Signal Processing
Part 2
As a rule, object information is detected by a pho-
Types of A/D conversion
tomultiplier tube (PMT). The PMT registers the spatial changes of object properties I(x) as a temporal
The quality of the image scanned depends on the
intensity fluctuation I(t). Spatial and temporal
type of A/D conversion which is employed. Two
coordinates are related to each other by the speed
types can be distinguished:
of the scanning process (x = t · vscan). The PMT con-
• Sampling : The time (t) for signal detection
verts optical information into electrical informa-
(measurement) is small compared to the time (T)
tion. The continuous electric signal is periodically
sampled by an analog-to-digital (A/D) converter
and thus transformed into a discrete, equidistant
per cycle (pixel time) (see figure 12).
• Integration: The signal detection time has the
same order of magnitude as the pixel time.
succession of measured data (pixels) (figure 12).
Integration is equivalent to an averaging of intensities over a certain percentage of the pixel time
known as pixel dwell time. To avoid signal distortion (and thus to prevent a loss of resolution), the
integration time must be shorter than the pixel
time. The highest resolution is attained with point
sampling (the sampling time is infinitesimally
short, so that a maximum density of sampling
points can be obtained). By signal integration, a
greater share of the light emitted by the specimen
contributes to the image signal. Where signals are
Fig. 12 Pointwise sampling of a continuous signal
T = spacing of two consecutive sampling points
t = time of signal detection (t<<T)
weak (e.g. fluorescence), this is a decisive advantage over point sampling with regard to the signal-
Intensity
to-noise ratio (SNR). Therefore, Carl Zeiss confocal
LSM systems operate in the integration mode, as a
200
rule. The absolute integration time can be modified by varying the scanning speed, which also
means a change of the pixel time.
150
100
50
0
Time
t
T
17
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 21
Nyquist theorem
If the number of sampling points per feature size is
smaller than that given by the Nyquist theorem
It is known from Part 1 that the information con-
(undersampling), part of the information will be
tent of the signal is limited by the resolving power
lost. This is evident in Figure 14c especially by the
of the microscope optics. A realistic estimate for
unresolved fine features.
the resolving power is the full width at half maxi-
A greater number of sampling points per feature
mum intensity (FWHMlat) of a point image (see
size (oversampling) means a greater number of
equation 3).
readings without a gain in information; simultane-
To avoid a loss of information during the scanning
ously, the time per pixel becomes shorter. Thus,
process, it is necessary to stick to the Nyquist
the volume of data to be processed is inflated, and
theorem. The optimal pixel spacing in scanning a
the noise of the measurement signal increases (see
periodic signal, as defined by the Nyquist theorem,
page 20)
is half the period of the feature spacing to be
Under unfavorable conditions, also artefacts may
resolved, or two pixels per resolvable structure
result out of the digitization process (aliasing). As
detail. Together with the resolving power defined
a rule, this is the case if the feature spacing in the
above, this results in a maximum pixel spacing of
specimen is equal, or nearly equal, to the pixel
dpix = 0.5 x FWHMlat.
spacing.
With a two-point object (see explanation on page
17), the pixel spacing needed to separate the
two Airy discs in the digitized image is 0.25 AU
(figure 13).
Fig. 14 Oversampling, correct
sampling and undersampling
of a continous signal.
Fig. 13 The graph illustrates the scanning of a two-point object
with the minimum number of sampling points needed to avoid
a loss of resolution (spacing of sampling points 0.25 AU).
a)
1
150
0.6
100
intensity
Relative intensity
0.8
0.4
50
0.2
AU
0.25
0.5
0.75
1
1.25
1.5
1.75
0
50
Oversampling
18
100
150
Pixels
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 22
Signal Processing
Part 2
Pixel size
For a Carl Zeiss confocal LSM, there is a simple formula, based on the edge length of the scanned
A quantity of decisive importance in this connec-
field in the intermediate image:
tion is the maximum scanning angle set via the
scanning zoom. By varying the scanning angle, it is
possible to directly influence the edge length of
dpix =
system constant
number of pixels . zoomfactor . magnificationobj
the scanned field in the intermediate image plane
Number of pixels = number of pixels per line
Zoom factor (Z) = scanning zoom set in the software
(Example: Zoom factor 2 reduces the edge length of the
scanned field by a factor of 2)
Magnificationobj = objective magnification
System constant = 8.94 mm with LSM 510, LSM 5 Pascal
(minimum zoom factor = 0.7); 12.77 mm with LSM 310,
LSM 410 (minimum zoom factor =1)
(or object plane), and thus the pixel size (at a given
number of pixels per line). The smaller the scanning angle, the smaller is the edge length of the
scanned field, and the smaller is the pixel (see the
example below).
In this way, the user of a Carl Zeiss confocal LSM
can control the sampling rate (pixel size). For setting the suitable scanning zoom for correct
The minimum scanning zoom needed to fullfill the
Nyquist sampling, the pixel size dPix in the object
Nyquist theorem can therefore be calculated as
plane is important.
follows:
Z≥
3.92 . NA . system constant
number of pixels .magnificationobj . exc
NA = numerical aperture of objective
λexc = excitation wavelength
b)
For example, with a 40 x objective (NA = 1.3),
c)
512 pixels per scan line and a wavelength of 488 nm,
the full resolving power (correct sampling) is
achieved with a scanning zoom of 4.56 as a mini-
150
150
mum ; the corresponding pixel size is 95.8 nm.
With lower factors of the scanning zoom the pixel
size itself will be the limiting factor for resolution
100
100
(pixel resolution). Higher factors will cause oversampling. Hence, the zoom factor influences not
50
only the total magnification but also the resolution
50
properties of the system.
With the more recent LSM systems of Carl Zeiss,
the number of sampling points can also be influ-
0
50
Correct sampling
100
150
Pixels
0
50
Undersampling
100
150
Pixels
enced by an increase in the number of pixels per
scan line.
(The number of pixels (X/Y) per image can be
freely selected between 4 x 2 and 2048 x 2048).
19
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 23
Noise
The main types of noise important in a confocal
With a pinhole diameter <1AU, resolution improves
LSM system are detector noise (dark noise, sec-
(better point separation thanks to a deeper dip),
ondary emission noise), laser noise, and shot noise
which is penalized by a drastic loss in energy.
of the light (see Details “Sources of Noise”). As a
Moreover, it should be considered that it depends
rule, these sources of noise are of a statistical
on the signal level wich noise source dominates.
nature. Periodic noise rarely occurs, and if it does,
With high-amplitude signals (number of detected
it tends to correlate with defective devices or
photons > 10,000), laser noise is the dominating
mechanical vibration in the setup; therefore it has
effect, whereas the quality of low signals (number
been left out of consideration here.
of detected photons < 1000) is limited by the shot
noise of the light.
As the graphs in figure 15 show, the number of
Therefore, laser noise tends to be the decisive
photons hitting the PMT depends not only on the
noise factor in observations in the reflection mode,
intensity of the fluorescence signal (see Details
while shot noise dominates in the fluorescence
“Fluorescence”), but also on the diameter of the
mode. With recent PMT models (e.g., from Hama-
pinhole. The graph shows the intensity distribu-
matsu), detector dark noise is extremely low, same
tion of a two-point object resulting behind the
as secondary emission noise, and both can be neg-
pinhole, in normalized (left) and non-normalized
lected in most practical applications (see Details
form (right). The pinhole diameter was varied
“Sources of Noise”).
between 2 AU and 0.05 AU. At a diameter of 1 AU
Therefore, the explanations below are focused on
the pinhole just equals the size of the Airy disk, so
the influence of shot noise on lateral resolution.
that there is only a slight loss in intensity. The gain
in resolution, is minimum in this case.
Relative intensity
Fig. 15 As shown in Part 1, small pinhole diameters lead to improved resolution
(smaller FWHM, deeper dip – see normalized graph on the left). The graph on the right
shows, however, that constricting the pinhole is connected with a drastic reduction in
signal level. The drop in intensity is significant from PH <1 AU.
1.0
1.0
d = 2.00 AU
0.8
d = 2.00 AU
d = 1.00 AU
0.6
0.8
d = 1.00 AU
0.6
d = 0.50 AU
d = 0.25 AU
d = 0.05 AU
0.2
0.5
20
d = 0.50 AU
0.4
0.4
1
1.5
2
[AU]
0.2
d = 0.25 AU
0.5
1
1.5
d = 0.05 AU
[AU]
2
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 24
Signal Processing
Part 2
Resolution and shot noise –
Figure 17 (page 22) shows the dependence of the
resolution probability
resolution probability on signal level and pinhole
diameter by the example of a two-point object
If the number of photons detected (N) is below
and for different numbers of photoelectrons per
1000, fluorescence emission should be treated as
point object. [As the image of a point object is
a stochastic rather than a continuous process; it is
covered by a raster of pixels, a normalization
necessary, via the shot noise, to take the quantum
based on pixels does not appear sensible.]
nature of light into account (the light flux is
Thus, a number of 100 photoelectrons/point
regarded as a photon flux, with a photon having
object means that the point object emits as many
the energy E = h⋅ν). Resolution becomes contin-
photons within the sampling time as to result in
gent on random events (the random incidence of
100 photoelectrons behind the light-sensitive
photons on the detector), and the gain in resolu-
detector target (PMT cathode). The number of
tion obtainable by pinhole constriction is deter-
photoelectrons obtained from a point object in
mined by the given noise level. Figure 16 will help
this case is about twice the number of photoelec-
to understand the quantum nature of light.
trons at the maximum pixel (pixel at the center of
As a possible consequence of the shot noise of the
the Airy disk). With photoelectrons as a unit, the
detected light, it may happen, for example, that
model is independent of the sensitivity and noise
noise patterns that change because of photon sta-
of the detector and of detection techniques
tistics, degrade normally resolvable object details
(absolute integration time / point sampling / signal
in such a way that they are not resolved every time
averaging). The only quantity looked at is the
in repeated measurements. On the other hand,
number of detected photons.
objects just outside optical resolvability may
appear resolved because of noise patterns modulated on them. Resolution of the “correct” object
structure is the more probable the less noise is
involved, i.e. the more photons contribute to the
formation of the image.
Therefore, it makes sense to talk of resolution
Fig. 16 The quantum nature of light can be made visible in two ways:
• by reducing the intensity down to the order of single photons and
• by shortening the observation time at constant intensity, illustrated
in the graph below: The individual photons of the light flux can be
resolved in their irregular (statistical) succession.
Power
probability rather than of resolution. Consider a
model which combines the purely optical under-
Time
standing of image formation in the confocal
microscope (PSF) with the influences of shot noise
of the detected light and the scanning and digiti-
Power
zation of the object. The essential criterion is the
discernability of object details.
Time
Photon
arrivals
Time
21
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 25
A resolution probability of 90% is considered ne-
The pinhole diameter selected in practice will
cessary for resolving the two point images.
therefore always be a trade-off between two qual-
Accordingly, the two-point object defined above
ity parameters: noise (SNR as a function of the
can only be resolved if each point produces at least
intensity of the detected light) and resolution (or
about 25 photoelectrons. With pinhole diameters
depth discrimination). The pinhole always needs a
smaller than 0.25 AU, the drastic increase in shot
certain minimum aperture to allow a minimum of
noise (decreasing intensity of the detected light)
radiation (depending on the intensity of fluores-
will in any case lead to a manifest drop in resolu-
cence) to pass to the detector.
tion probability, down to the level of indetermi-
Where fluorescence intensities are low, it may be
nateness (≤ 50% probability) at PH = 0.
sensible to accept less than optimum depth dis-
As another consequence of shot noise, the curve
crimination so as to obtain a higher signal level
maximum shifts toward greater pinhole diameters
(higher intensity of detected light = less noise, bet-
as the number of photoelectrons drops.
ter SNR). For most fluorescent applications a pin-
The general slight reduction of resolution proba-
hole diameter of about 1 AU has turned out to be
bility towards greater pinhole diameters is caused
the best compromise.
by the decreasing effectiveness of the pinhole
(with regard to suppression of out-of-focus object
regions, see Part 1).
Resolution
probability
1.0
100e50e-
0.9
30e0.8
20e-
0.7
10e6e4e3e-
0.6
0.5
2e0.4
0.3
0.2
0.1
0.25
22
0.5
0.75
1
1.25
1.5
Pinhole
size [AU]
Fig. 17 The graph shows the computed resolution
probability of two self-luminous points (fluorescence objects)
spaced at 1/2 AU, as a function of pinhole size and for
various photoelectron counts per point object (e-).
The image raster conforms to the Nyquist theorem
(critical raster spacing = 0.25 AU); the rasterized image
is subjected to interpolation. The photoelectron count per
point object is approximately twice that per pixel (referred
to the pixel at the center of the Airy disk). Each curve has
been fitted to a fixed number of discrete values, with each
value computed from 200 experiments.
The resolution probability is the quotient between successful
experiments (resolved) and the total number of experiments.
A resolution probability of 70% means that 7 out of 10
experiments lead to resolved structures. A probability >
90 % is imperative for lending certainty to the assumption
that the features are resolved. If we assume a point-like
fluorescence object containing 8 FITC fluorescence molecules (fluorochrome concentration of about 1 nMol) a laser
power of 100 µW in the pupil and an objective NA of 1.2
(n = 1.33), the result is about 45 photoelectrons / point
object on the detection side.
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 26
Signal Processing
Part 2
Possibilities to improve SNR
averaging method is the lower load on the specimen, as the exposure time per pixel remains con-
Pinhole diameters providing a resolution proba-
stant. Photon statistics are improved by the addi-
bility below 90% may still yield useful images if
tion of photons from several scanning runs (SNR =
one uses a longer pixel time or employs the signal
n·N; N = const., n = number of scans averaged).
averaging function. In the former case, additional
By comparison, a longer pixel time directly
photons are collected at each pixel; in the latter
improves the photon statistics by a greater num-
case, each line of the image, or the image as a
ber N of photons detected per pixel (SNR = N,
whole, is scanned repeatedly, with the intensities
N = variable), but there is a greater probability of
being accumulated or averaged. The influence of
photobleaching or saturation effects of the fluo-
shot noise on image quality decreases as the num-
rophores.
ber of photons detected increases. As fluorescence images in a confocal LSM tend to be shotnoise-limited, the increase in image quality by the
methods described is obvious.
Furthermore, detector noise, same as laser noise at
high signal levels, is reduced. The figures on the
right show the influence of pixel time (figure 18)
and the influence of the number of signal acquisitions (figure 19) on SNR in [dB]. The linearity
apparent in the semilogarithmic plot applies
to shot-noise-limited signals only. (As a rule,
signals are shot-noise-limited if the PMT high
voltage needed for signal amplification is greater
than 500 V).
Variation
Variationofofpixel
pixeltime
time
34
34
33
33
32
32
31
31
30
30
29
29
28
SNR 28
SNR
27
[dB]
[dB] 27
26
26
25
25
24
24
23
23
22
22
21
21
20
20
1
2
A doubling of pixel time, same as a doubling of
3
4
10
Fig. 18
Pixel
Pixeltime
time[s]
[s]
the number of signal acquisitions, improves SNR
by a factor of 2 (3 dB). The advantage of the
Variation
Variation of
of averages
averages
Figures 18 and 19 Improvement of the signalto-noise ratio. In figure 18 (top), pixel time is
varied, while the number of signal acquisitions
(scans averaged) is constant.
In figure 19 (bottom), pixel time is constant,
while the number of signal acquisitions is varied.
The ordinate indicates SNR in [dB], the abscissa
the free parameter (pixel time, scans averaged).
34
34
33
33
32
32
31
31
30
30
29
29
SNR 2828
SNR
[dB]
[dB] 2727
26
26
25
25
24
24
23
23
22
22
21
21
20
20
1
2
3
4
Number of
of averages
averages
Number
10
Fig. 19
23
337_Zeiss_Grundlagen_e
a)
25.09.2003
16:16 Uhr
Seite 27
The pictures on the left demonstrate the influence
of pixel time and averaging on SNR; object details
can be made out much better if the pixel time
increases or averaging is employed.
Another sizeable factor influencing the SNR of an
image is the efficiency of the detection beam path.
This can be directly influenced by the user through
the selection of appropriate filters and dichroic
beamsplitters. The SNR of a FITC fluorescence
image, for example, can be improved by a factor
of about 4 (6 dB) if the element separating the
excitation and emission beam paths is not a neutral 80/20 beamsplitter1 but a dichroic beamsplit-
b)
ter optimized for the particular fluorescence.
Fig. 20 Three confocal images of the same fluorescence
specimen (mouse kidney section, glomeruli labeled with
Alexa488 in green and actin labelled with Alexa 564
phalloidin in red).
All images were recorded with the same parameters,
except pixel time and average. The respective pixel times
were 0.8 µs in a), 6.4 µs (no averaging) in b),
and 6.4 µs plus 4 times line-wise averaging in c).
c)
1
24
An 80/20 beamsplitter reflects 20% of the laser light
onto the specimen and transmits 80% of the emitted
fluorescence to the detector.
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 28
Summary
This monograph comprehensively deals with the
quality parameters of resolution, depth discrimination, noise and digitization, as well as their mutual
interaction. The set of equations presented allows
in-depth theoretical investigations into the feasibility of carrying out intended experiments with a
confocal LSM.
The difficult problem of quantifying the interaction
between resolution and noise in a confocal LSM is
solved by way of the concept of resolution probability; i.e. the unrestricted validity of the findings
described in Part 1 is always dependent on a sufficient number of photons reaching the detector.
Therefore, most applications of confocal fluorescence microscopy tend to demand pinhole diameters
greater than 0.25 AU ; a diameter of 1 AU is a
typical setting.
25
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 29
Glossary
Aperture angle of a microscope objective
AU
Airy unit (diameter of Airy disc)
dpix
Pixel size in the object plane
FWHM
Full width at half maximum of an intensity
distribution (e.g. optical slice)
n
Refractive index of an immersion liquid
NA
Numerical aperture of a microscope objective
PH
Pinhole; diaphragm of variable size arranged in
the beam path to achieve optical sections
26
PMT
Photomultiplier tube (detector used in LSM)
PSF
Point spread function
RU
Rayleigh unit
SNR
Signal-to-noise ratio
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 30
Details
To give some further insight into Laser Scanning
Microscopy, the following pages treat several
aspects of particular importance for practical work
with a Laser Scanning Microscope.
Pupil Illumination
Optical Coordinates
Fluorescence
Sources of Noise
337_Grundlagen_Infoboxen_e
25.09.2003
16:17 Uhr
Seite 1
Details
Pupil Illumination
All descriptions in this monograph suggest a confocal LSM
corresponds to a truncation factor T = 1.3). The lateral coor-
with a ray geometry providing homogeneous illumination
dinate is normalized in Airy units (AU). From T = 3, the Airy
at all lens cross sections. The focus generated in the object
character is predominating to a degree that a further
has an Airy distribution, being a Fourier transform of the
increase in the truncation factor no longer produces a gain
intensity distribution in the objective’s pupil plane. However,
in resolution. (Because of the symmetry of the point image
the truncation of the illuminating beam cross-section need-
in case of diffraction-limited imaging, the graph only shows
ed for an Airy distribution causes a certain energy loss (a
the intensity curve in the +X direction). Figure 21 (right)
decrease in efficiency). [In Carl Zeiss microscope objectives,
shows the percentage efficiency as a function of pupil
the pupil diameter is implemented by a physical aperture
diameter in millimeter, with constant laser beam expansion.
close to the mounting surface].
The smaller the pupil diameter, the higher the T-factor, and
The Airy distribution is characterized by a smaller width at
the higher the energy loss (i.e. the smaller the efficiency).
half maximum and a higher resolving power. Figure 21 (left)
Example: If the objective utilizes 50 % of the illuminating
shows the intensity distribution at the focus as a function of
energy supplied, this means about 8 % resolution loss com-
the truncation factor T (the ratio of laser beam diameter
pared to the ideal Airy distribution. Reducing the resolution
(1/e ) and pupil diameter).
loss to 5 % is penalized by a loss of 70 % of the illuminating
The graph presents the relative intensity distributions at the
energy. In practice, the aim is to reach an optimal approxi-
focus (each normalized to 1) for different truncation fac-
mation to a homogeneous pupil illumination; this is one
tors. (The red curve results at a homogeneous pupil illumi-
reason for the fact that the efficiency of the excitation
nation with T > 5.2, while the blue one is obtained at a
beam path in a confocal LSM is less than 10 %.
2
Gaussian pupil illumination with T ≤ 0.5; the green curve
Fig. 21
Efficiency
0.9
0.9
0.81
Relative efficiency
Relative intensity
Intensity distribution at the focus
1
0.8
T < 0.3 (Gauss)
0.7
T = 1.3
0.6
T > 5,2 (Airy)
0.5
0.72
0.63
0.54
0.45
0.4
0.36
0.3
0.27
0.2
0.08
0.1
0.09
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Lateral distance [AU]
0.8
0.9
1
0
2
4
6
8
10
12
14
16
18
20
Pupil diameter [mm]
The trunction factor T is defined as the ratio of
dlaser
( -22 )
laser beamand pupil diameter of the objective lens used: T =
; the resulting efficiency is defined as = 1 - e T
dpupille
The full width at half maximum of the intensity
distribution at the focal plane is definied as
FWHM = 0.71 . . , with = 0.51 + 0.14 . In ( 1 )
1-n
NA
I
With T< 0.6, the Gaussian character, and with T>1 the Airy character predominates the resulting intensity distribution.
337_Grundlagen_Infoboxen_e
25.09.2003
16:17 Uhr
Seite 2
Details
Optical Coordinates
In order to enable a representation of lateral and axial
Thus, when converting a given pinhole diameter into AUs,
quantities independent of the objective used, let us intro-
we need to consider the system’s total magnification;
duce optical coordinates oriented to microscopic imaging.
which means that the Airy disk is projected onto the plane
of the pinhole (or vice versa).
Given the imaging conditions in a confocal microscope,
Analogously, a sensible way of normalization in the axial
it suggests itself to express all lateral sizes as multiples
direction is in terms of multiples of the wave-optical depth
of the Airy disk diameter. Accordingly, the Airy unit (AU)
of field. Proceeding from the Rayleigh criterion, the follow-
is defined as:
ing expression is known as Rayleigh unit (RU):
1AU =
1.22 . NA
1RU =
1.22 . NA2
NA= numerical aperture of the objective
λ = wavelength of the illuminating laser light
with NA = 1.3 and λ = 496 nm → 1 AU = 0.465 µm
n = refractive index of immersion liquid
with NA = 1.3, λ = 496 nm and n = 1.52 → 1 RU = 0.446 µm
The AU is primarily used for normalizing the pinhole
The RU is used primarily for a generally valid representation
diameter.
of the optical slice thickness in a confocal LSM.
II
337_Grundlagen_Infoboxen_e
25.09.2003
16:17 Uhr
Seite 3
Details
Fluorescence
Fluorescence is one of the most important contrasting
In principle, the number of photons emitted increases with
methods in biological confocal microscopy.
the intensity of excitation. However, the limiting parameter
Cellular structures can be specifically labeled with dyes
is the maximum emission rate of the fluorochrome mole-
(fluorescent dyes = fluorochromes or fluorophores) in vari-
cule, i.e. the number of photons emittable per unit of time.
ous ways. Let the mechanisms involved in confocal fluores-
The maximum emission rate is determined by the lifetime
cence microscopy be explained by taking fluorescein as an
(= radiation time) of the excited state. For fluorescein this is
example of a fluorochrome. Fluorescein has its absorption
about 4.4 nsec (subject to variation according to the ambi-
maximum at 490 nm. It is common to equip a confocal LSM
ent conditions). On average, the maximum emission rate of
with an argon laser with an output of 15 – 20 mW at the
fluorescein is 2.27·108 photons/sec. This corresponds to an
488 nm line. Let the system be adjusted to provide a laser
excitation photon flux of 1.26·1024 photons/cm2 sec.
power of 500 µW in the pupil of the microscope objective.
At rates greater than 1.26 ·1024 photons/cm2 sec, the fluo-
Let us assume that the microscope objective has the ideal
rescein molecule becomes saturated. An increase in the
transmittance of 100 %.
excitation photon flux will then no longer cause an increase
With a C-Apochromat 63 x/1.2W, the power density at
in the emission rate ; the number of photons absorbed
the focus, referred to the diameter of the Airy disk, then is
remains constant. In our example, this case occurs if the
2.58 ·105 W/cm2. This corresponds to an excitation photon
laser power in the pupil is increased from 500 µW to rough-
2
flux of 6.34 ·10 photons/cm sec. In conventional fluores-
ly 1 mW. Figure 22 (top) shows the relationship between
cence microscopy, with the same objective, comparable
the excitation photon flux and the laser power in the
lighting power (xenon lamp with 2 mW at 488 nm) and a
pupil of the stated objective for a wavelength of
visual field diameter of 20 mm, the excitation photon flux is
488 nm. Figure 22 (bottom) illustrates the excited-state
23
only 2.48 ·10 photons/cm sec, i.e. lower by about five
saturation of fluorescein molecules. The number of photons
powers of ten.
absorbed is approximately proportional to the number of
This is understandable by the fact that the laser beam in a
photons emitted (logarithmic scaling).
18
2
confocal LSM is focused into the specimen, whereas the
specimen in a conventional microscope is illuminated by
parallel light.
The table below lists the characteristics of some important
The point of main interest, however, is the fluorescence (F)
fluorochromes:
emitted.
The emission from a single molecule (F) depends on the
Absorpt.
max.(nm)
σ/10–16
Qe
σ*Q/10–16
molecular cross-section (σ), the fluorescence quantum
Rhodamine
554
3.25
0.78
0.91
yield (Qe) and the excitation photon flux (I) as follows:
Fluorescein
490
2.55
0.71
1.81
Texas Red
596
3.3
0.51
1.68
Cy 3.18
550
4.97
0.14
0.69
Cy 5.18
650
7.66
0.18
1.37
F = σ · Qe · I [photons/sec]
Source:
Handbook of Biological Confocal Microscopy, p. 268/Waggoner
In the example chosen,
F = 1.15 ·108 photons/sec or 115 photons/µsec
III
337_Grundlagen_Infoboxen_e
25.09.2003
16:17 Uhr
Seite 4
Incident photons
1.5 . 10
24
1.29 . 10
24
1.07 . 10
24
8.57 . 10
24
6.43 . 10
24
4.29 . 10
24
2.14 . 10
24
What has been said so far is valid only as long as the mol0
ecule is not affected by photobleaching. In an oxygen-rich
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Laser power [mW]
environment, fluorescein bleaches with a quantum efficiency of about 2.7·10–5. Therefore, a fluorescence molecule
can, on average, be excited n = 26,000 times (n = Q/Qb)
10
before it disintegrates.
n
, and referred to the maximum emission rate,
10
this corresponds to a lifetime of the fluorescein molecule of
10
Fmax
about 115 µs.
It becomes obvious that an increase in excitation power
can bring about only a very limited gain in the emission
rate. While the power provided by the laser is useful for
Absorbed photons
With t=
10
10
10
FRAP (fluorescence recovery after photobleaching) experi-
10
ments, it is definitely too high for normal fluorescence
10
applications. Therefore it is highly important that the exci-
21
20
19
18
17
16
15
14
17
10
tation power can be controlled to fine increments in the
18
10
19
10
20
10
21
10
22
10
23
10
24
10
25
10
2
Incident photons [1/s . cm ]
low-intensity range.
A rise in the emission rate through an increased fluorophore concentration is not sensible either, except within
Fig. 22 Excitation photon flux at different laser powers (top)
and excited-state saturation behavior (absorbed photons) of
fluorescein molecules (bottom).
certain limits. As soon as a certain molecule packing density
is exceeded, other effects (e.g. quenching) drastically
reduce the quantum yield despite higher dye concentration.
therefore, is the number of dye molecules contained in the
Another problem to be considered is the system’s detection
sampling volume at a particular dye concentration. In the
sensitivity. As the fluorescence radiated by the molecule
following considerations, diffusion processes of fluo-
goes to every spatial direction with the same probability,
rophore molecules are neglected. The computed numbers
about 80% of the photons will not be captured by the
of photoelectrons are based on the parameters listed
objective aperture (NA = 1.2).
above.
With the reflectance and transmittance properties of the
With λ = 488 nm and NA = 1.2 the sampling volume can
subsequent optical elements and the quantum efficiency of
be calculated to be V = 12.7 ·10 –18 l. Assuming a dye con-
the PMT taken into account, less than 10 % of the photons
centration of 0.01 µMol/l, the sampling volume contains
emitted are detected and converted into photoelectrons
about 80 dye molecules. This corresponds to a number of
(photoelectron = detected photon).
about 260 photoelectrons/pixel. With the concentration
In case of fluorescein (NA = 1.2, 100 µW excitation power,
reduced to 1 nMol/l, the number of dye molecules drops to
λ = 488 nm), a photon flux of F~23 photons/µsec results.
8 and the number of photoelectrons to 26/pixel.
In combination with a sampling time of 4 µsec/pixel this
Finally it can be said that the number of photons to be ex-
means 3 – 4 photoelectrons/molecule and pixel.
pected in many applications of confocal fluorescence
In practice, however, the object observed will be a labeled
microscopy is rather small (<1000). If measures are taken
cell. As a rule, the cell volume is distinctly greater than the
to increase the number of photons, dye-specific properties
volume of the sampling point. What is really interesting,
such as photobleaching have to be taken into account.
IV
337_Grundlagen_Infoboxen_e
25.09.2003
16:17 Uhr
Seite 5
Details
Sources of Noise
Sources of noise effective in the LSM exist everywhere in the
Dark noise
signal chain – from the laser unit right up to A/D conversion.
Dark noise is due to the generation of thermal dark electrons
Essentially, four sources of noise can be distinguished:
Nd, irrespective of whether the sensor is irradiated. Nd sta-
Laser noise q
PMT voltage of 1000 V; with lower voltages it progressively
Laser noise is caused by random fluctuations in the filling of
loses significance.
excited states in the laser medium. Laser noise is propor-
Dark noise can be reduced by cooling the sensor. However,
tional to the signal amplitude N and therefore significant
the reduction is significant only if N ≤ Nd, e.g. in object-free
where a great number of photons (N < 10000) are detected.
areas of a fluorescence specimen. In addition, the dark noise
Nd. Dark noise is specified for a
tistically fluctuates about must be the dominating noise source in order that cooling
Shot noise (Poisson noise)
effects a signal improvement; in most applications, this will
This is caused by the quantum nature of light. Photons with
not be the case.
the energy h·υ hit the sensor at randomly distributed time
Additional sources of noise to be considered are amplifier
intervals. The effective random distribution is known as
noise in sensor diodes and readout noise in CCD sensors. In
Poisson distribution. Hence,
the present context, these are left out of consideration.
SNR ≈ NPoisson = N
where N = number of photons detected per pixel time
(= photoelectrons = electrons released from the PMT cathode
by incident photons). With low photoelectron numbers
(N < 1000), the number N of photons incident on the sensor
can only be determined with a certainty of ± N.
The mean square deviation ∆N from the average (N + Nd) of
the photoelectrons and dark electrons registered,
N = se . (N+Nd ) (1+q2)
so that the total signal-to-noise ratio can be given as
SNR =
N can be computed as
N=
photons
QE() . pixel time
where QE (λ) = quantum yield of the sensor at wavelength λ;
1 photon = h·c/λ; c = light velocity; h = Planck’s constant
Secondary emission noise
Caused by the random variation of photoelectron multiplication at the dynodes of a PMT. The amplitude of secondary
N2
se (N+Nd ) (1+q2)
2
where
N = number of photoelectrons per pixel time
(sampling time)
se = multiplication noise factor of secondary emission
q = peak-to-peak noise factor of the laser
Nd = number of dark electrons in the pixel or sampling time
Example:
For N =1000, Nd =100, se = 1.2, and q = 0.05
emission noise is a factor between 1.1 and 1.25, depending
on the dynode system and the high voltage applied (gain).
Generally, the higher the PMT voltage, the lower the secondary emission noise; a higher voltage across the dynodes
improves the collecting efficiency and reduces the statistical
behavior of multiplication.
V
SNR =
1000 2
1.22 (1000+100) (1+0.052)
= 25.1
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 31
LITERATURE
1. Barton, D.L., Tangyunyong, P.,
Scanning Fluorescent Microthermal Imaging,
Proceedings of 23rd Int Symposium for Testing and
Failure Analysis (10/1997), Santa Clara, California
11. Oldenbourg, R. et al.,
Image sharpness and contrast transfer in coherent
confocal microscopy, Journal of Microscopy
Vol.172, pp. 31-39, (10/1993)
2. Barton, D.L., Tangyunyong, P.,
Infrared Light Emission from Semiconductor
Devices, ISTFA, pp. 9-17, (1996)
12. Pawley, J.,
Handbook of Biological Confocal Microscopy,
Plenum Press, 2nd Edition (1995)
3. Beyer, H.,
Handbuch der Mikroskopie, 2nd Edition,
VEB Verlag Technik Berlin, (1985)
13. Stelzer, E.H.K.,
The intermediate optical system of laser scanning
confocal microscopes; Handbook of Biological
Confocal Microscopy, pp. 139-154,
Plenum Press, 2nd Edition (1995)
4. Born & Wolf,
Priniples of Optics,
6th edition 1988, Pergamon Press
5. Brismar, H., Trepte, O., Ulfhake, B.:
Spectra and Fluorescence Lifetimes of Lissamine,
Rhodamine etc....: Influences of Some
Environmental Factors Recorded with a Confocal
Laser Scanning Microscope, The Journal of
Histochemistry and Cytochemistry, Vol. 43,
pp. 699-707, (7/1995)
6. Keller, H.E.,
Objective Lens for Confocal Microscopy,
Handbook of Biological Confocal Microscopy,
pp. 111-125, Plenum Press, 2nd Edition (1995)
7. Lackmann, F., et. al.,
An Automated Latch-up Measurement System
Using a Laser Scanning Microscope,
SPIE Vol 1028 Scanning Imaging, (1988)
8. Gröbler, B.,
Untersuchungen zur Bildübertragung in abtastenden Mikroskopen unter besonderer
Berücksichtigung der 3D-Abbildung, PhD thesis,
University of Potsdam, (1995)
9. Hell, S., et al.,
Aberrations in confocal fluorescence microscopy
induced by mismatches in refractive index,
Journal of Microscopy, Vol. 169, pp. 391-405
(3/1993)
10. Nitschke, R.,Wilhelm, S., et al.,
A modified confocal laser scanning microscope
allows fast ultraviolet ratio imaging of intracellular
Ca2+ activity using Fura 2, Euro. J. Physiologie,
Vol. 433: pp. 653-663, (1997)
14. Stelzer, E.H.K., et. al.,
Nondestructive sectioning of fixed and living
specimens using a confocal scanning laser
fluorescence microscope: Microtomoscopy; SPIE,
Vol. 809, pp. 130-136, (1987)
15. Tanke, H.J., van Oostveldt, P., et al.,
A parameter for the distribution of fluorophores in
cells derived from measurements of inner filter effect
and reabsorption phenomenon,
Cytometry Vol. 2, pp. 359-369 (6/1982)
16. Tsien, R.Y., Waggoner, A.,
Fluorophores for Confocal Microscopy,
Handbook of Biological Confocal Microscopy,
pp. 267-277, Plenum Press, 2nd Edition (1995)
17. Webb, R.H., Dorey, C.K.,
The Pixelated Image, Handbook of Biological Confocal
Microscopy, pp. 55-66, Plenum Press, 2nd Edition (1995)
18. Wilhelm, S.,
Über die 3-D Abbildungsqualität eines konfokalen
Laser Scan Mikroskops,
Dissertation, Fachhochschule Köln, (1994)
19. Wilson, T., Carlini, A.R.,
Three dimensional imaging in confocal imaging
systems with finite sized detectors; Journal of
Microscopy, Vol. 149, pp. 51-66, (1/1988)
20. Wilson, T., Carlini, A.R.,
Size of detector in confocal imaging systems;
Optical Letters Vol.12, pp. 227-229, (4/1987)
21. Wilson, T., Sheppard‚ C.J.R.,
Theory and Practice of Scanning Optical Microscopy,
Academic Press, 2nd Edition (1985)
337_Zeiss_Grundlagen_e
25.09.2003
16:16 Uhr
Seite 1
AUTHORS
Stefan Wilhelm, Bernhard Gröbler,
Martin Gluch, Hartmut Heinz †
(Carl Zeiss Jena GmbH)
Carl Zeiss
Advanced Imaging Microscopy
07740 Jena
GERMANY
Phone: ++49-36 41 64 34 00
Telefax: ++49-36 41 64 31 44
E-Mail: [email protected]
www.zeiss.de/lsm
Subject to change.
Printed on environment-friendly paper,
bleached without the use of chlorine.
We gratefully acknowledge the assistance of many other
staff members who contributed to this brochure.
45-0029 e/09.03
Microscopy from Carl Zeiss
Laser
Photomultiplier (PMT)
• Light source – projected into specimen
• Laser power: adjustable via attenuation device
(AOTF, AOM, MOTF) and tube current setting (Ar)
• Lifetime Ar: prolonged by using lower tube current;
but laser noise will be increased (8 A = minimum noise)
• Stand-by mode: prolongs laser lifetime; not suitable for
image acquisition
• Laser line: can be chosen via selection device (AOTF, MOTF)
dependent on fluorescent dye. Generally: the shorter
the wavelength, the higher the resolution
• Application goals: (1) Protect specimen (reduction of dye
bleaching and phototoxicity) by reduction of laser power.
(2) Maximize fluorescence signal (higher SNR) by longer
pixel dwell times or averaging
• Detector – pixelwise detection of photons emitted /
reflected by the respective specimen detail
• Parameters: "Detector Gain"= PMT high voltage,
"Amplifier Offset"= black level setting, "Amplifier
Gain"= electronic post-amplification
• Calibration: "Amplifier Offset" on image background
(object-free area), "Detector Gain" according to scanned
image (object); setting aid = "Range Indicator"
(➝ "Palette"). Goal: least number of overmodulated
(red, Gain) and undermodulated (blue, Offset) pixels
• Signal amplification: First exploit "Detector Gain" slider
before "Amplifier Gain" > 1
Detector
Emission filter
Confocal
pinhole
Confocal Pinhole
• Depth discrimination – confocal aperture to prevent
detection of out-of-focus light (optical sectioning)
• Diameter: determines thickness of optical slice; optimum
diameter: 1 Airy unit = best trade-off between depth
discrimination capability and efficiency
• x/y position: factory-adjusted for all beam path configurations; can be modified manually ( ➝"Maintain-Pinhole")
Scanning Mirrors
• Scanning unit – moves focused laser beam across
specimen line by line
• Scanning speed: defines frame rate (frames/sec) and
pixel time, i.e. time the specimen is illuminated
• Pixel time: influences SNR of image; the longer the pixel time,
the more photons per pixel, the less noise in the picture;
but bleaching of fluorochromes may increase
• Pixel resolution: maximum resolution can be achieved if
pixel size is set correctly (at least 4 x 4 pixels (x, y) per
smallest detail) ➝ directly adjustable via scan zoom
• x/y frame size: variable from 4 x 2 up to 2048 x 2048 pixels;
maximum frame rate with 512 x 512 pixels 5 frames/sec
(bidirectional scan
); unidirectional scan
: slower by
factor 2
Beam Splitter
• Fluorescence beam path – definable by combination
of main (HFT) and secondary (NFT) dichroic mirrors and
emission filters (BP = bandpass, LP = longpass,
KP = shortpass) (➝ "Acquire"– "Config")
• HFT: separates excitation and emission light
• NFT: effects spectral division of (different) fluorescence
emissions (e.g. NFT 545: reflects light of λ < 545nm and
transmits light of λ > 545nm)
• BP, LP, KP: determines bandwidth of fluorescence
emission for the respective channel (e.g. LP 505:
λ ≥ 505 nm ➝ detection)
Laser source
Collimator
Main dichroic
beamsplitter
Scanning
mirrors
Objective
Z-Motor
• Focusing the specimen – acquisition of image stacks
or x-z sections
• z-interval: distance between two optical slices (step size
of z-motor: min. 25 nm)
• Optimum z-motor step size: 0.5 x optical slice thickness
(compare: min. slice thickness about 340 nm for NA = 1.4,
n = 1.52, λ = 488 nm)
• Optional: fast z-scanning stage (HRZ)
fast piezo objectiv focus
1
2
3
Objective Lens
Specimen
• Optical image formation – determines properties
of image quality such as resolution (x, y, z)
• Numerical Aperture (N.A.): determines imaged spot size
(jointly with wavelength), and substantially influences
the minimum optical slice thickness achievable
• Refractive index (n): match n immersion liquid with
n specimen mounting medium for better image quality.
• Best confocal multifluorescence images (VIS, UV):
use water immersion objectives with apochromatic
correction (C- Apochromat)
Focal plane
Z-motor
3 Steps to Get a Confocal Image
How to Enhance Image Quality
View specimen in VIS mode
More signal!
Focus the specimen in epi-fluorescence mode using the binocular
and center the part of interest; select fluorescence filter cube according
to application (e.g. FITC or Cy3) via SW (window "Microscope Control");
match the field of view: change to appropriate objective magnification
(consider use of correct immersion medium).
• Change to longer pixel dwell times by reducing scanning speed
• Use "Average" method: Calculation of "Sum"or "Mean" value
of pixels of consecutive "Line" or "Frame" scans.
• Increase bandwidth of emission filter (e.g. LP instead of BP).
• Enlarge pinhole diameter; Note: optical slice thickness increases accordingly.
• Increase excitation energy (laser power); but pay attention
to bleaching, saturation and phototoxic effects.
Load an LSM configuration
Activate LSM mode (operate manual tube slider or
button "LSM"). Open window "Configuration control", and
select a predefined configuration from list (Single Track).
A click on "Apply" automatically sets up the system: laser lines, attenuation,
emission filters, beam splitters (HFT, NFT), pinhole diameter, detector settings
(channels, gain, offset). Or: Click on "Reuse" button (stored image/image
database window) to restore settings of a previous experiment.
More details !
• Use objective with higher numerical aperture (NA); x/y-resolution ~ 1/NA,
z-resolution ~ 1/NA2.
• Increase "FrameSize"= number of pixels per line + lines per frame,
e.g. 1024 x 1024 or 2048 x 2048 (min. 4 x 2).
• Optimize scan zoom (Z), i.e. pixel size ≤ 0.25 x diameter of Airy disk
(e.g.: Objective 40x, NA 1.3, l = 488 nm => Z = 4.56).
• Increase dynamic range (change from 8 to 12 bit per pixel).
Scan an image
Click on "Find" button (right row in window "Scan Control")
=> System automatically opens image window, optimizes detector
settings (matches PMT gain and offset to dynamic range of 8 or 12 bit),
and scans an image.
See operating manual for scanning a stack of slices, time series etc.
More reliability !
• Use Multitracking: very fast switching of excitation wavelengths; prevents
crosstalk of signals between channels; predefined configurations available.
• Use ROI (Region Of Interest) function: significantly reduces excited area
of specimen and increases acquisition rate at constant SNR; several ROIs
of any shape can be defined and used simultaneously.
Carl Zeiss · 07740 Jena · Germany · E-mail: [email protected] · www.zeiss.de/lsm
45-0024 e/08.03
The Confocal Laser Scanning Microscope
We make it visible.
Microscopy from Carl Zeiss
Methods
Confocal
Laser Scanning Microscopy
Applications in Research and Teaching.
Design, Functions, Methods.
We make it visible.
Contents
Confocal
Laser Scanning Microscopy
Having decoded the human genome, biomedical
High Resolution in Space and Time
3
The Confocal Principle
4
Two-Dimensional Images
6
Three-Dimensional Images
8
research today is focused on exploring the interaction between cellular components. Scientists want to
find out which protein is where, and at what time,
and what other structural and functional modules it
interacts with.
In the search for answers to these questions, imaging
systems based on the classical light microscope have
Time Series
10
come to play an unprecedented role, thanks to many
technical innovations and a high degree of automa-
Multifluorescence – The Crosstalk
tion. Many experiments have only become possible
Problem and Its Solution
12
Spectral Imaging
14
Courses
22
Summary
23
Literature / Links
24
because of the new functions provided by modern
microscopes.
Triple staining of a primary culture
of a rat’s cortical neurons.
Nucleus: blue (DAPI),
Nestin: green (Cy2),
Doublecortin: red (Cy3).
Specimen: Dr. H. Braun, FAN GmbH,
Magdeburg, Germany.
2
High Resolution in Space and Time
Confocal laser scanning microscopes (LSMs) are
cence lifetimes. With such information it is possi-
distinguished by their high spatial and temporal
ble to increase the number of fluorescent labels
resolving power. They clearly outperform classical
used in an experiment, or to use fluorochrome
light microscopes especially by their axial resolu-
combinations unthinkable with conventional
tion – a quality that enables users to acquire opti-
detection methods. The advantages are obvious:
cal sections (slices) of a specimen. An object can
the more components in a cellular process that are
thus be imaged completely in three dimensions
observed simultaneously, the greater the yield of
and subsequently visualized as a 3D computer
information.
image. In another group of applications, exactly
Because of its versatility, laser scanning micros-
defined areas of a specimen can be selectively illu-
copy has become an established mainstream
minated by laser light. This functionality is essen-
method in biomedical research – a tool permitting
tial for quantitative investigations of dynamic proc-
scientists to follow innovative experimental paths.
esses in living cells using techniques such as FRAP
This article will show which basic functions and
(fluorescence recovery after photobleaching),
applications of laser scanning microscopy can be
FRET (fluorescence resonance energy transfer),
taught in academic tuition. The modern method of
photoactivation and photoconversion.
confocal laser scanning microscopy can be taught
New acquisition methods on the LSM permit the
on the basis of classical light microscopy, an estab-
detection of additional properties of the emitted
lished part of fundamental biomedical teaching.
light including spectral signatures and fluores-
Triple-labeled tissue section
of mouse intestine
In the non-confocal image, the interesting information of the
focal plane mixes with unwanted information from extrafocal
specimen planes; differently stained details result in a color mix.
In the confocal image, object details
blurred in the non-confocal image
are visible clearly and in greater contrast.
3
Detector
The
Confocal Principle
In this chapter, the mode of operation of an LSM
will be explained using a fluorescence-labeled spe-
Pinhole in the
confocal plane
Laser light
source
cimen as an example. Fluorescent dyes, also known
as fluorochromes, are used as markers in most biomedical applications to make the structures of inter-
Principal dichroic
beam splitter
Collimator
est visible. But laser scanning microscopes can just
Scanning mirrors
as well be combined with other microscopic contrast
Objective
techniques such as reflected light or polarization.
An LSM can be easily understood as a modified
Specimen
light microscope supplemented by a laser module
Focal plane
that serves as a light source, and a scanning head
(attached to the microscope stand) that is used to
detect the signal. Signal processing is effected by
an electronic system contained in a box. The
whole system is controlled by a computer.
To generate a confocal LSM image, let us first
excite the fluorescence marker in a defined specimen area with a laser. For this purpose, mono-
13
chromatic light from the laser module is coupled
12
into the scanning head via a fiber optic. In the
10
scanning head, the beam is made parallel by
means of a collimator, and reflected into the
12
11
microscope’s light path by the principal dichroic
10
beam splitter. The objective focuses the excitation
9
11
beam onto a small three-dimensional specimen
10
10
region called the excitation volume. The spatial
9
extension of this volume is directly related to the
16
9
system’s resolving power. The greater the numeri-
3
cal aperture of the objective, the smaller the focal
volume, and the higher the resolution. The posi-
4
tion of the excitation volume can be shifted laterally (in X and Y) by means of two scanning mir-
14
rors, and vertically (in Z) with the microscope’s
5
focusing knob. The current Z position marks the
system’s focusing plane.
6
7
4
8
15
2 1
Excitation light path
The laser focused through the objective
forms a double cone of excitation light
inside the specimen. While the excitation
intensity is strongest at the center of the
double cone (in the focal plane), it is sufficiently high in the planes above and below
the focus to excite fluorescence there, too.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Detection light path
The only fluorescence that
reaches the detector is that
emitted in the focal plane.
Light originating from
other planes is blocked by
a pinhole diaphragm.
The laser light illuminating a particular detail of
The pinhole is essential to the generation of sharp
the specimen in order to excite fluorescence is
images and for the optical sectioning capability.
focused by the objective into the focal plane.
The very designation of confocal laser scanning
Fluorescence excitation and emission are most
microscopy refers to the pinhole, as this is in a
efficient within the focal volume. Therefore, the
plane conjugated to that of the focal plane (con-
fluorescence from labeled structures in the focal
focal plane). The thickness or Z dimension of an
plane forms a sharp image. The laser light, whilst
optical section can be set by motor-driven adjust-
less efficient, is still intensive enough to also excite
ment of the pinhole diameter. Fluorescence light
fluorescently labeled structures above and below
from the focal plane, having passed the pinhole, is
the focal plane. Light emitted there would be
then detected by a photomultiplier. As an LSM
superimposed onto the sharp focal plane image
image is formed sequentially, i.e. pixel by pixel,
and blur it. This is prevented by a pinhole
the detector does not require any spatial resolu-
diaphragm arranged in the ray path, which only
tion. It merely measures the fluorescence intensity
permits light emitted in the focal plane to reach
as a function of time. The image proper is formed
the detector.
only when the intensity measured by the detector
Fiber (from laser source)
Motor-driven collimators
Beam combiner
Primary dichroic beam splitter
Scanning mirrors
Scanning lens
Objective
Specimen
Secondary dichroic beam splitters
Pinholes
Emission filters
Photomultipliers
META detector
Gray filter
Monitor diode
Fiber output
is assigned to the corresponding site of the laser
focus in the specimen. The laser beam is directed
by the two independent scanning mirrors to scan
the specimen in a line-by-line mode. The result of
the scanning process is an XY image that represents a two-dimensional optical section of the
specimen.
5
Two-Dimensional Images
For examining flat specimens such as cell culture
monolayers, it is usually sufficient to acquire one XY
image to obtain the desired information. The same
applies if the specimen is a three-dimensional tissue
section of which a single optical section is representative.
The thickness of the optical section (slice) and the
focal position are selected so that the structures of
interest are contained in the slice. The lateral resolution of a 2D image is defined by the pixel size in
X and Y. The pixel size, in turn, varies with the
objective used, the number of pixels per scan
field, and the zoom factor. Pixels that are too large
degrade resolution, whereas pixels too small
require longer scanning times, bleach the specimen and generate superfluous data volumes. The
optimum pixel size for a given objective and a
given zoom factor can be set by selecting the
number of pixels with a mouse click.
The procedure for a two-dimensional image
1 Position and focus on the specimen in the Vis(ual) mode
2 Select the configuration to match the fluorochromes used
3 Define pixel resolution, scanning speed and, where required, Average Mode
4 Set the optical slice thickness by means of the pinhole diameter
5 Adapt the dynamic range to the specimen;
automatically via Find, or manually via Gain and Offset
6 Adapt the scanning field to specimen substructures, using the Crop function
6
Confocal section through the cerebellum
of a rat.
Green: astroglia cells (GFAP labeling);
red: superoxide dismutase in neurons.
Double labeling of a
Drosophila retina.
Green: actin; red: Crumbs.
Specimen: Dr. O. Baumann,
University of Potsdam,
Germany.
7
Three-Dimensional Images
To record the three-dimensional structure of a spe-
If the sample to be examined is labeled with more
cimen, several two-dimensional optical sections are
than one fluorochrome, it is necessary to adjust
made in different focal planes. The result is an XYZ
the optical slice thicknesses of the various image
image stack, which can be visualized, processed and
channels. The slice thickness is a function of the
analyzed.
numerical aperture of the objective, the wavelength used, and the pinhole diameter. It differs
The optical section is selected by shifting the posi-
for channels detecting light of different wave-
tion of the focus in the specimen. This can be
lengths. In the systems of the Zeiss LSM 510 family,
effected by moving either the objective or the
every detector is equipped with a separate pin-
specimen stage along the Z axis, according to the
hole. This makes it easy to equalize the optical
microscope stand design. Whether the image
slice thicknesses in the software – an important
acquisition exhausts the resolving power given by
condition for 3D colocalization analyses or for
the objective’s numerical aperture depends on the
reconstructing 3D images.
thickness of the optical slice and on the spacing of
Once a 3D stack of images has been recorded, the
two successive sections (the Z interval). According
user has various presentation options. The data
to the Nyquist criterion, the optimum Z interval is
may be displayed as a gallery of depth-coded
equal to half the optical slice thickness. If the pin-
images or as orthogonal projections of the XY, XZ
hole diameter is selected to equal one Airy unit
and YZ planes. To create a 3D impression on a 2D
(1 AU), an optimum compromise between contrast
monitor, animations of different viewing angles
and intensity is achieved for the XY image. The
versus time, shadow projections, and surface ren-
respective settings can be made by a mouse click
dering techniques are possible.
in the software.
The procedure for a three-dimensional image
1 Optimize the recording conditions for an XY image at the center
of the three-dimensional specimen (see box for 2D images)
2 Define the Start and Stop stack limits in the Z Setting menu
3 Define the optimum Z interval in the Z Slice menu
4 Acquire the Z stack
5 Display and analyze the stack in one of the Gallery, Ortho or Cut display modes
6 With multiple-labeled specimens, equalize the optical slice thicknesses in the Z Slice menu
8
Three-dimensional specimen
10µm
The orthogonal projection of the three-dimensional image data
stack permits the raw data stack to be sectioned anywhere in
any of the three mutually perpendicular planes.
Bottom left (above): Horizontal section through the center
of the pollen grain (XY image).
Top: Projection of a vertical section along the horizontal axis
in the XY image.
Bottom right: Vertical section along the vertical axis in the
XY image.
Surface-rendered projection of the pollen grain.
XYZ image stack of a pollen grain. A series of XY images
acquired in different focus positions represents the Z dimension
of the specimen.
9
Time Series
Dynamic processes in living specimens can be
Within a time series, the LSM 510 permits selec-
recorded by means of time series. Data thus
tive, point-accurate illumination of ROIs with laser
acquired can be analyzed “off-line”, i.e. after image
light.
acquisition, or “on-line”, i.e. right during the experi-
This function is useful for generating a photo-
ment, for example in the Online Ratio mode.
bleaching routine within a FRAP experiment
(fluorescence recovery after photobleaching), for
Time series are defined by a start time and the
analyzing dynamic processes, and for the photo-
time interval between two successive images. The
activation or photoconversion of suitable fluo-
series can be started by a mouse click, automati-
rochromes. Complex time series experiments, with
cally at a preselected time, or by some external
different images to be taken at different sites
trigger. To analyze a time series, the Physiology
within a specimen according to a defined time
software option allows fluorescence intensity
pattern, can be defined by means of a special soft-
changes to be quantified in defined regions of
ware option.
interest (ROIs).
The procedure for a time series
1 Define the image dimensions to be recorded versus time (XY image, Z stack, λ stack)
2 Optimize the recording conditions at minimum laser output to avoid or minimize bleaching
3 Define the number of images to be taken and the time interval between
two successive images (Time Interval or Time Delay)
4 Combine with a photobleaching routine if required: define the region to be bleached,
the laser line and its power, the number of bleaching actions,
and the bleaching start time within the series
5 Start the time series with the Start button, at a preselected time, or by an external trigger
10
Evaluation of the experiment
Selection of ROIs
within the specimen.
ROI 1: Cytosol
ROI 2: Cytoplasmic membrane
Ratio
Intensity (I)
Investigation of protein movements
Time series of PKC-GFP transfected HeLa cells.
The stimulation of the cells with PMA at the time t=1 min
leads to a redistribution of PKC from the cytosol to the
cytoplasmic membrane (times in minutes).
Specimen: Dr. S. Yamamoto,
Medical University of Hamamatsu, Japan.
The individual
intensities (upper
graph) and the ratio
of intensities in the
two ROIs marked in
the top picture (lower
graph) illustrate PKC
redistribution.
Colors are assigned
correspondingly.
Time (min)
11
Multifluorescence –
The Crosstalk Problem and Its Solution
If a specimen is labeled with more than one fluo-
One can distinguish between two kinds of cross-
rochrome, each image channel should only show
talk: emission and excitation crosstalk.
the emission signal of one of them.
In a pure emission crosstalk between two fluoIf, in a specimen labeled red and green, part of the
rochromes A and B, the two emission spectra will
green light is detected in the red channel, the phe-
overlap, but the laser lines will excite the dyes
nomenon is known as crosstalk or bleed-through.
independently of each other; i.e. there is no over-
This may lead to misleading results, especially in
lap of the excitations.
colocalization experiments.
Excitation crosstalk would occur if the laser that
excites fluorochrome A also partially excited fluorochrome B.
The problem of emission crosstalk can be solved
by sequential excitation and detection (Multitracking) of the fluorochromes. In case of a combination of excitation and emission crosstalk,
additional spectral information is needed for separating the emission signals.
Emission crosstalk
Section through a mouse kidney, doublelabeled with Alexa 488 (wheat germ
agglutinin) and Alexa 568 (phalloidin).
Simultaneous excitation with 488 and
543 nm. The emission of Alexa 488 is
detected in both the green (BP 505-530 nm)
and red (BP 560-615 nm) channels.
Because of this bleed-through, the areas
labeled with Alexa 488 appear yellow in
the superposition and could be misinterpreted as colocalization with the Alexa 568.
Elimination of emission crosstalk
by Multitracking
If Alexa 488 and 568 are excited and
detected sequentially, no green signal is
detected in the red channel.
Structures labeled with Alexa 488 appear
green in channel superposition.
12
543 nm
laser line
Relative fluorescence
488 nm
laser line
Emission crosstalk of Alexa Fluor 488 and 546
The excitation efficiency of the two fluorochromes is determined
by the point of intersection between the laser line used and the
excitation spectrum (dotted line).
Accordingly, Alexa Fluor 488 is excited to about 80 %,
Alexa Fluor 546 to about 60 %.
At a level of about 5%, the excitation spectrum of Alexa
Fluor 546 is also intersected by the 488 nm laser line (arrow).
Theoretically, this constitutes excitation crosstalk, as one line
excites both markers, but it is inefficient enough to be negligible.
Contrary to this, the emission spectra of the two dyes overlap
significantly.
The red area marks the emission crosstalk of Alexa Fluor 488
occurring if Alexa Fluor 546 is detected to the right of the
543 nm laser line.
350
400
450
500
550
600
650
700
543 nm
laser line
Combined excitation and emission crosstalk
If GFP is used together with YFP, the emission spectra will
overlap considerably.
The red area marks the emission crosstalk between GFP and
YFP occurring if YFP is detected to the right of the 514 nm laser line.
In addition, there will be a pronounced excitation crosstalk.
The 488 nm line excites not only GFP but also YFP to an
efficiency of about 30 % (arrow).
Source: http://home.ncifcrf.gov/ccr/flowcore/welcome.htm;
modified
Relative fluorescence
488 nm
laser line
Wavelength in nm
350
400
450
500
550
600
650
700
Wavelength in nm
13
Spectral Imaging
The acquisition of spectral data becomes necessary
Whereas the beam paths for conventional and
where the overlapping emission signals of multiple-
META detection are identical on the excitation
labeled specimens have to be separated, or where
side, the emission beam for spectral detection,
the cellular parameter to be measured is coded by
after having passed the pinhole, hits a reflective
changes of the emission spectrum (e.g., FRET and
grating. The grating spreads the beam into a spec-
ratio imaging of ion concentrations).
trum and projects it onto the surface of the linear
detector array. Each of the 32 PMT elements in
The LSM 510 META is a system for the fast acqui-
that array thus registers a different part of the
sition of images of high spectral resolution. The
spectrum, each part having a width of 10 nm. The
hardware enabling this functionality consists of a
result is a lambda stack of XY images in which
spectrally dispersive element, a photomultiplier
each image represents a different spectral window.
(PMT) with 32 parallel detection channels (META
Detector), and special electronic circuitry for
detector control and signal amplification.
2
3
META Detector
Part of the beam path in the
LSM 510 META scanning head
1 Confocal pinhole
2 Reflective grating for spectral
dispersion
3 META Detector with
32 separate PMT elements
1
The procedure for a lambda stack
1 Select the spectral range in the Lambda-Mode
2 Activate the excitation laser lines
3 Carefully control the dynamic range to avoid over- and underexposed pixels (Range Indicator)
4 In multiple-labeled specimens, vary the power of the respective laser lines instead of the Amplifier Gain,
in order to match the signal intensities of the fluorochromes
5 Record the lambda stack
6 Display the data in one of the modes: Gallery, Single, Slice, Max or λ-coded
14
λ
X
Y
Lambda stack
All images show the same area, but different
spectral windows of the specimen.
The marker dyes are represented by different parts
of the stack depending on the emission spectrum.
By connecting adjacent detector elements (bin-
The META Detector is good not only for recording
ning), the spectral width of the images can be
lambda stacks, but also as a channel detector in
extended. From a lambda stack, the intensity of
the conventional mode. By binning the respective
the signal for each pixel of the image can be
detector elements in this mode, the optimum
extracted as a function of wavelength. These
spectral bandwidth can be adjusted for any fluo-
spectral “fingerprints” can easily be obtained for
rescent dye.
any image area by means of the Mean of ROI
function. Lambda stacks can be recorded as time
series, Z stacks, or as Z stacks versus time. In the
last-named case, the result would be a five-dimensional image file with the coordinates, X, Y, Z,
lambda and time.
15
Spectral Imaging
With Emission Fingerprinting, autofluorescences
Emission Fingerprinting
are simply included in the unmixing process. The
Emission Fingerprinting is a method for the com-
user can subsequently decide between switching
plete separation (unmixing) of overlapping emission
the autofluorescence channel off and using it to
spectra. It is used with specimens labeled with more
obtain structural information possibly contained in
than one fluorescent dye, exhibiting excitation and
the specimen.
emission crosstalk.
The reference spectra can either be loaded from a
spectra database, or directly extracted from the
The typical raw data for Emission Fingerprinting
lambda stack. For the latter version, the user has
are lambda stacks. The previous chapter described
two options. One is to define spectra via ROIs.
how they are recorded by means of the META
The other uses a statistical method, Automatic
Detector. The second step is to define reference
Component Extraction (ACE), to find the refer-
spectra for all spectral components contained in
ence spectra. In either case, the images of the
the specimen. As a rule, these are dyes interna-
lambda stack must contain structures marked with
tionally used for labeling the specimen. Other pos-
a single fluorochrome only.
sible components are autofluorescent and highly
The third step of Emission Fingerprinting is Linear
reflecting structures. Autofluorescences, in partic-
Unmixing, which converts the lambda stack into a
ular, often have rather broad emission spectra that
multichannel image. Each spectral components of
overlap with the fluorescent markers; this makes
the specimen is then displayed in one channel
them an added source of “impurities” degrading
only. The accuracy of the technique allows the
the signals in conventional laser scanning micros-
complete unmixing even of such dyes whose
copy.
spectra have almost identical emission maxima.
Linear Unmixing
Linear mathematical algorithm for spectral unmixing.
If we regard a pixel of a lambda stack that represents a locus in the specimen
where three fluorescent dyes A, B and C with their spectra S(λ)dye A, B and C overlap,
the cumulative spectrum ΣS(λ) measured can be expressed as
Σ S(λ) = [intensity· S(λ)]dye A + [intensity· S(λ)]dye B+ [intensity· S(λ)]dye C
By means of known reference spectra S(λ)dye A, B and C , the equation can be
solved for the intensities of the dyes A, B and C, which yields the intensity
shares of the three dyes for this pixel. If this calculation is made for each pixel,
a quantitatively correct 3-channel image results, in which each channel
represents a single dye.
16
Intensity
4000
3000
2000
GFP
YFP
1000
0
500
510 520 530 540 550 560 570
Emission wavelength in nm
The 3 Steps of Emission Fingerprinting
1 Recording of a lambda stack
The illustration shows an 8-channel
image of a cell culture transfected
with GFP and YFP. Each image shows
the mean wavelength of the channel.
2 Definition of reference spectra
3 Linear Unmixing
The reference spectra were obtained by
means of lambda stacks of cells singlemarked with GFP and YFP, respectively.
Top: Lambda-coded projections of a cell
marked with GFP (left) or YFP (right).
Bottom: Reference spectra for GFP
(green) and YFP (red).
Using the reference spectra from the
lambda stack, the Linear Unmixing
function generates a two-channel image,
in which each channel represents only
one of the two fluorochromes.
Top: GFP
Center: YFP
Bottom: Both channels superimposed
17
Spectral Imaging
Channel Unmixing
As raw data for unmixing in this case, it is sufficient to have multichannel images in which one
If the emission spectra of fluorescent markers over-
of the marker dyes dominates in each channel.
lap only slightly, the signals can be separated with
Such images can be acquired without the META
the Channel Unmixing function.
Detector, i.e. with an LSM PASCAL, LSM 510 or a
CCD camera.
Channel Unmixing also allows unmixing based on
the excitation behavior of dyes, if the raw data are
multichannel images in which the channels differ
only by their excitation wavelength.
Double labeling of the nervous system
of a zebra fish embryo
Two-channel single-track images with emission crosstalk.
The nerve labeled with Alexa 488 can be discerned (arrows)
in the green (top) and, faintly, in the red channel (center).
Bottom: Superposition of the two channels.
Specimen: Prof. M. Bastmeyer, Friedrich Schiller University of Jena,
Germany.
18
The same images after Channel Unmixing.
The Alexa 488-positive nerve is visible in the green channel
(top) only but vanished from the red one (Center).
Bottom: Superposition of the two channels.
Online Fingerprinting
Here, a reference spectrum is assigned to each
image channel before image acquisition starts.
The functionality of Online Fingerprinting can be
During the experiment proper, lambda stacks are
used to separate overlapping emissions even while
acquired and immediately unmixed in a back-
a time series is being recorded. This may be of
ground operation. The user sees the unmixed mul-
decisive importance where dynamic processes are
tichannel image during the acquisition of the time
investigated.
series. Online Fingerprinting is of advantage especially in spectral FRET experiments and in studies
of dynamic processes with fluorescent proteins.
Visualization of FRET by means of acceptor photobleaching
Expression of a FRET-positive protein construct (CFP linker citrin)
in HEK 293 cells.
Recording conditions: Simultaneous excitation with 458 and 514 nm.
Spectral detection from 462 to 655 nm in Lambda Mode.
Online Fingerprinting and simultaneous display of the two-channel image
(CFP blue, citrin green).
During the combined time-&-bleaching series (bleaching region marked),
citrin (green channel) as a FRET partner is destroyed by intensive
irradiation with 514 nm. The decrease in FRET is visible as an increase
in CFP fluorescence (blue channel).
Specimen: PD Dr. M. Schäfer,
Charité University Hospital,
Berlin, Germany
19
Spectral Imaging
Excitation Fingerprinting
The infrared (IR) light emitted by such lasers can
penetrate tissues to greater depths than visible
By means of tunable excitation lasers such as those
light can. Due to its low phototoxicity, IR light is
used in multiphoton systems, it is possible to detect
suitable for long-time observation of live samples.
also the excitation spectra of fluorochromes. These
Usually, the emission wavelength of these lasers
can be used for unmixing as an alternative to emis-
can be varied continuously to excite the respective
sion spectra.
fluorochrome used in the multiphoton mode.
In Excitation Fingerprinting, this property is used
Multiphoton systems are a special class of confo-
for the acquisition of excitation lambda stacks. For
cal laser scanning microscopes, distinguished from
that purpose, the multiphoton laser is controlled
classical one-photon systems essentially by an
by the LSM software to shift its excitation wave-
additional light source, known as a multiphoton
length by a defined interval before every new
or NLO (non-linear optics) laser.
image. The image stacks thus recorded can be
used for the unmixing of spectral components
differing by their excitation properties, analogously
to the (emission) lambda stacks described before.
For more information on multiphoton microscopy,
refer to the literature cited on the rear cover.
The procedure for Excitation Fingerprinting
1 Define an excitation lambda stack (wavelength range and interval size)
in the Excitation Fingerprinting macro
2 Record the excitation lambda stack
3 Define the reference spectra via single-labeled specimen regions,
single-labeled reference samples, or by using the ACE function
4 Run Linear Unmixing
20
Excitation Fingerprinting separates
widely overlapping emission signals
by their excitation spectra.
1.0
Intensity
0.8
0.6
0.4
0.2
0.0
740
760
780
Excitation wavelength in nm
Retina of a Drosophila fly, labeled for actin
(Alexa Fluor 586 phalloidin);
autofluorescence and emission signal can be cleanly separated
by Emission Fingerprinting.
Specimen: PD Dr. O. Baumann, University of Potsdam, Germany
21
Courses
Laser scanning microscopy
and related techniques
Laser scanning microscopy has become a mainstream technique in biomedical research.
Thorough familiarity with its theoretical principles
and application know-how is a prerequisite for
successful experimentation.
■ Laser scanning microscopy
in biomedical applications
■ Multiple fluorescence labeling
and quantitative colocalization
■ Dynamic investigation of living cells –
Carl Zeiss offers seminars that teach the funda-
FRAP, FLIP, FRET,
mental theory and explain biomedical research
photoactivation and photoconversion
methods, followed by intensive practical hands-on
■ Confocal laser scanning microscopy in
training in small groups.
materials research & quality inspection –
The participants in these courses have access to
principles and applications
various combinations of microscopes and latest
generation LSM systems made by Carl Zeiss.
There is no better way to efficiently acquire new
■ Fluorescence correlation spectroscopy
in biomedical research
know-how and skills in using up-to-date LSM
equipment.
For details, see
www.zeiss.de/courses
22
Summary
Modern laser scanning microscopes are versatile
tools for visualizing cellular structures and analyzing dynamic processes in biomedical research.
Apart from mere imaging, Carl Zeiss laser scanning microscopes are designed for the quantification and analysis of image-coded information.
Among other things, they allow easy determination of fluorescence intensities, distances, areas
and their changes over time. The LSM 510 META,
in particular, is capable of quickly detecting and
quantitatively unmixing the spectral signatures
of fluorescent dyes that closely resemble each
other.
Many software functions analyze important
parameters such as the degree of colocalization
of labeled structures, or the ion concentration in
a specimen.
With their capabilities for acquiring, evaluating
and presenting experimental data, LSM systems
made by Carl Zeiss are tailored to the requirements of scientists of today and tomorrow.
23
REFERENCES
Amiri H., Schulz G., Schaefer M. (2003)
FRET-based analysis of TRPC subunit stoichiometry.
Cell calcium, 33(5-6):463-70.
Carl Zeiss (2000)
Confocal Laser Scanning Microscopy –
Principles. 45-0029 e/09.03
Carl Zeiss (2002)
Colocalization – Analysis and Visualization.
45-0012 e/11.02
Carl Zeiss (2002)
Spectral Separation of Multifluorescence Labels
with the LSM 510 META.
40-546 e/05.02
Diaspro A. (2001)
Confocal and Two-Photon Microscopy :
Foundations, Applications and Advances,
Wiley-Liss, New York
Dickinson M.E. (2001)
Multi-spectral imaging and Linear unmixing
add a whole new dimension to laser scanning
fluorescence microscopy.
Bio Techniques 31/6, 1272-1278.
Gordon G,W. (1998)
Quantitative Fluorescence Resonance
Energy Transfer Measurements Using
Fluorescence Microscopy.
Biophys J, May 1998, p. 2702-2713, Vol. 74, No. 5
Pawley J.B. (1995)
Handbook of Biological Confocal Microscopy.
Plenum Press, New York
Selvin P.R. (2000)
The renaissance of fluorescence resonance energy
transfer. Nat Struct Biol. 7(9):730-4. Review.
LINKS
Carl Zeiss website on
contrasting techniques in light microscopy
www.zeiss.de/contrasts
EAMNET website on FRAP
www.embl-heidelberg.de/eamnet/html/
teaching_modules.html
Carl Zeiss
Advanced Imaging Microscopy
07740 Jena, Germany
Phone: +49 36 41-64 34 00
Fax:
+49 36 41-64 31 44
E-Mail: [email protected]
www.zeiss.de/lsm
Subject to change.
Printed on environment-friendly paper,
bleached without the use of chlorine.
For further information, please contact:
45-0039 e/04.05
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement