tcvgbh1
Febr 2003
Printer HP4500PS
TCVGBH1.DOC
Lifetime Imaging Techniques for Optical Microscopy
Wolfgang Becker, Axel Bergmann
Becker & Hickl GmbH, Berlin, becker@becker-hickl.com, bergmann@becker-hickl.com
1
Contents
Introduction......................................................................................................................................................................... 3
Biological Applications of Lifetime Techniques ............................................................................................................... 3
Fluorescence Quenching ................................................................................................................................................ 3
Resonance Energy Transfer............................................................................................................................................ 4
Separation of Different Chromophores .......................................................................................................................... 5
Diffusion in Cells ........................................................................................................................................................... 5
Microscopy Techniques ...................................................................................................................................................... 6
Wide-Field Fluorescence Microscopy............................................................................................................................ 6
Laser Scanning Microscopy ........................................................................................................................................... 6
Optical Near-Field Microscopy...................................................................................................................................... 7
Light Sources ...................................................................................................................................................................... 8
Titanium-Sapphire Lasers .............................................................................................................................................. 8
Frequency Doubled Titanium-Sapphire Lasers .............................................................................................................. 8
Fibre Lasers.................................................................................................................................................................... 8
Pulsed Diode Lasers....................................................................................................................................................... 8
Modulated CW Lasers.................................................................................................................................................... 9
Mode-locked CW Lasers................................................................................................................................................ 9
Pulse Pickers and Cavity Dumpers................................................................................................................................. 9
Detectors ............................................................................................................................................................................. 9
Photomultiplier Tubes (PMTs) ...................................................................................................................................... 9
Image intensifiers ......................................................................................................................................................... 11
Avalanche photodiodes ................................................................................................................................................ 12
Cornerstones of Fluorescence Lifetime Imaging............................................................................................................... 13
Time Resolution ........................................................................................................................................................... 13
Signal-to-Noise Ratio ................................................................................................................................................... 13
Detection efficiency ..................................................................................................................................................... 13
Recording efficiency .................................................................................................................................................... 14
Sample Saturation ........................................................................................................................................................ 14
Photobleaching............................................................................................................................................................. 14
Signal Processing Techniques........................................................................................................................................... 15
Gated Image Intensifiers .............................................................................................................................................. 15
Modulation Techniques................................................................................................................................................ 17
Single Channel Modulation Techniques.................................................................................................................. 18
Modulated Image Intensifiers.................................................................................................................................. 20
Gated Photon Counters ................................................................................................................................................ 23
Time-Correlated Single Photon Counting (TCSPC) .................................................................................................... 26
TCSPC Imaging ...................................................................................................................................................... 27
Multi Wavelength TCSPC Imaging......................................................................................................................... 29
Features of the TCSPC imaging techniques ............................................................................................................ 32
Other TCSPC Techniques ............................................................................................................................................ 35
Application of TCSPC to diffusion in cells............................................................................................................. 35
TCSPC Wide Field Imaging.................................................................................................................................... 35
Comparison of Signal Processing Techniques .................................................................................................................. 36
Summary ........................................................................................................................................................................... 37
References......................................................................................................................................................................... 38
2
Introduction
Since their broad introduction in the early 90s confocal and two-photon laser scanning microscopes
have initiated a breakthrough in biomedical imaging [1-5]. The applicability of multi-photon
excitation, the optical sectioning capability and the superior contrast of these instruments make
them an ideal choice for fluorescence imaging of biological samples.
However, the fluorescence of organic molecules is not only characterised by the emission spectrum,
it has also a characteristic lifetime. Any energy transfer between an excited molecule and its
environment in a predictable way changes the fluorescence lifetime. Since the lifetime does not
depend on the concentration of the chromophore fluorescence lifetime imaging is a direct approach
to all effects that involve energy transfer [6-10]. Typical examples are the mapping of cell
parameters such as pH, ion concentrations or oxygen saturation by fluorescence quenching, or
fluorescence resonance energy transfer (FRET) [6,7,8] between different chromophores in the cell.
Furthermore, combined intensity / lifetime imaging is a powerful tool to distinguish between
different fluorescence markers in multi-stained samples and between different natural fluorophores
of the cells themselves. These components often have ill-defined fluorescence spectra but are clearly
distinguished by their fluorescence lifetime [70].
Fluorescence lifetime measurements require a pulsed or modulated excitation source, a sufficiently
fast detector and a suitable recording electronics. This article gives an overview about applicable
time-resolved signal recording techniques and their benefits and drawbacks.
Biological Applications of Lifetime Techniques
When a dye molecule absorbs a photon it goes into an Excited State
excited state from which it can return by the emission Population
of a fluorescence photon, by converting the absorbed
Laser
energy internally, or by transferring the energy to the
Emission +
environment (fig. 1) [6,9,10]. The probability that one
Internal Conversion
of these effects occurs is independent of the time after
e -t/ 0
e -t/
the excitation. If a large number of similar molecules
Energy
transferred to Environment
with similar local environment is excited by a short
laser pulse the fluorescence decay function is therefore
Time
single exponential. As long as no energy is transferred
to the environment the lifetime is the ‘natural Fig. 1: Return of molecules from the excited state
fluorescence lifetime’, τ0 which is a constant for a given molecule and refraction index of the
solvent. The fluorescence decay times of the fluorophores commonly used in microscopy are of the
order of a few ns.
Fluorescence Quenching
If energy is transferred to the environment the actual Intensity
Laser
fluorescence lifetime, τ, is less than the natural lifetime,
τ0. For almost all dyes the energy transfer rate depends
e -t/
unquenched lifetime
more or less on the concentration of ions, on the oxygen
concentration, on the pH value or on the binding to
proteins in a cell [6,9,10]. Therefore, specifically designed
e -t/
quenching
dyes can be used to probe the concentration of biological
relevant ions such as Na+, Mg++, or Ca++, the oxygen
concentration or the pH value inside a cell. There is a Fig. 2: Fluorescence quenching
direct relation between the lifetime and the quencher
0
Time
3
concentration, fig. 2.
Fluorescence markers used to reveal particular protein structures in cells often bind to a variety of
slightly different targets. Although this often does not cause significant changes in their spectral
behaviour the lifetime can be clearly different due to different quenching efficiency. Therefore the
lifetime of markers can be used as an additional probe technique in cells [11, 12]. A wide variety of
chromophores including CFP, GFP and YFP clearly show variations in their fluorescence lifetimes.
Although most of these effects are not investigated in detail yet there is probably a large potential
for ‘intelligent’ markers based on lifetime changes.
Fluorescence can be almost entirely quenched in aggregates of fluorophore molecules. Lifetimes as
short as 20 ps have been found. The fast relaxation is often considered unfavourable for
photodynamic therapy applications. However, if the aggregates monomerise inside a tumor cell the
effect can be of particular interest [52].
Resonance Energy Transfer
A particularly efficient energy transfer process is fluorescence resonance energy transfer, or FRET.
FRET occurs if two different dyes are present with the emission band of one dye overlapping the
absorption band of the other [6,7,8]. In this case the energy from the first dye, the donor, goes
immediately into the second one, the acceptor. This results in an extremely efficient quenching of
the donor fluorescence and, consequently, decrease of the donor lifetime, see fig. 3.
Intensity
Laser
Intensity
Donor
Absorption
Donor
Emission
Acceptor
Absorption
Laser
Acceptor
Emission
e -t/
D
D
A
A
0
Emission
e -t/
FRET
quenched
donor
Wavelength
unquenched donor
Time
Fig. 3: Fluorescence Resonance Energy Transfer (FRET)
FRET works only over a distance shorter than a few nm. Therefore, it can be used to probe the
distance between different subunits in the cell.
It is difficult to obtain quantitative FRET results from steady-state images. The fluorescence
intensity does not only depend on the FRET efficiency but also on the unknown concentration of the
dyes. Moreover, some of the acceptor molecules are excited directly, and the donor emission band
extends into the acceptor emission. Up to eight measurements at different excitation wavelength
and in different emission wavelength bands are required to obtain calibrated FRET results from
steady state data [4,5,7]. FRET results can also be obtained by measuring the donor fluorescence,
then photobleaching the acceptor, and measuring the Intensity
donor once more. The FRET efficiency is given by the
ratio of the two donor images. Although this procedure b quenched
looks reliable at first glance photobleaching products
b e -t /
may induce damage in the cell, and in a living cell
diffusion may replace a part of the photobleached
unquenched
acceptor molecules.
a
FRET
ae
-t /
0
In lifetime data, however, FRET shows up as a
time
dramatical decrease of the donor lifetime [6,8,68,72].
Fig. 4: Fluorescence decay components in
The fluorescence decay functions contain the FRET systems
fluorescence of quenched and of unquenched donor
4
molecules and are therefore double-exponential, fig. 4. Qualitative FRET results can be obtained
from the lifetime of a single exponential approximation of the decay curve. Quantitative
measurements require double exponential decay analysis that delivers the lifetimes, τ0 and τFRET,
and the intensity factors, a and b, of the two decay components [68,72]. The relative numbers of
quenched and unquenched molecules is given by the ratio of the two intensity components, b/a,
while the average coupling efficiency of the FRET pairs is given by τ0 / τFRET. In principle, both the
ratio of quenched and unquenched molecules and the coupling efficiency can be derived from a
single donor lifetime measurement.
Separation of Different Chromophores
Steady-state multi-wavelength techniques have been developed that efficiently separate different
fluorescence markers by unmixing their fluorescence spectra [13]. However, not all marker
combinations can efficiently be resolved. Even the well known GFP and YFP are difficult to unmix.
Autofluorescence images of cells and tissue show a wide variety of fluorescence components with
ill-defined, variable, and often unknown spectra. When spectral unmixing fails the components can
usually be distinguished by their different lifetimes [70,75]. The relative concentration of two
components can be determined by double-exponential decay analysis. Even if only a single
exponential approximation, i.e. an average lifetime is measured, the contrast in the fluorescence
images can be considerably improved. [70,75]. Moreover, changes in the relative concentration and
the lifetime of autofluorescence components can possibly be used as diagnostic tools.
Significant progress can be expected from combining spectral unmixing and lifetime analysis.
Recording of time- and wavelength resolved data is technically possible by TCSPC techniques (see
‘Multi Wavelength TCSPC Imaging’). Combining the two methods requires to develop suitable
data analysis software.
Diffusion in Cells
Diffusion time constants in cells are usually in the ms range and below. They are usually determined
by fluorescence correlation (FCS) techniques [14,15,16]. The problem is that the correlation
technique is a single point measurement. Moreover, the measurement is usually not done in the
same setup as the cell imaging. This makes it difficult to identify the measured spot in a particular
cell with sufficient accuracy. Basically the photon counting techniques used for lifetime
measurement are able to run a combined FCS and lifetime measurement at a single, well defined
point of the sample. This could not only help to solve the positioning problem but also to identify
single marker molecules [17] and to reveal conformational changes of the diffusing marker/protein
clusters [18]. Although appropriate photon counters are available [63] no combined FLIM / FCS
setup has become known yet.
5
Microscopy Techniques
Wide-Field Fluorescence Microscopy
As in any traditional microscope, the sample is uniformly
illuminated by the excitation light. The image is detected by a
CCD camera (fig. 5). To obtain time resolution in a wide-field
setup the excitation light is pulsed or modulated and a gain
modulated or gated intensified CCD camera is used [19,20,21,23].
The restriction to a special detection technique with sub-optimal
efficiency is a drawback of wide-field imaging. Furthermore,
wide-field imaging does not have the inherent depth resolution of
the scanning techniques (see section below). However, there is a
remarkable technique to obtain 3D images in whole-field
microscopes [22], and a time-resolved solution with a gated CCD
camera is described in [23]. The most severe drawback of the
wide-field technique is that it cannot be used for two-photon
excitation - a technique that allows deep tissue imaging in
scanning microscopes.
Gated or
modulated
intensified
CCD Camera
Pulsed or
modulated
Light Source
Dichroic
Mirror
Excited
Objective
Lens
Sample
Fig. 5: Whole-field fluorescence
lifetime imaging
A benefit of wide-field imaging is that it probably causes less photobleaching than the scanning
techniques. Provided the same number of detected photons and the same detection efficiency, the
overall excitation dose for wide-field imaging is the same as for the scanning techniques described
below. However, the excitation density is much smaller because the light is not focused. Since the
increase of photobleaching with the power is nonlinear it should be expected that wide field
imaging is less affected by photobleaching than scanning techniques. Wide field techniques are
therefore superior for time-lapse imaging, i.e. for recording sequences of images in millisecond
intervals. For fluorescence lifetime imaging the situation may be less favourable because the
detection techniques applicable for wide-field imaging are not the most efficient ones [39].
Laser Scanning Microscopy
The general optical principle of a laser scanning
microscope [1,2] is shown in fig. 6.
Laser
Pinhole
Pinhole
Laser
fs Pulses
The laser is fed into the optical path via a
Dichroic
Dichroic
dichroic mirror and focused into the sample by
Mirror
Mirror
Detector
Detector
the microscope objective lens. In the traditional
Scanner
Scanner
confocal setup used for one-photon excitation
Detector
(fig. 6, left) the light from the sample goes back
Dichroic
Mirror
through the objective lens, through the scanner,
is diverted by a dichroic mirror and goes
Objective
Objective
through a pinhole in the upper image plane of
Lens
Lens
Excited
Excited
the objective lens. Light from outside the focal
Sample
Sample
plane is not focused into the pinhole and
One-Photon Excitation
Two-Photon Excitation
therefore substantially suppressed. X-Y imaging
is achieved by optically scanning the laser spot
Fig. 6: General setup of a laser scanning microscope
over the sample, Z imaging (optical sectioning)
is possible by moving the sample or the microscope up and down.
With a fs Ti:Sa laser the sample can be excited by two-photon absorption [3,54,78]. Excitation
occurs only in the focus, so that no pinhole is required to reject light from outside the focal plane.
6
Therefore, the fluorescence light need not be fed back through the scanner and through the pinhole.
It can be diverted by a dichroic mirror directly behind the microscope objective. This setup is called
‘non-descanned detection’ in contrast to the ‘descanned detection’ shown above.
Two-photon excitation in conjunction with non-descanned detection can be used to image tissue
layers as deep as 500 µm. Since the scattering and the absorption at the wavelength of the twophoton excitation are small the laser beam penetrates through relatively thick tissue. Even if there is
some loss on the way through the tissue it can easily compensated by increasing the laser power.
The increased power does not cause much photodamage because the power density outside the
focus is small. However, as long as there are enough ballistic (non-scattered) excitation photons in
the focus the fluorescence is excited. Of course, the fluorescence photons are heavily scattered on
their way out of the tissue and therefore emerge from a relatively large area of the sample surface
which is out of the focus of the objective. However, for non-descanned detection there is no need to
focus the fluorescence light anywhere. Therefore the fluorescence photons can be efficiently
transferred to the detector.
It is sometimes believed that lifetime imaging is somehow connected to two-photon excitation. This
is, of course, not correct. Depending on the signal processing technique used, lifetime imaging
requires a pulsed or modulated laser. Although a Ti:Sa laser is the ideal source lifetime imaging is
possible with the frequency doubled Ti:Sa laser, with pulsed diode lasers, or with modulated CW
lasers.
Optical Near-Field Microscopy
The optical near-field microscope (SNOM or NSOM) combines the principles of the atomic force
microscope and the laser scanning microscope [24-28]. A sharp tip is scanned over the sample and
kept in a distance comparable to the diameter of a single molecule. The tip can be the end of a
tapered fibre through which the laser is fed to the sample (fig. 7, left). Or, the tip is illuminated by
focusing the laser through the microscope objective on it and the evanescent field at the tip is used
to probe the sample structure (fig. 7, right). In any case, the fluorescence photons are collected
through the microscope objective.
Laser
Fibre
lever
lever
lever
scan
lever
scan
Detection
Sample
Sample
Microscope
Objective
Microscope
Objective
Detection
Laser
Fig. 7: Optical near-field microscope
The optical near-field microscope reaches a resolution of a few 10 nm, i.e. about 10 times less than
the laser scanning microscope. Imaging cells with this technique is difficult and possibly restricted
to special cases [27,28]. The realm of the optical near-field microscope are certainly applications
where fluorescing molecules or nano-particles are fixed on a flat substrate.
Generally, the SNOM principle can be combined with flourescence lifetime imaging in the same
way as the normal laser scanning microscope. Since only a small number of photons can be
obtained from the extremely small sample volume photon counting techniquesare used to obtain
lifetime information [78]. The problem to be expected is that the proximity of the tip changes the
7
fluorescence lifetime of the sample. Whether this effect makes lifetime imaging in a SNOM useless
or whether it can be exploited for completely new techniques is hard to say.
Light Sources
Depending on the signal processing technique used, fluorescence lifetime imaging requires either a
pulsed excitation source with a repetition rate in the MHz range or a modulated light source with a
possible variable modulation frequency of 50 MHz to 1 GHz and a near perfect modulation depth.
Titanium-Sapphire Lasers
The ultimate solution is the femtosecond Ti:Sa laser. These lasers deliver pulses with 70 to 80 MHz
repetition rate, 80 to 200 fs pulse width and up to several Watts average power. The wavelength is
in the NIR from 780 nm to 950 nm. To excite the sample which usually absorbs below 500 nm,
simultaneous two photon excitation is used. Due to the short pulse width and the high energy
density in the focus of the microscope the two-photon process works very efficiently. Therefore the
traditional frequency doubling of the Ti:Sa radiation is not often used for laser scanning
microscopes.
Frequency Doubled Titanium-Sapphire Lasers
Frequency doubled titanium-sapphire lasers can be used to excite the sample via the traditional onephoton absorption. Frequency doubling is achieved by a nonlinear crystal. The output power is in
the mW range. Less than 50µW are required to excite a typical sample so that the available power is
by far sufficient. Whether one-photon or two-photon excitation gives less photobleaching is still
under discussion. In a some cases considerably higher count rates and less photodamage for onephoton excitation were reported [72].
Fibre Lasers
Another useful excitation source are fibre lasers. Fibre lasers are available for a wavelength of
780 nm and deliver pulses as short as 100 to 180 fs [29]. The average power is 10 to 20 mW. This is
less than for the Ti:Sa laser but still sufficient for two-photon excitation. As a rule of thumb, the
maximum useful power for biological samples and fs NIR excitation is 5 to 10 mW. A higher power
kills the cells or cooks the sample. The benefit of the fibre laser is the small size, the high reliability
and a lower price compared to the Ti:Sa laser. The drawback of the fibre laser is that it is not
tuneable.
Pulsed Diode Lasers
A reasonable cost solution for one-photon excitation are
pulsed diode lasers which are available for the NUV, blue,
red, and near-infrared range [30-32]. These lasers deliver
pulses with 40 to 400 ps duration and up to 80 MHz
repetition rate. The average power is up to a few mW.
However, the pulse width increases with the power. For
pulses shorter than 100 ps the average power is of the order
of a few 100 µW. Fig. 8 shows the typical pulse shape of a
405 nm picosecond diode laser.
Pulsing diodes with less than 100ps width requires special
driving techniques that are not commonly available.
However, pulses as narrow as 300 ps from red diodes can
easily be obtained by connecting a commercially available
pulse generator to a bare laser diode.
8
Fig. 8: Pulse shape of a blue diode laser. BDL405, recorded with R3809U MCP and SPC-830
TCSPC module. FWHM is 80 ps.
It has been shown that diode lasers can be used for time-resolved microscopy with good results [33,
33a]. Unfortunately the beam quality of diode lasers is not very good. Therefore it can be difficult to
obtain a diffraction-limited resolution. However, if only the central part of the beam is used, the
result can be quite acceptable. Discarding a large fraction of the beam causes a considerable loss of
power. This loss is, however, not substantial because 50 µW in the focal plane are sufficient to
excite the sample.
Modulated CW Lasers
For signal processing techniques based on phase measurement of modulated signals CW Ar+ and
HeNe lasers in conjunction with an acousto-optical modulator can be used. However, care must be
taken to avoid any crosstalk of the modulation frequency into the detection system. The smallest
amount of crosstalk makes an accurate phase measurement impossible.
Using modulated CW lasers in conjunction with photon counting techniques is not recommended.
These techniques require pulses rather than sinewave signals. Acousto-optical modulators are
resonance systems unable to deliver sufficiently short pulses with high on/off ratio.
Mode-locked CW Lasers
Ar+ lasers can be actively mode-locked. By introducing a modulator into the laser cavity pulses as
short as 100 ps with 80 to 120 MHz repetition rate are obtained. The pulses can be used directly or
to pump a jet-stream dye laser that delivers ps pulses at a wavelength tuneable from 500 to 600 nm .
Although the light from these lasers can be used for fluorescence excitation [65] the systems are
often unstable and require permanent maintenance, checking and re-adjustment. We strongly
discourage to use these lasers as excitation sources for time-resolved microscopy.
Pulse Pickers and Cavity Dumpers
Pulse pickers and cavity dumpers are used to obtain a lower repetition rate from lasers running at a
high, fixed repetition rate. For fluorescence measurements they are sometimes used to measure
lifetimes longer than the original period of the laser. The problem of the devices is the poor on-off
ratio and the electrical noise they often produce. It is by far better to take some incomplete decay
into regard in the data analysis than to cope with electrical noise and satellite pulses. Furthermore, a
low repetition rate reduces the useful count rate in photon counting setups and possibly increases
photobleaching. Don’t use pulse pickers if they are not absolutely necessary.
Detectors
Photomultiplier Tubes (PMTs)
The most common detectors for low level detection of light
are photomultiplier tubes. A conventional photomultiplier
tube (PMT) is a vacuum device which contains a
photocathode, a number of dynodes (amplifying stages) and
an anode which delivers the output signal (fig. 9).
D2
PhotoCathode
D1
D3
D4
D6
D5
D7
D8
Anode
The operating voltage builds up an electrical field that
Fig. 9 Principle of a conventional PMT
accelerates the electrons from the photocathode to the first
dynode D1, from D1 to D2, further to the next dynodes, and
from D8 to the anode. When a photoelectron emitted from the photocathode hits D1 it releases
several secondary electrons. The same happens for the electrons emitted by D1 when they hit D2.
The overall gain reaches values of 106 to 108. The secondary emission at the dynodes is very fast,
therefore the secondary electrons resulting from one photoelectron arrive at the anode within a few
9
ns or less. Due to the high gain and the short response a single photoelectron yields a easily
detectable current pulse at the anode.
A similar gain effect is achieved in the channel plate of a
microchannel PMT, fig. 10. The channel plate consists of
millions of narrow parallel channels. The channels have a
diameter below 10 µm and are coated with a conductive
material. When a high voltage is applied across the plate the
walls of the channels act as secondary emission targets. With
two plates in series, a gain of the order of 106 is achieved.
MCP PMTs deliver extremely fast pulses with low transit
time jitter.
Channel
Plate
Channel Plate
Anode
Electrons
to
Anode
Photo
Electron
Electrical Field
Cathode
Fig. 10 Multichannel PMT
There are two parameters that characterise the time resolution of a photomultiplier - the ‘Single
Electron Response’, SER, and the ‘Transit Time Spread’, TTS.
The SER is the output pulse for the detection of a single photon. The width of the SER limits the
resolution of a PMT when it is used as a linear detector, i.e. a oscilloscope of fast digitizer. Some
typical SER shapes for PMTs are shown in fig. 11.
Iout
1ns/div
Standard PMT (R928)
1ns/div
Fast PMT (R5600, H5783)
1ns/div
MCP-PMT (R3809U)
Fig. 11: Single Electron Response (SER) for typical PMTs
For photon counting applications the resolution is not limited by the width of the single electron
response. For these techniques only the Transit Time Spread (TTS), i.e. the uncertainty of the delay
between the photon detection and the output pulse is important. The TTS can be 10 times smaller
than the width of the SER - a serious argument to use photon counting techniques in conjunction
with photomultipliers.
Due to the random nature of the detector gain, the pulse amplitude is not
stable but varies from pulse to pulse. The pulse height distribution can be
very broad, up to 1:5 to 1:10. Fig. 12 (right) shows the SER pulses of an
R5600 PMT. The pulse height jitter introduces an additional noise factor
into all measurements that use the PMT as a linear detector. The signal-tonoise ratio of photon counting measurements is not - or almost not impaired by the pulse height jitter.
If a PMT is operated near its full gain the peak current of the SER pulses is
of the order of a few mA. This is much more than the allowed continuous
output current. Consequently, for high repetition rate signals or steady state
operation the PMT delivers a train of random pulses rather than a
continuous signal. Because each pulse represents the detection of an
individual photon the pulse density - not the pulse amplitude - is a measure
for the light intensity at the cathode of the PMT. Obviously, the pulse
10
Fig. 12: Amplitude jitter of
SER pulses (R5600)
density is measured best by counting the PMT pulses within subsequent time intervals. Therefore,
the application of photon counting techniques is the logical consequence of the high gain and the
high speed of photomultipliers.
The efficiency, i.e. the probability that a particular photon causes a pulse at the output of the PMT,
depends on the efficiency of the photocathode. Unfortunately the sensitivity S of a photocathode is
usually not given in units of quantum efficiency but in mA of photocurrent per Watt incident power.
The quantum efficiency QE is
hc
QE = S ---- =
eλ
S
---λ
.
1.24.106
The efficiency for the commonly used
photocathodes is shown in fig. 13 (right). The
QE of the conventional bialkali and
multialkali cathodes reaches 20 to 25 %
between 400 and 500 nm. The recently
developed GaAsP cathode reaches 45 %. The
GaAs cathode has an improved red sensitivity
and is a good replacement for the multialkali
above 600 nm.
Wm
----A
Sensitivity
1000
mA/W
QE=0.5
GaAsP
GaAs
100
QE=0.2
QE=0.1
bialkali
multialkali
10
Generally, there is no significant difference
between
the
efficiency
of
similar
photocathodes in different PMTs and from
1
different manufacturers. The differences are of
300
400
500
600
700
800
900
the same order as the variation between
nm
Wavelength
different tube of the same type. Reflection
type cathodes are a bit more efficient than Fig. 13: Sensitivity of different photocathodes [34]
transmission type photocathodes. However,
reflection type photocathodes have non-uniform photoelectron transit times to the dynode system
and therefore cannot be used in ultra-fast PMTs. A good overview about the characteristics of PMTs
is given in [34] and [79].
Image intensifiers
Image intensifiers are vacuum devices consisting of
a
photocathode,
an
acceleration
and/or
multiplication system for the photoelectrons, and a
two-dimensional image detection system.
First generation systems used a electron-optical
imaging system that accelerates the photoelectrons
to an energy of some keV and sends them to a
fluorescent screen. The image from the screen was
detected by a traditional camera or later with a
CCD camera. First generation devices had a
relatively low gain and strong image distortions.
-HV
Photocathode
Multichannel Plate
Fluorescence Screen
CCD Camera
Fig. 14: Intensified CCD camera
11
Second generation image intensifiers use multichannel plates for electron multiplication (fig. 14).
One plate gives a typical multiplication factor of 1000 so that a gain of 106 can be achieved by two
plates in series.
The CCD chip can be placed inside the tube to detect the electrons directly. These EBD CCDs
(Electron Bombarded CCDs) give higher gain than a CCD behind a fluorescent screen.
Gating of an image intensifier can be accomplished by a grid behind the photocathode. Gain
modulation can also be achieved by modulating the voltage between the cathode and the channel
plate or the voltage across the multichannel plate.
In general, image intensifiers use the same photocathodes as photomultiplier tubes. Therefore, the
detection efficiency is approximately the same. There can, however, be an appreciable loss of
photons or signal-to-noise ratio due to gating or modulating.
Avalanche photodiodes
Avalanche photodiodes (APDs) use a multiplication effect
Quenching Circuit
due to a strong electric field in a semiconductor structure
200V
(fig. 15). For use as a linear detector, a gain factor of the
order of 100 can be achieved. However, cooled avalanche
Photon
photodiodes can be used to detect single photons if they are
Avalanche
Output
operated close to or slightly above the breakdown voltage.
The generated electron-hole pairs initiate an avalanche
breakdown in the diode. Active or passive quenching
15: Single Photon Avalanche Photodiode
circuits must be used to restore normal operation after each Fig.
(SPAPD)
photon [36]. Therefore, a single photon avalanche
photodiode (SPAPD) can only be used for photon counting. The advantage of an APD is the high
quantum efficiency that can reach 90% in the near infrared. Single photon avalanche photodiode
(SPAPD) modules are available from Perkin Elmer [35]. Although a resolution as fast as 20 ps has
been reported for especially manufactured SPAPDs [36] the time resolution of these modules is in
the order of 300 to 800 ps and depends on the count rate. This makes them less useful for lifetime
measurements. SPAPDs are, however, excellently suited for fluorescence correlation measurements
which do not require sub-ns resolution.
12
Cornerstones of Fluorescence Lifetime Imaging
Time Resolution
The lifetimes of highly efficient fluorescence markers are typically in the region of a few ns.
However, these dyes are selected for high quantum yield. Lifetimes of less efficient chromophores
can easily be below 100 ps. Quenching effects can reduce the lifetime down to a few 10 ps, and the
lifetime of the quenched donor fluorescence in FRET experiments is in the order of 100 to 300 ps.
Therefore, a good lifetime system should resolve fluorescence decay functions down to the order of
10 ps.
The fluorescence decay curves of in biological samples are often multi-exponential. There can be
several chromophores in the same part of a cell, a single chromophore can be quenched witch nonuniform efficiency, or there can be quenched and unquenched molecules in the same part of the cell.
Therefore, the ability to resolve multi-exponential decay functions is an absolute requirement to get
quantitative results. Resolving two or even more exponential terms in a decay function requires data
with an excellent signal-to-noise ratio.
Signal-to-Noise Ratio
Due to the short lifetimes, the measurement of fluorescence decay functions requires a detection
bandwidth in the GHz range. The high bandwidth does not only demand for very fast detectors and
detection electronics, it poses also a noise problem. The noise in fast optical measurements is almost
essentially shot noise, i.e. the fluctuation of the number of photons detected within the resolving
time of the measurement system. The best signal-to-noise ratio, SNR, that can be achieved is
SNR = n1/2
with n being the number of photons detected within the resolved time interval. Actually the SNR
can be even lower due to background signals originating in the detector or coming from the
environment, random gain fluctuations in the detector, and inefficient acquisition of the detected
photons in the subsequent signal processing chain.
Acceptable steady state images can be obtained for less than 100 photons detected per pixel of the
image. However, lifetime measurements actually deliver a stack of images for - in case of timedomain methods - different times after the excitation or - in case of frequency domain methods - for
different phases and modulation frequencies. Therefore, the number of photons required to get
lifetime information is much larger. Although rough lifetime information for single exponential can
be obtained from only 100 detected photons high accuracy measurements for multi-exponential
decay analysis can easily require 10.000 or 100.000 photons per pixel [80].
Unfortunately the number of photons that can be emitted from the sample in a given time interval is
limited by the sample itself.
Detection efficiency
The detection efficiency for the photons emitted be the sample depends on the optical system and on
the detector. The efficiency of a microscope depends on the numerical aperture of the microscope
objective, NA, and increases with NA2. The effective NA can be doubled by using the 4Pi technique
[37].
It is often claimed that the efficiency for non-descanned detection is considerably higher than for
descanned detection. This is certainly true for deep-tissue imaging when the emission light is
scattered in the sample and cannot be fed through a pinhole. However, state-of the art microscopes
have pinholes with adjustable diameter. For imaging single cells there is no noticeable difference
between descanned and non-descanned detection.
13
Appreciable loss of photons can occur in the filters used to select the desired emission wavelength
range. Using the right filter for a particular chromophore can improve the efficiency considerably.
Commonly used detectors for lifetime imaging are described below. Although there is no
appreciable difference between detectors of the same cathode type not all cathodes may be available
for a particular detector. Techniques that are not restricted to a special detector, i.e. the photon
counting techniques, can use high efficiency detectors, i.e. PMTs with GaAs cathodes or single
photon avalanche photodiodes.
Recording efficiency
Different signal processing techniques differ considerably in terms of recording efficiency, i.e. in the
exploitation of the detected photons. Taking into regard that the available number of photons is
limited by photobleaching in the sample, the recording efficiency is the most important parameter
next to the time resolution. The quality of a signal recording technique can be described by a ‘figure
of merit’ [38, 39], F , that describes the ratio of the signal-to-noise ratio SNR of an ideal
measurement to the signal-to-noise ration actually achieved, i.e.
SNRideal
F = ------------SNRact
F values for different methods were determined in [39]. Since the SNR is proportional to the square
root of the number of detected photons, n, the efficiency of a technique in terms of photons required
to obtain a given SNR is
E=1/F2
The figure of merit and the efficiency will be used in the discussion of the commonly used lifetime
techniques.
Sample Saturation
The sample volume from which the photons are detected in a laser scanning microscope is of the
order of 1 fl. For a chromophore concentration of 10-6 mol/l this volume contains only 600
molecules, and for 10-9 mol/l the number of molecules is of the order of 1. Lifetime measurements
require a pulsed or modulated light source. To avoid saturation effects for pulsed excitation only a
small fraction of the molecules can be excited by each pulse. Therefore, lasers with repetition rates
in the kHz range cannot reasonably be used in conjunction with scanning. There is, however, no
saturation problem if lasers with a repetition rate above 50 MHz such as Ti:Sa lasers, YLF lasers or
pulsed laser diodes are used.
Photobleaching
The most severe limit for the emission intensity is set by photobleaching in the sample. The
chromophores are not infinitely stable but are destroyed after a large number of absorption and
emission cycles. The reasons of photobleaching are not clear in detail and are probably different for
one-photon and two-photon excitation . Possible reasons are simultaneous multi-photon absorption,
intersystem crossing, reactions from the triplet state and excited-state absorption [40-45].
It is often claimed that photodamage and photobleaching is smaller for two-photon excitation.
Certainly, there is no absorption outside the focus and consequently above and below the focal
plane. Moreover, a cell can withstand much more power in the NIR than in the NUV because the
fraction of absorbed power is much smaller [46,47]. However, if the cell contains dyes with a strong
absorption around 400 nm and photobleaching is compared for the same number of emitted
14
fluorescence photons the situation is less clear. It has also been found that photobleaching is more
rapid for two-photon excitation [41,42].
For two-photon excitation the dependence of the photobleaching efficiency on the excitation power
is highly nonlinear. For photobleaching versus excitation power exponents of 2.5 [45] and from 3 to
5 for have been found [43]. At the same time the emission followed the excitation intensity by the
expected power of 2. That means photobleaching increases more than linearly with the emission
intensity. Therefore two-photon excitation can easily cause 10 times faster photobleaching than onephoton excitation for the same emission intensity [43].
Although photobleaching is the most crucial constraint for scanning microscopy the question about
the excitation method is still open. The consequence from the controversial situation is not to rely
on two-photon excitation alone. For all lasers commonly used for two-photon excitation frequency
doubling is available and delivers sufficient power for one-photon excitation.
Signal Processing Techniques
Gated Image Intensifiers
Gating an image intensifier is achieved by placing a
grid behind the photocathode. The principle is
similar as for the grid in a radio tube. As long as the
grid voltage is negative referred to the
photocathode the photoelectrons cannot pass the
grid. When a positive pulse is applied the electrons
pass through the meshes of the grid and are
accelerated towards the multichannel plate or into
the acceleration system.
-HV
Photocathode
Gate
Gate
pulse
Multichannel Plate
Fluorescence Screen
CCD Camera
Fig. 16: Gated Image Intensifier
Although gating of an image intensifier looks straightforward at fist glance it is anything but simple,
particularly if sub-ns resolution is to be achieved. Even if a sufficiently short gating pulse can be
generated electronically the electrical field between the photocathode and the grid must follow the
pulse at the same speed. Because the conductivity of the photocathode is relatively low the time
constant formed by the gate-cathode capacitance and the cathode resistance limits the switching
speed. Furthermore, a variable lateral field builds up in front of the photocathode that distorts the
image and impairs the image resolution. Manufacturers counteract these effects by using high
conductivity photocathodes which, however, compromises sensitivity. High efficiency GaAs and
GaAsP photocathodes as they are used in PMTs have particularly low conductivity and are most
likely not applicable for gated image intensifiers.
Another RC time constant exists between the grid and the multichannel plate. Although the change
of the field in front of the plate has only small influence on the gating performance it induces a
lateral current in the multichannel plate that heats the device at high gate repetition rates.
The gating resolution can also be impaired be electron-optical effects. When the gate voltage in the
setup of fig. 16 is negative a cloud of photoelectrons builds up between the cathode and the grid.
When a gate pulse is applied to the grid these electrons pass the grid and enter the detection system.
Depending on the grid geometry, the lifetime of the photoelectrons between the grid and the cathode
can be of the order of some 100 ps.
The effects described above can be reduced by additional grids. Even then a lateral change of the
gate delay due to the wave propagation in the grid structure remains. This effect is, however,
predictable and can be corrected in the recorded data.
15
Standard gated image intensifier devices have a minimum gate width of the order of a few ns. A
device with 5ns gate width has been used to determine single exponential decay constants down to a
few ns by deconvolution [33]. The shortest gate width obtained with gated image intensifiers is 50
ps for low repetition rate applications and 200 ps at a repetition rate of 80 MHz [20,51].
The general setup of a wide field and a scanning microscope with a gated image intensifier is shown
in fig. 17.
Reference
Pulsed
Laser
Pulsed
Laser
Variable
Delay
Reference
Gate
Scan
Control
Variable
Delay
Scanning Head
Scan
Mirror
Telescope
Gate
Generator
Gate
Scan
Control
Scan
Mirror
Gate
Generator
Scan
Lens
Gate
Gate
CCD
CCD
Image
of
Sample
-HV
Microscope
Objective
Sample
Image
of
Sample
Gated
Image
Intensifier
+ MCP -
Gated
Image
Intensifier
+ MCP -
-HV
Objective
Sample
Fig. 17: Wide field microscope (left) and scanning microscope (right) with gated image intensifier
For time-resolved imaging a series of images is recorded for different delays of the gate pulse
referred to the laser pulse (fig. 18).
In a wide field microscope with a high repetition rate laser,
such as a Ti:Sa or YLF laser with frequency doubler, the gate
delay can be controlled idependently of the laser pulse
sequence. The acqusition time for each image is simply
chosen to get a sufficient signal-to-noise ratio. If a laser with
kHz repetition rate is used, the gate scan has to be
synchronised with the laser, i.e. for each image a defined
number of laser shots has to be acquired.
Gate
Gate Scan
Delay
Gate pulse
Recorded
photons
Fluorescence
If a gated image intensifier is used in conjunction with a
scanning microscope [20,33], the gate scan must be
Fig. 18: Scanning a fluorescence decay
synchronised with the frame scan rate of the microscope.
function with the gate pulse
A serious drawback of the gated image intensifier is the low
counting efficiency. Due to the gating process, most of the photons collected from the sample are
gated off. The counting efficiency is about the ratio of the gate width to the fluorescence lifetime
and becomes more and more significant for shorter gate width. For a lifetime of 3.5 ns and a gate
width of 200 ps [20] the efficiency is only 5.7 %. The F value [39], i.e. the ratio of the ideal to the
actual SNR is 4.18.
The low efficiency must be compensated by a longer acqusition time with correspondingly more
photobleaching in the sample. Although the multiple beam technique [48, 49] can be used to
considerably reduce the acquisition time it does not really improve the detection efficiency.
16
The counting efficiency can be improved by using a very wide gate and measuring the fluorescence
with only two gate delays. Single exponential decay constants can be derived from the intensities in
the two time windows analog to multi-gate photon counting [50]. Since the measurements for the
two gates have to be done one after another the counting efficiency of such a measurement is close
to 0.5, i.e. by a factor of two less than for multi-gate photon counting.
In scanning microscope applications the gated image intensifier narrowly beats the multiple gate
photon counting method for time resolution. The Pico Star System of La Vison [51] has a minimum
gate width of 200 ps compared to 500 ps of the gated photon counting method [57,58]. However,
the multiple gate photon counting method has a near-ideal counting efficiency, resulting in a
correspondingly higher signal to noise ratio (SNR) for a given sample exposure.
In a scanning microscope the image intensifier cannot compete with time-correlated photon
counting (TCSPC) imaging [67,68,69] in terms of time resolution and counting efficiency. TCSPC
currently achieves 25 ps resolution and an F value and a counting efficiency close to one. TCSPC is
even able to record in several wavelength intervals simultaneously [68] - a feature that pushes the
efficiency of TCSPC beyond the theoretical limit of any single channel detection technique.
A gated image intensifier can be used for wide field illumination, for scanning with one-photon
excitation and for scanning with two-photon excitation. However, it cannot be used in a confocal
setup, and deep tissue two-photon images are blurred by scattering.
The realm of gated image intensifiers is clearly the wide field microscope [23]. If wide field
illumination has to be used for whatever reason, there is currently no replacement for the image
intensifiers.
Modulation Techniques
Modulation techniques use modulated light to excite
the fluorescence. Referred to the excitation light, the
fluorescence light has a phase shift and a reduced
modulation degree (fig. 19). Both depend on the
fluorescence lifetime and on the modulation
frequency:
Excitation
Fluorescence
Sample
Excitation
Fluorescence
tan ϕf = ω τf
Mf / Mex = 1 / sqrt ( 1 + ω2 τf 2 )
ω angular frequency of modulation,
Mex modulation of excitation, Mf modulation of
fluorescence, ϕf phase lag of fluorescence,
τf fluorescence lifetime
t
Fig. 19: Modulation technique
Both the phase and the modulation can be used to determine the fluorescence lifetime. However,
phase measurements are much more accurate than measurements of the modulation degree.
Therefore normally the phase is used for lifetime measurements. The optimum frequency depends
on the lifetime and is
ω = 1 / τf
or f = 1 / 2Π τf
Since fluorescence lifetimes are of the order of nanoseconds or picoseconds a modulation frequency
between 50 MHz and several 100 MHz is used. To resolve the components of multi-exponential
decay functions phase measurements at different frequencies are necessary.
17
If a frequency between 100 MHz and 1 GHz is used the modulation method gives a time resolution
in the ps range. For one-photon excitation the light source can be a modulated laser diode or a CW
laser with an external modulator. For two-photon excitation a Ti:Sa laser is used and the phase is
measured at the fundamental pulse frequency of the laser and at its harmonics.
Taking into consideration the high frequency, the wide amplitude range of the fluorescence signal
and the low signal-to noise ratio the phase measurement is anything but simple. Consequently, there
are several modifications of the method depending on different excitation sources, detectors and
phase measurement methods.
Single Channel Modulation Techniques
The general principle of the single channel modulation
technique is shown in fig. 20.
The light from the laser is modulated at a frequency in the
range of 30 MHz to 1 GHz. The fluorescence light from the
sample is detected by a PMT or a photodiode. The AC
component of the detector signal is amplified and fed into
two mixers that mix the signal with the modulation
frequency at 0° and 90° phase shift. Mixing means actually a
multiplication of the signals, therefore the outputs of the
mixers deliver a DC component that represents the 0° and
90° components of the amplified detector signal. After low
pass filtering the phase can be calculated from the mixer
signals. The principle is analog to a dual-phase lock-in
amplifier and often called lock-in detection [52,53].
Continuous Laser
Sample
Modulator
Detector
Generator
Amplifier
Mixer 0°
90° Mixer
Phase
Calculation
Fig. 20: Modulation technique
The modulation technique can be used to record the fluorescence of several chromophores
simultaneously in several parallel mixer systems. Several lasers which are modulated at different
frequencies provide different excitation wavelengths. The emission of the sample is split into
several wavelength ranges and detected by separate detectors. The detector signals are mixed with
the modulation frequencies of the individual lasers in several parallel groups of mixers [52, 53].
Unfortunately high frequency mixers do not work well for
DC output signals. Therefore, often a heterodyne principle,
similar to that in a radio, is used (fig. 21). The detector
signal and the modulation frequency are mixed with an
oscillator frequency slightly different from the modulation
frequency. The result are two signals at the difference
frequency. The phase shift between the two signals is the
same as between the detector signal and the modulation
signal. Depending on the frequency difference, the output
signals of the mixers are usually in the kHz range. Therefore
they can directly be digitised. The results of the AD
conversion are filtered and used for phase measurement. The
advantage of direct digitising is that effective digital filtering
algorithms can be applied and the phase can be determined
via fast Fourier transform [54, 55].
Laser
Sample
Modulator
Detector
Generator
f mod
Oscillator
Mixer
Mixer
f osc
digitiser
f mod - f osc
A/D
A/D
digitiser
digital filter
phase calculation
Fig. 21: Heterodyne principle
18
A phase fluorometer of this type can be built up by using a commercially available network analyser
and a laser diode. The setup is very simple - unless you have to build the network analyser. The
principle is shown in fig. 22. The network analyser runs a
frequency sweep over a selectable frequency interval. The
Network Analyser
output signal of the network analyser drives a laser diode
RF out
RF in
that is used to excite the sample. The detected fluorescence
signal is fed back into the signal analyser. The measurement
delivers the phase and the amplitude of the signal as a
Diode Bias
function of the frequency. The network analyser is even able
Amplifier
to correct the results by using reference data recorded with a
Sample
Laser
Amplifier
scattering solution in place of the sample.
Detector
Diode
The mixers used in fig. 20 and 21 can be replaced with a
modulated detector (fig. 23). A setup of this type has been
used for lifetime imaging in a two-photon laser scanning
microscope [54]. A commercially available frequency
synthesiser generates the modulation frequency, the
oscillator frequency and the difference of both with a high
frequency and phase stability. Mixing is accomplished by
modulating the gain in a photomultiplier. The mixed signal
at the output of the PMT has a frequency of 25 kHz and is
directly digitised. Further filtering and phase calculation is
done digitally.
The benefit of the detector modulation is that the modulation
frequency is not limited by the bandwidth of the gain system
of the PMT. Therefore relatively high modulation
frequencies can be used. The drawback is that the detector is
a very poor mixer. This results in a poor efficiency of the
detector gain modulation technique.
Fig. 22: Phase fluorometer with a network
analyser
Laser
Sample
Modulator
Frequency
Synthesiser
Gain
modulated
Detector
fmod
f osc
f mod - f osc
digitiser
A/D
f mod - f osc
A/D
digitiser
digital filter
phase calculation
If a Ti:Sa laser is used in a modulation system, e.g. for twophoton excitation, the 80 MHz fundamental repetition
frequency of the laser and its harmonics can be used as Fig. 23: Mixing by gain-modulating the detector
modulation frequency. Due to the short pulse width the
spectrum of the laser modulation is actually a frequency comb that extends up to THz frequencies.
By proper tuning of the frequency synthesiser, any of the harmonics of the fundamental repetition
frequency within the detector bandwidth can be used for the phase measurement.
A heterodyne system with a Ti:Sa laser, two-photon excitation and modulated PMT delivered a
lifetime accuracy of ± 300 ps for lifetimes between 2.4 and 4.2 ns [54]. This relatively poor
accuracy is probably due to the short acquisition time of only 160 µs per pixel or 10 seconds for a
256 x 256 pixel image. Although the photon detection rate was estimated to be more than 108
photons/s - which is probably an overestimation - the accuracy is of the same order as for TCSPC
imaging with 106 photons per second and the same acquisition time and pixel number.
The efficiency of the single channel modulation technique in terms of photon counts strongly
depends on the operating conditions. Results of Monte-Carlo simulations of the efficiency for
single-exponential decay are given in [39]. The simulations yield ‘F values’, i.e. the ratio of the
ideal SNR to the SNR achieved by the investigated method.
For a setup with separate detector and mixer, sine-wave-modulated excitation, sine-wave mixing
and 100% modulation of the excitation F was found to be 3.7. If the modulation is less than 100% a
19
fraction of unmodulated light is detected which contributes to the noise, but not to the phase
measurement signal. The SNR drops dramatically for less than 75% modulation [39]. Interestingly,
in the same setup 100 % square-wave modulation of the excitation gives an F = 1.2, and Dirac pulse
excitation even F = 1.1. This is a strong argument to use pulsed lasers, i.e. Ti:Sa or diode lasers
rather than modulating CW lasers.
For a gain-modulated detector the efficiency is much worse than for a detector and a separate
mixer. The reason is that a gain modulated detector is a very bad mixer. True mixing means to
multiply the detected signal with a sine wave, i.e. to invert the polarity of the signal for the negative
half-period of the mixing signal. Of course, gain modulation at best means to change the gain
between 100% and 0. Therefore a large fraction of the input signal remains unused, but contributes
with its shot noise to the noise of the result. The F values for the modulated detector are around 4.
That means, the technique needs 16 times more photons than techniques with a near-ideal F.
For extremely low light levels the situation can be even worse. A high gain detector, such as a PMT,
delivers a train of extremely short current pulses rather than a continuous signal. The pulses are due
to the detection of the randomly arriving photons. If this pulse train is fed to a phase measurement
circuitry described above a phase signal is produced only for a short time after the detection of a
photon. In the times between the phase detector is unable to deliver a reasonable signal. What then
happens depends on the phase detection electronics or the equivalent software algorithm. A normal
phase detector in the absence of a signal delivers a huge noise. This noise is averaged with the
randomly appearing phase signals for the individual photons. The result is a dramatic drop of the
signal-to-noise ratio for low photon rates.
For application with a scanning microscope it must be taken into regard that phase measurements
are not compatible with the fast scanning speed of a state-of-the art laser scanning microscope. For
the lock-in method the low pass filters must have time to settle, and the heterodyne method requires
to record at least some periods of the difference frequency fmod - fosc. Therefore, minimum pixel
dwell times are in the order of 40 to 160 µs. It is not clear whether a long pixel dwell time increases
the photodamage in the sample. At least, the implementation of single channel modulation
techniques in commercial scanning microscopes requires changes in the scan control of the
microscope.
Another problem can arise for deep tissue imaging with a confocal or two-photon microscope. In
this case the presumption that the time-domain response of the sample is a sum of exponentials is
questionable. Scattered excitation light and broadening of the fluorescence decay functions by
scattering remain unnoticed unless a continuous sweep over wide frequency interval is performed.
The conclusion is that the single channel modulation technique is not the best lifetime technique for
scanning microscope applications. The true realm of the single channel modulation technique is in
the infrared where no detectors with single photon sensitivity are available.
Modulated Image Intensifiers
The modulated image intensifier technique uses the same vacuum device as the gated image
intensifier technique. However, the grid is driven by a sine wave or square wave signal rather than
by a short gating pulse [55,56]. Modulation can also be achieved without a grid by modulating the
voltage between the photocathode and the channel plate.
As for the gated image intensifier, the problem is the low conductivity of the photocathode. The
time constant formed by the gate-cathode capacitance and the cathode resistance limits the
modulation frequency and introduces lateral phase variation across the image. Furthermore, a
20
variable lateral field builds up in front of the photocathode that distorts the image and impairs the
image resolution.
Modulating the voltage across the channel plate [55] has also been attempted. The drawback of this
method is the heating of the channel plate due to dielectric losses and the low degree of modulation.
The microscope setup for a modulated image intensifier is shown in fig. 24. The setup is very
similar to that for the gated image intensifier.
Freqency
Synthes.
CW
Laser
AOM
Freqency
Synthes.
Pulsed
Laser
AOM
Driver
AOM
Driver
AOM
Variable
Phase Shift
Scan
Mirror
Telescope
Amplifier
Scanning Head
Scan
Mirror
Variable
Phase Shift
Amplifier
Scan
Lens
Grid
Grid
CCD
CCD
Image
of
Sample
-HV
Image
of
Sample
Image
Intensifier
+ MCP -
Microscope
Objective
Sample
-HV
Image
Intensifier
+ MCP -
Objective
Sample
Fig. 24: Modulated image intensifier used at a wide field microscope (left) and at a scanning microscope (right)
A frequency synthesiser generates a modulation frequency which is typically in the range of
50 MHz to 500 MHz. This frequency is used to modulate the excitation light and the image
intensifier. A variable phase shifter is used to change the phase relation between the modulation of
the light source and the modulation of the image intensifier. From three images acquired at different
phase shifts the phase of the fluorescence signal referred to the excitation can be calculated.
Instead of the phase shifter a heterodyne technique can be used. In this case the image intensifier is
modulated with a slightly different frequency than the laser. This causes the phase between the two
modulations to change continuously. If the difference frequency is a few Hz a sequence of images
can be obtained for each period of the difference frequency [55]. The heterodyne technique requires
a good frequency synthesiser that delivers the two modulation frequencies and the difference of both
with high frequency and phase stability.
A single phase measurement at a fixed frequency allows to determine the time constant of a single
exponential fluorescence decay function. To resolve the components of a multi-exponential decay
measurements at different frequencies are required. Ideally, the modulation frequencies should be of
the order of
fmod = 1 / 2Πτf with
fmod = Modulation Frequency, τf = Fluorescence decay time constant
At this frequency the change of the phase is at maximum for a given change of the decay time
constant. For decay components below a few 100 ps the optimum frequency is around 1 GHz.
Modulation frequencies of this order are almost impossible to achieve with standard image
21
intensifiers. The dielectric losses cause heat dissipation in the intensifier tube structures, and there
is a considerable phase shift between different parts of the image area.
A modulated image intensifier can be used for one-photon and multi-photon fluorescence imaging
with high repetition rate pulsed fs laser. The pulse train of the laser contains all harmonics of the
fundamental repetition frequency. In the frequency domain the laser pulse train is actually a
frequency comb that extends to THz range. Generally, by modulation the intensifier at the
harmonics of the laser frequency a multi-frequency measurement can be accomplished [55,56]. A
problem associated with this technique is that the gain modulation is nonlinear and therefore not
ideally sinusoidal. Therefore the measurement signal obtained at a particular harmonic of the laser
also contains a small fraction of higher harmonics.
A drawback of the image intensifier technique is that it cannot be used in a confocal scanning
microscope. Therefore depth resolution can only be achieved by two-photon excitation with a fs
laser [55]. Even then the spatial resolution is limited by scattering of the emission light in the
sample. The multiple beam technique described for the gated image intensifier in [48,49] is also
applicable for the modulated intensifier technique.
The efficiency of the modulated image intensifier in terms of photon exploitation has been
investigated theoretically by Monte-Carlo simulations [39]. The F value, i.e. the SNR of an ideal
measurement compared to the actually obtained SNR is 10 for sine-wave modulated exciatation and
4.3 for dirac modulation. That means the image intensifier needs 18.5 to 100 times more photons
than techniques with perfect exploitation of the photons, e.g. time-correlated single photon
counting.
The reason for the high F values is that the gain can only be modulated between 0 and 1. Ideal
mixing requires gain modulation between -1 and +1. Thus the modulation depth is only 50%. The F
values are even worse if the modulation of the gain in the intensifier tube is less than 50% - which is
certainly the case if a device without a grid is used. Moreover, the modulation is an efficiency
modulation rather than a gain modulation - the photoelectrons from the photocathode are either
rejected or transmitted into the channel plate. Therefore the efficiency is worse than for a detector
with a subsequent mixer.
The time resolution of the modulated image intensifier technique was reported to be down to a few
100 ps [55,56]. It is likely that much shorter lifetimes can be resolved if a sufficiently long
acquisition time is used.
Generally the resolution of frequency domain techniques cannot directly be compared to the
resolution of time domain methods. The decay times are calculated from the phase (i.e. from a part
of the Fourier spectrum) of the fluorescence signal. This calculation includes reference
measurements for a reference sample or zero decay time for each frequency. Correctly, the
resolution of frequency domain methods should be compared with the resolution of time-domain
methods with deconvolution of the decay data. Deconvolution of TCSPC data obtained with an
MCP PMT resolves single exponential decay times down to a few ps.
Due to the low efficiency the modulated image intensifier technique it cannot be recommend as a
lifetime method for laser scanning microscopes. However, the technique is applicable to wide field
illumination. If wide field illumination has to be used for whatever reason there is currently no
replacement for the gated and modulated image intensifier techniques.
22
Gated Photon Counters
As all photon counting techniques, gated photon
counting uses a fast, high gain detector that is usually a
PMT or a single photon avalanche photodiode. The
single photon pulses from the detector are counted
within one or more time intervals. Due to the moderate
time resolution of the gating technique there are no
special requirements to the transit time spread of the
detector. However, the transit time distribution should
not have a long tail, or bumps and prepulses or
afterpulses and should be stable over the detection area.
The principle of a gated photon counter with a single
gate is shown in fig. 25.
Sample
Pulsed Laser
Detector
Reference
Variable Delay
Gate
Counter
Gated Photon Counting (single gate)
Fig. 25: Gated Photon counter with single gate
To record the shape of a fluorescence decay function the gate pulse can be scanned over the time
interval of interest by a variable delay generator. This method yields a decay curve with as many
points as the delay unit has delay steps. However, the majority of the photons is discarded by the
gating. Thus, the counting efficiency is low and the F value is of the same order as for the gated
image intensifier. Although ready-to use gated photon counters are available nobody seriously
considers to use this technique for lifetime imaging.
A counting efficiency close to one can theoretically be
achieved by a multiple gate architecture, fig. 26. The system
contains several parallel counters with individual gates. The
gates are controlled via separate gate delays and by separate
gate pulse generators. If the measured decay curve is
completely covered by subsequent gate pulses all detected
photons are counted. Moreover, the parallel counter
structure can be operated at extremely high count rates. Even
for count rates of several MHz there is almost no counting
loss due to the short dead time of the counters. The practical
implementation of the method is described in [57, 58].
Another benefit is that no special vacuum devices as for the
image intensifier technique are required. The method
immediately benefits from new detector developments, e.g.
PMTs with high efficiency cathode materials [59,60].
Sample
Pulsed Laser
Detector
Reference
Variable Delay
Variable Delay
Variable Delay
Variable Delay
Gate
Counter
Gate
Counter
Gate
Counter
Gate
Counter
Gated Photon Counting (multiple gate)
Fig. 26: Gated photon counting with several gates
Limitations of the method result from the relatively long gate duration and from the limited number
of gate and counter channels. From the of signal-theory-point of view, the function of photon
density versus time is heavily ‘undersampled’. In other words, the maximum frequency detectable in
the signal spectrum is sampled less than 2 times per signal period. Undersampled signals cannot be
reconstructed from the sample values without presumptions about the signal shape.
23
Fortunately, decay curves are either single exponentials
or a sum of a few exponentials. Ideally, the lifetime of a
single exponential decay can be calculated from the
photons collected in only two time windows, see fig. 27.
The optimum gate width was determined to be T = 2.5 τf
[38]. It has been shown that the components of multiexponential decay functions can be determined if the
number of gates is increased [61].
T
T
e -t/tau
Ia
However, practically the photons distribution is the
convolution of the fluorescence decay function with the
instrument response function (IRF), i.e. the excitation
pulse shape, the detector TTS, the pulse dispersion in the
optical system, and the effective shape of the gating
pulse. The resulting fluorescence signal cannot be
considered to be a sum of exponentials (fig. 28). To
obtain the fluorescence lifetime from the intensity in two
time windows these must be placed in a part of the
fluorescence signal where the IRF has dropped below the
noise level, i.e. in a region where the fluorescence signal
is exponential. This, however, reduces the counting
efficiency by discarding the most intense portion of the
signal. Practically it is probably better to tolerate some
error in the lifetime calculation and to calibrate the
system by using known fluorescence lifetimes. However,
for double- or multi-exponential decay functions a
calibration appears difficult if not impossible.
tau =
T
ln ( Ia / Ib )
Ib
Fig. 27: Calculation of fluorescence lifetime from
intensities in two time intervals, ideal IRF
Fluorescence Signal
Instrument
Response
Function
(IRF)
T
~~ e -t/tau
Ia
T
tau =
T
ln ( Ia / Ib )
Ib
Fig. 28: Calculation of fluorescence lifetime from
intensities in two time intervals, real IRF
The shortest gate width that has practically been used is of the order of 500 ps. Although lifetimes
down to 70 ps have been measured with reasonable accuracy it is difficult to measure lifetimes
below 200 ps. Double exponential decay functions of FRET systems with a fast component of 100
to 300 ps probably cannot be resolved.
Practical devices based on multi-gate photon counting work with a direct readout of the counters
after the acquisition of each pixel of the image [57]. This sets a lower limit to the pixel dwell time
in the 10 us range. Consequently, the maximum scan speed of a laser scanning microscope cannot
be exploited. It is not clear whether this increases the photobleaching of the sample via
accumulation of triplet states or heat concentration in the scanned spot. The problem - if it exists could easily be overcome by implementing a memory that accumulates the results in memory
locations according to the x,y location of the laser spot in the scanning area.
For samples containing different chromophores the counting efficiency and the amount of
information in the data can be improved by recording the fluorescence in several wavelength
channels simultaneously. Steady-state multi-wavelength imaging has been successfully used to
separate different chromophores in stead-state images [13]. The gated photon counting technique
basically can be used with several detectors if the number of gate and counter channels is increased.
Several gates can be driven by the same delay and gate pulse generator so that the increase of
system complexity is less than proportional. Although a multi-wavelength gated SPC system has not
become known yet it could improve the selectivity for different chromophores considerably.
Theoretical investigation of the efficiency [39] delivers an F value of 1.5 for two gates of a width of
2.5 τ. For 8 gates with unequal width F = 1.1 can be achieved. However, practically τ is unknown
and varies throughout the image so that the gate width is usually not optimal. Furthermore, the first
24
part of the signal can probably not be used because it is convoluted with the system response.
Assuming a loss of a factor of two in the number of counted photons the +method ends up at F = 1.5
to 2.1 which is much better than for the image intensifiers and noticeably better than sine-wave
modulation techniques.
Acquisition times for biological samples stained with highly fluorescent dyes are in the range 10 to
100 seconds [58,11]. Generally the acquisition times for comparable lifetime accuracy should be
expected in the same range as for TCSPC imaging (see below) unless count rates exceeding 5 MHz
are available. As for TCSPC imaging, the acquisition time can be reduced by decreasing the number
of pixels of the scan.
Multi-gate photon counting has its merits in applications that require to record lifetimes in the range
of a few ns at high count rate and with short acquisition times. Typical applications are lifetime
imaging with marker dyes for Ca++, Na+, O-- or other parameters within cells and tissue [11b].
Probing of DNA and RNA by the lifetime change of SYTO13 is described in [11, 11a].
25
Time-Correlated Single Photon Counting (TCSPC)
Time-Correlated Single Photon Counting is based on the detection of single photons of a periodical
light signal, the measurement of the detection times of the individual photons and the reconstruction
of the waveform from the individual time measurements [62, 63]. The principle is shown in fig. 29.
The fluorescence is excited by a pulsed laser with a
repetition rate of typically 80 MHz. The fluorescence
photons are detected by a fast PMT, MCP or a single
photon avalanche photodiode. For precision
measurements these detectors allow count rates of the
order of a few MHz, i.e. much less than the laser
repetition rate. The TCSPC method makes use of the
fact that the detection of several photons in one laser
period is small enough to be neglected. When a photon
is detected the time of the corresponding detector pulse
in the laser pulse sequence is measured. The measured
times are used to address a histogram memory in which
the photons are accumulated. After acquiring a large
number of photons the photon density versus time
builds up in the memory.
The method may look circumstantial at first glance, but
it has some striking benefits - a near-ideal counting
efficiency and an ultra-high time resolution.
Optical Waveform
Detector Signal:
Time
Period 1
Period 2
Period 3
Period 4
Period 5
Period 6
Period 7
Period 8
Period 9
Period 10
Period N
Result after
collecting many
Photons
memory location
Fig. 29: Time-correlated single photon counting
As long as the count rate is not so high that the conversion time of the time measurement and
histogramming procedure becomes noticeable there is no loss of photons. The F value describing
the ratio of the ideal SNR to the actually obtained SNR is < 1.1 [38,39].
The resolution is limited by the transit time spread in the detector, by the timing accuracy of the
discriminator that receives the detector pulses and by the accuracy of the time measurement. With
an MCP-PMT the width of the instrument response function is of the order of 25 to 30 ps. The
width of the time channels of the histogram can be made less than 1 ps [63]. Therefore, the
instrument response function is sufficiently sampled to determine lifetimes down to a few ps.
The TCSPC technique is often believed to be extremely slow and unable to reach short acquisition
times. This ill reputation comes from older NIM systems used in conjunction with low repetition
rate light sources which are indeed not applicable for time-resolved imaging.
However, high count rate techniques exist [64] since the late 80s. An early - probably the first implementation of TCSPC in a microscope is described in [65].
State-of-the-art TCSPC devices use a new conversion principle and achieve count rates of several
106 photons per seconds [63]. A system with four parallel TCSPC channels and 20 MHz useful
count rate is described in [66]. The system was used in a slow-scan setup for optical tomography
and can be combined with a multi-detector technique [63,66]. Although a combination with a
scanning microscope has not become known yet the system is basically able to record images in up
to 32 detector channels with slow scan rates at count rates comparable to multi-gate photon
counting [57].
26
TCSPC Imaging
TCSPC Imaging is an advanced TCSPC technique used in conjunction with scanning microscopes.
The method builds up a three-dimensional histogram of the photon number over the time and the
coordinates of the scan area. The principle is shown in figure 30. The recording electronics consists
of a time measurement channel, a scanning interface, and a large histogram memory.
The time measurement channel contains the usual TCSPC building blocks. Two constant fraction
discriminators, CFD, receive the single photon pulses from the detector and the reference pulses
from the laser. The time-to-amplitude converter, TAC, measures the time from the detection of a
photon to the next laser pulse. This ‘reversed start-stop‘ principle is the key to the processing of
high laser repetition rates. The analog-digital converter, ADC, converts the TAC output voltage into
an address for the memory.
Detector
Timing
Start
Time
Measurement
CFD
TAC
ADC
Time within decay curve
t
Stop
CFD
from Laser
Frame Sync
Counter Y
Line Sync
y
Histogram
Memory
Scanning
Interface
Pixel Clock
from
Microscope
Counter X
x
Location within scanning area
Fig. 30: TCSPC Imaging
The scanning interface is a system of counters. It receives the scan control pulses from the
microscope and determines the current position of the laser beam in the scanning area.
When a photon is detected the device determines the time, t, within the fluorescence decay curve
and the location of the laser spot within the scanning area, x, y. These values are used to address the
histogram memory. Consequently, in the memory the photon distribution versus t, x, and y builds
up. The result can be interpreted as a stack of images for different times after the excitation pulse or
as an array of pixels containing a complete fluorescence decay function each.
The implementation of TCSPC imaging into Zeiss microscopes is described in [67-69]. The
described instrument consist of a Zeiss LSM-510 NLO Laser scanning microscope, a femtosecond
Ti:Sa laser, a fast PMT, and a Becker & Hickl SPC-730 TCSPC imaging module [63]. The setup is
shown in fig. 31.
27
Zeiss
PMH-100
R3809U
b&h
SPC-730 TCSPC Imaging Module
200ps
<50ps
Fibre Output
Detector
start
LSM-510
Scanning
Head
stop
Ti:Sa Laser
80 MHz 200 fs
Pixel Clock, Line Clock, Frame Clock
User I/O
Axiovert
Microscope
Control
Box
Fig. 31: Setup of the lifetime microscope
The setup does not require any changes in the microscope or in the associated control software. The
fibre output option of the LSM-510 scanning head is used to feed the detected light into the
photomultiplier. Depending on the required resolution either a PMH-100 detector head of Becker &
Hickl or an R3809U MCP-PMT of Hamamatsu is used. For each photon, the detector delivers a
start pulse to the SPC-730 TCSPC imaging module. The stop pulse for the time measurement is
obtained from the monitor diode of the laser. The recording in the SPC-730 TCSPC module is
synchronised with the scanning via the Pixel Clock, Line Clock and Frame Clock signals from the
control box of the microscope.
The system response for a Hamamatsu R3809U
MCP is shown in fig. 32. The instrument response
width is 30 ps for an MCP supply voltage of
-3000 V.
Usually, deconvolution of the measured
fluorescence decay from the system response
delivers decay at least times 10 time shorter than
the system response, i.e. less than 3 ps.
Deconvolution requires to record the instrument
response function (IRF), which is simple in a onephoton microscope. However, in a two-photon
microscope the IRF is difficult to record because
the excitation wavelength is blocked by filters, or
detectors are used which are insensitive at the
excitation wavelength. Furthermore, the sample is
excited by a two-photon process while the detector
sees the response via one-photon photoemission.
For standard applications a ‘best guess’ system
response can be calculated from the rising edge of
the fluorescence decay functions themselves. The
shortest lifetime that can reasonably be determined
this way is in the range of the IRF width, i.e. 20 to
30 ps. For precision measurements an ultra-short
fluorescence or a nonlinear crystal can be used to
record the IRF.
Lifetime images obtained by TCSPC imaging are
28
fwhm=30ps
Fig 32: Instrument response for the R3809U-50 MCP
0.3
Lifetime, ns
1.2
Fig. 33: Lifetime image of a sample stained with three
different dyes (Leica TCS SP2, SPC-730)
presented in [67-75]. An example is shown in fig. 33. The sample is stained with three different
dyes which are clearly distinguished in the lifetime image.
An application to FRET imaging (after [68]) is shown in figure 34.
1
1
2
2
1ns
2ns
Lifetime, single exponential
approximation
1
2
Ratio of intensity
coefficients, fast/slow
Fig. 34: HEK cell containing CFP and YFP in the α and β subunits of the sodium channels.
Left: Lifetime image of donor, CFP, centre: Decay curves of selected pixels, right: FRET image showing ratio of
intensity coefficients of quenched and unquenched fluorescence components
Data from [68], Zeiss LSM-510, Becker & Hickl SPC-730
The sample is an HEK cell containing CFP and YFP in the α and β subunits of the sodium
channels. Since the emission spectrum of CFP overlaps the absorption spectrum of YFP FRET is
expected in regions where the subunits are close together. Fig. 34, left, shows a lifetime image of
the donor, CFP, obtained from a single exponential fit through the decay functions in the individual
pixels.
Fig. 34, centre, shows decay curves of two selected pixels from the yellow (short lifetime) and blue
(long lifetime) area of the lifetime image. Double exponential decay analysis yields two decay
components of about 380 ps and 1.8 to 2 ns throughout the image. However, the intensity
coefficients, a, differ considerably in different parts of the cell. Assuming that the lifetime
components belong to the quenched and unquenched donor molecules the ratio of the intensity
coefficients, afast / aslow, should represent the ratio of the number of quenched and unquenched CFP
molecules. Therefore the ratio of the coefficients can be used as a measure for the FRET intensity.
Fig. 34, right shows a FRET image displaying the number of photons as brightness and afast/aslow as
colour.
Multi Wavelength TCSPC Imaging
For samples containing different chromophores the counting efficiency and the amount of
information in the data can be increased by recording the fluorescence in several wavelength
channels simultaneously. Multi-wavelength imaging has been successfully used to separate different
chromophores in stead-state images [13]. TCSPC detection of single molecules in four wavelength
channels considerably improves the selectivity for different molecules [17].
TCSPC imaging can be combined with a multidetector technique [63,66,76,77] and be used to
record the photon density over time, wavelength and the coordinates of the scanning area [72]. The
principle of the multi-detector technique is shown in fig. 35.
29
R5900L16
Discriminators Encoder
Detectors
3 bit
Channel
Number
1
Don’t
Discriminators Encoder
1
1
2
2
3
3
4
4
2 bit
Channel
Number
Don’t
Count’
Count
16
Photon Pulse
Photon Pulse
R3809U
Summing Amplifier
Inverting Amplifier
Fig. 35: Multi-detector technique for TCSPC. Left: Multichannel PMT, Right: Individual PMTs
The technique works both with a multichannel PMT [72a,77] and with several individual PMTs or
MCPs [72b,66,76]. It makes use of the fact that the detection of several photons in different detector
channels in one laser period is unlikely. Therefore, the single photon pulses from all detector
channels can be combined into a common photon pulse line and send through the normal time
measurement procedure of the TCSPC module. The output of each PMT channel is connected to a
discriminator. If the channel detects a photon the corresponding discriminator responds and sends a
pulse to the subsequent encoding logic. The encoder delivers the number of the PMT channel that
detected the photon. The channel number is used as an additional dimension in the multidimensional histogramming process of the TCSPC imaging technique.
The recording electronics for multi-wavelength TCSPC imaging consists of a time measurement
channel, a scanning interface, a detector channel register, and a histogram memory (fig. 36).
from
Polychromator
Channel
Channel register
Timing
Start
CFD
n Channel / Wavelength
Detector
Time
Measurement
TAC
ADC
t
Time within decay curve
Stop
CFD
from Laser
Frame Sync
Counter Y
Line Sync
Pixel Clock
y
Scanning
Interface
Counter X
x
Histogram
Memory
Detector
channel 1
Histogram
Memory
Detector
channel ...
Histogram
Memory
Detector
channel ...
Histogram
Memory
Detector
channel 16
Location within scanning area
from
Microscope
Fig. 36: Multi-wavelength TCSPC lifetime imaging
For each photon, the time measurement channel determines the detection time (t) referred to the
next laser pulse. The scanning interface determines the current location (x and y) of the laser spot in
the scanning area. Synchronously with the detection of a photon, the detector channel number (n)
for the current photon is read into the detector channel register. If a polychromator is used in front of
the detector, n represents the wavelength of the detected photon.
The obtained values for t, x, y and n are then used to address the histogram memory. Thus, in the
memory the distribution of the photons over time, wavelength, and the image coordinates builds up.
The result is a four-dimensional data array that can be interpreted as a set of image stacks for
30
different wavelengths. Each stack contains a number of images for subsequent times after the
excitation pulse.
Fig. 37 shows a HEK 293 cell expressing a hybrid protein in which the cyan (CFP) and yellow
(YFP) shifted mutants of the green fluorescent protein are linked together by a short amino acid
chain. The setup used for this measurement [72a] consisted of a laser scanning microscope
(LSM-510, Zeiss), a polychromator (250is, Chromex), and a 16 channel TCSPC detector head [77]
connected to an SPC-730 TCSPC imaging module [63]. The setup recorded the wavelength range
from 410 nm to 635 nm in 16 wavelength channels, covering the emission bands of CFP and YFP.
The image below was obtained by summing the photons from all time channels of the CFP
fluorescence.
CFP
Fig. 37: HEK 293 cell expressing a CFP-YFP hybrid protein
Left: Intensity image of CFP. Right: Fluorescence decay curves of CFP and YFP in a selected region (square).
Fluorescence decay analysis in a selected region (small square) reveals a double-exponential decay
both for CFP and YFP. The intensity coefficient of the fast component is positive for CFP and
negative for YFP, indicating that energy is transferred from CFP to YFP. In Fig. 38 the intensity is
represented by the brightness, the ratio of the coefficients by the colour of a pixel. The results are
shown for the CFP and the YFP fluorescence.
CFP
0.40
YFP
0.80
-0.65
-0.15
Fig. 38: FRET images for CFP (left) and YFP (right) of a HEK cell expressing a hybrid protein in which CFP and YFP are linked
together by a short peptide. The brightness represents the intensity, the colour the ratio of the amplitudes of the fast and slow decay
components. From [72a]
31
Features of the TCSPC imaging techniques
It should be pointed out that neither single wavelength nor multi-wavelength TCSPC imaging use
any time gating or wavelength scanning. Therefore, both techniques yield a near-perfect counting
efficiency and a maximum signal to noise ratio for a given acquisition time. Due to the short dead
time of the TCSPC imaging electronics (125 to 180 ns ) [63] there is almost no loss of photons for
count rates up to a few 105/s as they are typical for cell imaging.
The time resolution of the TCSPC techniques is given by the transit time spread in the detector. A
system response of only 25 ps FWHM is achieved with MCP PMTs. In conjunction with a
minimum time channel width of the TCSPC modules of less than 820 fs [63] lifetimes down to a
few ps can be measured.
No special vacuum devices as for the image intensifier technique are required. The method
immediately benefits from new detector developments, e.g. ultra-fast MCP PMTs or new high
efficiency cathode materials.
Implementing TCSPC imaging into a laser scanning microscope requires no more changes than
attaching a fast detector to the microscope. In most microscopes this is possible either via a fibre
output or via the non-descanned port.
TCSPC imaging works at the full scan rate of the laser microscope, i.e. with pixel dwell times of
order of 1 µs. Due to the synchronisation via the scan control pulses the zoom and image rotation
functions of the microscope can be used also for lifetime imaging.
Currently available TCSPC imaging devices have a limited memory space. Therefore, a tradeoff
between the number of pixels, the number of time channels and the number of detector channels has
to be made. However, with the introduction of the SPC-830 module of bh memory space is not
longer a severe constraint. Some combinations of image size, and number of time and wavelength
channels for the SPC-830 are given in the table below [63].
Detector Channels
Resolution x,y
Time Channels
Min. Time Channel Width (ps)
1
1
1
1
1
2048 x 2048
1024 x 1024
512 x 512
256 x 256
128 x 128
4
16
64
256
1024
50
12.5
3.125
0.8
0.8
4
4
4
4
1024 x 1024
512 x 512
256 x 256
128 x 128
4
16
64
256
50
12.5
3.125
0.8
16
16
16
16
512 x 512
256 x 256
128 x 128
64 x 64
4
16
64
256
50
12.5
3.125
0.8
Parallel operation of two modules connected to the same detector can be used to simultaneously
obtain images with high pixels resolution and moderate time resolution and images with moderate
pixel resolution and high time resolution. Although no such system has become known yet the
implementation with current TCSPC modules is straightforward.
The maximum photon count rate of the TCSPC technique is limited by the time required to measure
the time of a detected photon and to store it in the histogram memory. This ‘dead time’ is of the
order of 125 to 180 ns for state-of-the art devices. Becker & Hickl define a ‘maximum useful count
rate’ which means the recorded photon rate at which 50% of the photons are lost in the dead
time [63]. Maximum useful count rates are in the range of 3 to 4 MHz. The counting efficiency and
32
the ‘figure of merit’ [39] as a function of the detector count rate for different techniques are shown
in fig. 38. The values for the image intensifiers and single channel nmodulation techniques were
taken from [39], the value for multi-gate photon counting from [81].
1
TCSPC, 4 Channels
Efficiency
4
F
Figure
of Merit
3.4
TCSPC, 1 Channel
0.8
Modulation
Image Intensifiers
Modulation
3.2
3
0.6
2.8
2.6
Multi Gate SPC
0.4
2.4
2.2
2
1.8
0.2
1.6
1.4
Modulation
Modulation
1.2
Image Intensifiers
0
10 kHz
100 kHz
TCSPC, 4 Channels
Multi Gate SPC
TCSPC, 1 Channel
1
1 MHz
10 MHz
Detector Count Rate
10 kHz
100 kHz
10 MHz
1 MHz
Detector Count Rate
Fig. 38: Counting efficiency and figure of merit F of TCSPC imaging compared to other techniques
F is the ratio of theoretical SNR to obtained SNR, F=1 means ideal SNR
Surprisingly, TCSPC imaging beats the other methods even for detector count rates of the order of 5
to 10 MHz.
It is unlikely that such high count rates can be obtained from living cells. Anyway, a count rate
above 5 MHz imposes overload problems to most detectors. Although traditional PMTs still work at
10 MHz the timing performance may not longer meet the high standard of the TCSPC method.
MCP PMTs, i.e. the fastest detectors for TCSPC, are clearly overloaded above 1 MHz. If a system is
to be operated at extremely high count rates the solution is the multidetector technique. If the light
is split into several detection channels the load for the individual detectors is reduced.
Another solution is to operate several TCSPC modules with individual detectors in parallel.
Although no such application has become known yet the implementation is straightforward. The
Becker & Hickl TCSPCs are designed to work in packages of up to four devices [63].
The acquisition time for TCSPC lifetime measurements can vary in a wide range. In vivo lifetime
measurements at the human ocular fundus in conjunction with an ophthalmologic scanner delivered
single exponential lifetimes for an array of 128x128 pixels within a few seconds [75, 75b]. On the
other hand, for the double exponential decay data of FRET measurements acquisition times from 5
to 30 minutes were used [68].
In practice the acquisition time depends on the required lifetime accuracy, on the number of
lifetime components to be resolved, on the number of pixels and wavelength channels and on the
count rate that can be obtained from the sample without photobleaching or photodamage. Fig. 39
shows the acquisition time as a function of the product of the number of pixels and the number of
wavelength channels. The left diagram is for a high count rate of 106 /s. Count rates of this order can
be obtained by one-photon excitation from cells containing high chromophore concentrations. For
two-photon excitation a count rate of 106 /s requires extremely stable samples stained with very
efficient chromophores.
The right diagram is for a very low count rate of 104 /s. Count rates of this order are typical for
autofluorescence of cells and tissue and for samples with low photostability. An HEK cell
33
containing CFP and YFP measured by two-photon excitation with 150 fs pulses at 820 nm [68]
delivered only 104 photons/s. Nevertheless, it photobleached to 50% intensity within 10 minutes.
10000s
Acquisition
Time
1000s
10000s
Acquisition
Time
1000s
Count Rate
10 6 /s
100s
100,000
100s
100,000
10,000
Count Rate
10 4 /s
1000
10,000
10s
10s
100
1000
1s
1s
Photons/
Pixel
100
100ms
100ms
Photons/
Pixel
10ms
1ms
10ms
1ms
0.1ms
0.1ms
1
10
100
1000
10,000
100,000
Pixels x Wavelength Channels
1
10
100
1000
10,000
100,000
Pixels x Wavelength Channels
Fig. 39: Acquisition times for a count rate of 106 /s (left) and 104 /s (right) for different numbers of photons per pixel.
Photons per pixel range from 100 for rough single exponential decay mapping to 105 for precision double exponential
decay analysis.
Fig. 39 shows that relatively long acquisition times must be expected, particularly for large numbers
of pixels and precision measurements of samples at a low count rate. However, it should be pointed
out that long acquisition times are not a special feature of TCSPC imaging - they simply result from
the fact that much more photons per pixel are required to determine the lifetime than to record the
intensity.
Shorter acquisition times can be achieved if the scanning area is confined to a few pixels or only
one pixel is measured. For single pixels the acquisition time can be in the ms or sub-ms range. That
makes it possible to investigate diffusion processes by monitoring the lifetime change in a single
pixel. Even fast FRET measurements appear feasible. FRET requires to resolve the components of a
double exponential decay function. This requires about 10,000 photons which can be obtained in
10 ms.
Even if the emission intensity of biological samples is limited there is probably still some potential
to increase the count rate by improving the detection efficiency. The transmission band of filters
used to select the emission from a particular chromophore are often narrower than the emission
band. Optimising the filters can easily yield a factor of two in sensitivity. Moreover, in a normal
microscope the photons are collected from one side of the sample only. The 4Pi technique [37]
collects the fluorescence from both sides and therefore gives a factor of two improvement in
collection efficiency.
In cases where the full diffraction-limited resolution of the objective is not required the count rate
can probably be increased by increasing the excited sample volume. Photobleaching is reported to
be highly nonlinear, therefore the larger volume allows to use more excitation power and
consequently to get more photons [43,45]. By introducing defined distortions into the wavefront of
the excitation path or by under-illuminating the aperture of the objective the excited volume can be
increased without impairing the photon collection efficiency.
Recently new photomultiplier cathodes with a 3-fold improved sensitivity compared to traditional
bialkali and multialkali cathodes became available. Fast PMTs and MCPs with the new cathodes
have an IRF width of 300 ps and 100 ps respectively [59,60].
34
Other TCSPC Techniques
Application of TCSPC to diffusion in cells
Diffusion constants in cells are usually determined by
fluorescence correlation (FCS) techniques. Recording
an FCS function requires to detect the fluorescence
intensity in a fixed spot of the sample and to calculate
the autocorrelation function of the intensity or of the
photon detection times [14,15,16]. Although FCS is
not an imaging technique it is often used in
combination with imaging to define the location in a
cell where an FCS measurement has to be run.
Generally, the TCSPC technique is not only able to
record time-resolved images in laser scanning
microscopes, but also to recorded FCS data. An FCS
measurement can even be combined with a single
point lifetime measurement.
The multi-detector
technique described above for TCSPC imaging is also
available for FCS measurements. With the
introduction of the SPC-830 it is not longer necessary
to use independent modules for imaging and FCS. The
SPC-830 works for imaging and FCS recording as
well.
ps time from TAC / ADC
Laser
Detector
Channel
Laser
Photon
micro time
macro time
resolution 25 ps
FIFO
Buffer
time from start of experiment
Start
of
experiment Photons
resolution 50 ns
micro time
Det. No
macro time
micro time
Det. No
macro time
.
.
.
.
.
.
micro time
Det. No
macro time
micro time
Det. No
macro time
Readout
Histogram of
micro time
Hard disk
Fluorescence
decay
curves
Autocorrelation of
macro time
Fluorescence
correlation
spectra
FCS data recording does not build up histograms as
picoseconds
ns to seconds
the TCSPC imaging techniques do. Instead, it records Fig. 40: Combined liftime / FCS data acquisition by
the full information about each photon. Each entry
TCSPC
contains the time of the photon in the laser pulse
sequence, the time from the start of the experiment, and the detector channel. The data structure is
shown in Fig. 40. For each detector an individual correlation spectrum and a fluorescence decay
curve can be calculated. An instrument like this was used to detect and identify single molecules on
a substrate. By using four wavelength channels and a piezo scanning stage different molecules could
be identified in a time of the order of 1 ms [17]. Moreover, data from different detectors can be
cross-correlated. By detecting different chromophores in different wavelength intervals crosscorrelation can show whether the molecules of both chromophores and the associated protein
structures are linked or diffuse independently.
TCSPC Wide Field Imaging
TCSPC can be used for wide field imaging if a special detector is used. The detector consists of the
usual photocathode, a microchannel plate and a delay line system or a four-element anode system at
the output. By measuring the delay of the output pulses at the ends of the delay lines or the charge
distribution at the four anode sectors the location of a photon can be determined, see fig. 41.
35
Cathode
Cathode
Multichannel
Plate
Multichannel
Plate
A2
A1
A2
A1
A4
A3
Delay Line Anode
Quadrant Anode
Fig 41 : Wide field detectors for TCSPC
Wide-field TCSPC imaging requires several TCSPC channels or one TCSPC channel with four low
noise charge detection channels. The technique features high counting efficiency and relatively good
time resolution. Because additional signal processing is needed to obtain the position of a photon it
is difficult to reach high count rates. The most severe drawback is that the technique depends on
complex detectors that are not commonly available.
Comparison of Signal Processing Techniques
A comparison of the different lifetime techniques is given below.
The figure of merit, F, was taken from [39]. Small F values indicate high efficiency. For multiwavelength techniques it was assumed that two chromophores are measured which give
approximately the same count rates. Therefore, the efficiency, E, increases by a factor of 2 and F is
reduced by a factor of 1.4. The resulting F can be smaller and E greater than one because the values
are referred to a single detector measurement with ideal SNR.
For the time resolution of the gated image intensifier and the TCSPC technique it was assumed that
lifetimes down to 1/10 of the instrument response width can be de-convoluted, which is certainly a
conservative assumption. The resolutions given for the modulation techniques are the practically
obtained values from [19, 21,52-54]. The applicability to multi-exponential decay functions is more
or less theoretical. Resolution of double exponential decay functions in FRET systems has been
proved for TCSPC only.
The maximum count rate of the TCSPC method was defined as the rate at which the counting
efficiency drops to 50%. Therefore, the values are smaller that the reciprocal dead time.
36
Technique
F
E
Figure Effiof Merit ciency
Resolution
ps
Count Multi- Fast Con- Remark
Rate
Epon. Scan focal
MHz Decay
Gated Image Intensifier 200ps
4
0.057
20
unlim.
Single Channel mod.
Technique
Sinewave
Dirac
3.7
1.1
0.07
0.95
300
300
Modulated Image
Intensifier
Sinewave
Dirac
10
4.3
0.01
0.054
300
300
Multi-Gate SPC
2 λ Channels
2 Gates
8 Gates
2 Gates,
4 Detectors
TCSPC Imaging,
1 TCSPC Channel
100 kHz 4)
1 MHz 4)
1
1.06
TCSPC Imaging
4 TCSPC Channels,
4 λ Channels,
2 Chromophores
100 kHz 4)
1 MHz 4)
4 MHz 4)
10 MHz 4)
TCSPC Multiwavelength Imaging,
2-16 λ Channels,
2 Chromophores
100 kHz 4)
1 MHz 4)
yes
yes
no
applicable for
wide field
unlim. yes
unlim. yes
no
no
yes
yes
slow scan only
unlim.
unlim.
yes
yes
yes
yes
no
no
applicable for
wide field
1.5-2.1 0.7-0.8 100
1.1-1.5 0.8-0.95 <100
0.78-1.06 0.97-1.13 100
30
30
120
no
yes
no
no
no
no
yes
yes
yes
1
0.88
2.5 3)
2.5 3)
4 1)
4 1)
yes
yes
yes
yes
yes
yes
1 detector
0.71 5)
0.72
0.75
1.07
2 5)
1.94
1.76
0.88
2.5 3)
2.5 3)
2.5 3)
16 2)
16 2)
16 2)
yes
yes
yes
yes
yes
yes
yes
yes
yes
4 detectors,
4 parallel TCSPC
channels
0.71 5)
0.75
2 5)
1.76
2.5 3)
2.5 3)
4 1)
4 1)
yes
yes
yes
yes
yes
yes
up to 16 detector
channels recording
simultaeously in
1 TCSPC channel
1) Count rate for 50% counting loss. Dead time 150ns
2) Count rate for 50% counting loss. Four parallel channels with a dead time of 150ns each
3) With MCP-PMT
4) Overall count rate
5) F < 1 and E > 1 because 2 chromophores are detected simultaneously and the system is compared with a single
channel detecting only 1 chromophore
Summary
A wide variety of lifetime imaging techniques for time-resolved microscopy is available - gated and
modulated image intensifiers, single channel modulation techniques and gated and time-correlated
single photon counting. At the same time two generally different microscopy techniques exist - the
wide field microscope and the confocal or two-photon laser scanning microscope.
For wide field microscopy the gated and modulated image intensifiers and can be used. Wide-field
lifetime systems do not have the intrinsic depth resolution and optical sectioning capability of the
confocal and two-photon scanning systems.
For laser scanning microscopes the best results are achieved with multi-gate photon counting and
with the TCSPC imaging technique. Multi-gate photon counting can be used up to very high count
rates and efficiently detects single exponential lifetimes down to a few 100ps. It does, however, not
reach the efficiency and the time-resolution of TCSPC. It is not clear how far the high count rate of
the technique can be practically exploited because photobleaching sets a limit to the fluorescence
intensity obtained from the excited sample volume.
TCSPC is able to detect lifetimes down to a few ps and is able to resolve the components of multiexponential decay functions. It has a near-ideal efficiency and is able to detect in several wavelength
intervals simultaneously. Moreover, the TCSPC technique can be used to obtain combined
37
FCS / lifetime data in selected spots of a sample. TCSPC imaging works at an extremely high
scanning speed and is therefore compatible to almost any laser microscope.
Among all methods TCSPC imaging with confocal or two-photon microscopes scanning
microscopes is most suitable to meet the requirements of cell imaging - optical sectioning, multiwavelength recording, high detection efficiency, high time resolution, resolution of multiexponential decay functions, and applicability of FCS techniques.
References
[1] M. Minsky : Memoir on inventing the confocal microscope. Scanning 10, 128-138 (1988).
[2] J.G. White, W.B. Amos, M. Fordham : An evaluation of confocal versus conventional imaging of biological
structures by fluorescence light microscopy. J. Cell Biol. 105, 41-48 (1987).
[3] W. Denk, J.H. Strickler, W.W. Webb, Two-photon laser scanning fluorescence microscopy. Science 24 (1990) 7376
[4] A. Periasami, M. Elangovan, H. Wallrabe, J.N. Demas, M. Barroso, D.L. Brautigan, R.N. Day, Methods in cellular
imaging (ed. by A. Periasamy), 295-308, Oxford University Press, New York
[5] B. Herman, Fluorescence Microscopy, 2nd. edn. Springer-Verlag, New York 1998
[6] J.R. Lakowicz, Principles of Fluorescence Spectroscopy, 2nd. Ed., Plenum Press, New York 1999
[7] A. Periasami, R.N. Day, Visualizing protein interactions in living cells using digitized BFG imaging and FRET
microscopy, in: Green Fluorescent Proteins, ed. by K.F. Sullivan, S.A Key, Academic Press, 1999
[8] M. Elangovan, R.N. Day, A. Periasami, Nanosecond fluorescence resonance energy transfer-fluorescence lifetime
imaging microscopy to localize the protein interactions in a single cell. J. Microsc. 205 (2002) 3-14
[9 ] W.R.G. Baeyens, D. de Keukeleire, K. Korkidis, Luminescence techniques in chemical and biochemical analysis,
M. Dekker, New York, 1991
[10] M.R. Eftink, Fluorescence quenching: Theory and application, in: J.R. Lakowicz, Topics in fluorescence
spectrocopy, Vol. 2 (1991) 53-126
[11] M.A.M.J. van Zandvoort, W. Engels, C.J. de Grauw, H.C. Gerritsen, D.W. Slaaf, Lifetime imaging of the vital
DNA/RNA probe SYTO13 in helthy and apoptitic cells. Proc. SPIE 4626-83 (2002)
[11a] Van Zandvoort, M.A.M.J., de Grauw, C.J., Gerritsen, H.C., Broers, J.L.V., Egbrink, M.G.A. Ramaekers, F.C.S.,
Slaaf, D.W. Discrimination of DNA and RNA in cells by a vital fluorescent probe: Lifetime imaging of SYTO13 in
healthy and apoptotic cells. Cytometry 47 (2002) 226-223
[11b] Gerritsen, H.C., Sanders, R., Draaijer, A., Levine, Y.K. Fluorescence lifetime imaging of oxygen in cells. J.
Fluoresc. 7 (1997) 11-16
[12] Jens-Peter Knemeyer, Nicole Marmé, Markus Sauer, Probes for detection of specific DNA sequences at the single
molecule level. Anal. Chem. 72 (2000) 3717-3724
[13] M. E. Dickinson, C.W. Waters, G. Bearman, R. Wolleschensky, S. Tille, S.E. Fraser, Sensitive imaging of
spectrally overlapping fluorochromes using the LSM 510 Meta. Proc. SPIE 4620 (2002)
[14] Thomson, N,L., Fluorescence Correlation Spectrocopy. Topics in Fluorescence Spectroscopy, Vol. 1, ed. by J.R.
Lakowicz, 337-378. Plenum Press, New York, 1991
[15] K.M. Berland, P.T.C. So, E. Gratton, Two-photon fluorescence correlation spectroscopy: Method and application
to the intracellular environment. Biophys. J. 88 (1995) 694-701
[16] P. Schwille, S. Kummer, A.H. Heikal, W.E. Moerner, W.W. Webb, Fluorescence correlation spectroscopy reveals
fast optical excitation-driven intramolecular dynamics of yellow fluorescent proteins. PNAS 97 (2000) 151-156
[17] Michael Prummer, Christian Hübner, Beate Sick, Bert Hecht, Alois Renn, Urs P. Wild, Single-Molecule
Identification by Spectrally and Time-Resolved Fluorescence Detection. Anal. Chem. 2000, 72, 433-447
[18] J. Schaffer, A. Volkmer, C. Eggeling, V. Subramaniam, C. A. M. Seidel, Identification of single molecules in
aqueous solution by time-resolved anisotropy. Journal of Physical Chemistry A, 103 (1999) 331-335
38
[19] P.T.C. So, T. French, E. Gratton, A frequency domain microscope using a fast-scan CCD camera. Proc. SPIE 2137
(1994) 83-92
[20] M. Straub, S. W. Hell, Fluorescence lifetime three-dimensional microscopy with picosecond precision using a
multifocal multiphoton mocroscope. Appl. Phys. Lett. 73 (1998) 1769-1771
[21] A. Squire, P.J. Verveer, P.I.H. Bastiaens: Multiple frequency fluorescence lifetime imaging microscopy. J.
Microsc. 197, 136-149 (2000).
[22] M.A. Neil, A. Squire, R. Juskaitas, P.I. Bastiaens, T. Wilson, Wide-field fluorescence microscopy with laser
illumination. J. Microsc. 197 (2000) 1-4
[23] M.J. Cole, J.Siegel, R. Dowling, M.J. Dayel, D. Parsons-Karavassilis, P.M. French, M.J. Lever, L.O. Sucharov,
M.A. Neil, R. Juskaitas, T. Wilson, Time-domain whole-field lifetime imaging with optical sectioning. J. Microsc.
203 (2001) 246-257
[24] Betzig, E., Trautmann, J.K., Near-field optics: Microscopy, spectroscopy and surface modification beyond the
diffraction limit. Science 257, 189-195
[25] Betzig, E., Trautmann, J.K.,Harris, T.D., Weiner, J.S., Kostelak, R.L., Breaking the diffraction barrier: Optical
microscopy on a nanometric scale. Science 251 (1991) 1468-1470
[26] Dunn, R.C. Near-field scanning optical microscopy. Chem. Rev. 99 (1999) 2891-2928
[27] Hollars, C.W., Dunn, R.C., Submicron fluorescence, topology, and compliance measurements of phase-separated
lipid monolayers using tapping-mode near-fiel scanning optical microscopy. J. Phys. Chem. B. 101, 6313-6317
[28] Kirsch, A.K., Subramaniam, V., Jenei, A., Jovin, T.M., Fluorescence Resonance transfer detected by scanning nearfield optical microscopy. J. Microsc. 194 (1999) 448-454
[29] Femtolight series fiber lasers, IMRA America Inc., www.imra.com
[30] Pulsed diode lasers, PicoQuant GmbH, www.picoquant.com
[31] BDL-450 picosecond diode laser. www.becker-hickl.com
[32] BHL-250 pulsed diode lasers. www.becker-hickl.com
[33] M. Kress, Th. Meier, T. A. A. El-Tayeb, R. Kemkemer, R. Steiner, A. Rück, Short-pulsed diode lasers as an
excitation source for time-resolved fluorescence applications and confocal laser scanning microscopy in PDT, Proc.
SPIE 4431(2001)
[33a] A. Rück, F. Dolp, C. Happ, R. Steiner, M. Beil, Fluorescence lifetime imaging (FLIM) using ps pulsed diode
lasers in laser scanning microscopy. Proc. SPIE 4962-44 (2003)
[34] Hidehiro Kume (Chief Editor), Photomultiplier Tube, Hamamatsu Photonics K.K., 1994
[35] SPCM-AQR series. www.perkinelmer.com/opto
[36] S. Cova, S. Lacaiti, M.Ghioni, G. Ripamonti, T.A. Louis, 20-ps timing resolution with single-photon avalanche
photodiodes, Rev. Sci. Instrum. 60, 1989, 1104-1110
[37] M. Nagorni, S. W. Hell, 4Pi-confocal microscopy provides three-dimensional images of the microtubule network
with 100- to 150-nm resolution. J. of Structural Biology 123 (1998) 236-247
[38] Ballew, R.M., Demas, J.N., An error analysis of the rapid lifeetime determination method for the evaluation of
single exponential decays. Anal. Chem. 61 (1989) 30-33
[39] K. Carlsson, J.P. Philip, Theoretical Investigation of the Signal-to-Noise ratio for different fluorescence lifetime
imaging techniques. Proc. SPIE 4622-08 (2002)
[40] J. Widengreen, R. Rigler, Mechanisms of photobleaching investigated by fluorescence correlation spectroscopy.
Bioimaging 4 (1996) 149-157
[41] A.A. Heikal, W.W. Webb, One- and two-photon time-resolved fluorescence spectroscopy of selected fluorescent
markers: photobleaching, triple-, and singlet-state dynamics. Biophys. J. 73 (1999) 260
[42 E.J. Sánchez, L. Novotny, G.R. Holtom, X.S. Xie. Room-Temperature fluorescence imaging and spectroscopy of
single molecules by two-photon fluorescence. J. Phys. Chem. A 101 (1997) 7019-7023
[43] G.H. Patterson, D.W. Piston, Photobleaching in two-photon exciatation microscopy. Biophys. J. 78 (2000) 21592162
39
[44]G. J. Brakenhoff, M. Müller, R.I. Grauharali, Analysis of efficiency of two-photon versus single-photon absorption
for fluorescence generation in biological objects. J. Microsc. 183 (1996) 140-144
[45] Hopf, A., Neher, E., Highly nonlinear Photodamage on two-photon fluorescence microscopy. Biophysical Journal
80 (1002) 2029-2036
[46] K. König, P.T.C. So, W.W. Mantulin, B.J. Tromberg, E. Gratton, Two-Photon excited lifetime imaging of
autofluorescence in cells during UVA and NIR photostress. J. Microsc. 183, 197-204
[47] K. Koenig, T.W. Becker, P. Fischer, I. Riemann, K.-J. Halbhuber, Pulse-length dependence of cellular response to
intense near-infrared pulses in multiphoton microscopes. Opt. Lett. 24 (1999) 113-115
[48] J. Bewesdorf, R. Pick, S. W. Hell, Multifocal multiphoton microscopy. Opt. Lett. 23 (1998) 655-657
[49] A.H. Buist, M. Müller, J. Squier, G.J. Brakenhoff, Real time two-photon absorption microscopy using multi-point
excitation. Journal of Microscopy 192 (1998) 21767-226
[50] P. Urayama, J.A. Beamish, F.K. Minn, E.A. Hamon, M.-A. Mycek, A UV fluorescence lifetime imaging
microscope to probe endogenous cellular fluorescence. CLEO 2002, Technical Digest, CTh062
[51] Pico Star Camera, www.lavision.com
[52] K. Carlsson, A. Liljeborg, Simultaneous confocal lifetime imaging of multiple fluorophores using the intensitymodulated multiple-wavelength scanning (IMS) technique. J. Microsc. 191 (1998) 119-127
[53] K. Carlsson, A. Liljeborg, Confocal fluorescence microscopy using spectral and lifetime information to
simultaneaously record four fluorophores with high channel separation. J. Microsc. 185 (1997) 37-46
[54] P.T.C. So, T. French, W.M. Yu, K.M. Berland, C.Y. Dong, E. Gratton, Time-resolved fluorescence microscopy
using two-photon excitation. Bioimaging 3 (1995) 49-63
[55] P.T.C. So, T. French, E. Gratton, A frequency domain microscope using a fast-scan CCD camera. Proc. SPIE 2137
(1994) 83-92
[56] A. Squire, P.J. Verveer, P.I.H. Bastiens: Multiple frequency fluorescence lifetime imaging microscopy. J. Microsc.
197, 136-149 (2000).
[57] E.P. Buurman, R. Sanders, A. Draaijer, H.C. Gerritsen, J.J.F. van Veen, P.M. Houpt, Y.K. Levine, Fluorescence
lifetime imaging using a confocal laser scanning microscope. Scanning 14, 155-159 (1992).
[58] J. Sytsma, J.M. Vroom, C.J. de Grauw, H.C. Gerritsen, Time-gated fluorescence lifetime imaging and microvolume
spectroscopy using two-photon excitation. Journal of Microscopy, 191 (1998) 39-51
[59] H7422 data sheet, Hamamatsu Photonics, www.hamamatsu.com
[60] R3809U data sheet, Hamamatsu Photonics, www.hamamatsu.com
[61] Scully, A.D., Ostler, R.B., Phillips, D., O’Neill, P.O., Townsend, K.M.S., Parker, A.W., McRobert, A.J.,
Application of fluorescence lifetime imaging microscopy to the investigation of intracellular PDT mechanisms.
Bioimaging 5 (1997) 9-18
[62] D.V. O’Connor, D. Phillips, Time Correlated Single Photon Counting, Academic Press, London 1984
[63] SPC-134 through SPC-730 TCSPC Modules, Operating Manual and TCSPC Compendium. Becker & Hickl
GmbH, www.becker-hickl.com
[64] W. Becker, H. Stiel, E. Klose, Flexible Instrument for time-correlated single photon counting, Rev. Sci. Instrum. 62
(1991) 2991-2996
[65] I. Bugiel, K. König, H. Wabnitz, Investigations of cells by fluorescence laser scanning microscopy with
subnanosecond resolution. Lasers in the Life Sciences 3 (1989) 47-53
[66] W. Becker, A. Bergmann, H. Wabnitz, D. Grosenick, A. Liebert, High count rate multichannel TCSPC for optical
tomography. Proc. SPIE 4431, 249-254 (2001)
[67] W. Becker, A. Bergmann, K. Koenig, U. Tirlapur, Picosecond fluorescence lifetime microscopy by TCSPC
imaging. Proc. SPIE 4262 (2001) 414-419
[68] Wolfgang Becker, Klaus Benndorf, Axel Bergmann, Christoph Biskup, Karsten König, Uday Tirplapur, Thomas
Zimmer, FRET Measurements by TCSPC Laser Scanning Microscopy, Proc. SPIE 4431(2001) 94-98
40
[69] Wolfgang Becker, Axel Bergmann, Georg Weiss, Lifetime Imaging with the Zeiss LSM-510. Proc. SPIE 4620
(2002) 30-35
[70] K. Koenig, C. Peuckert, I. Riemann, U. Wollina, Optical tomography of human skin with picosecond time
resolution using intense near infrared femtosecond laser pulses, Proc. SPIE 4620-36
[71] S. Ameer-Beg, P.R. Barber, R. Locke, R.J. Hodgkiss, B. Vojnovic, G.M. Tozer, J. Wilson, Application of
multiphoton steady state and lifetime imaging to mapping of tumor vascular architecture in vivo. Proc. SPIE 4620
(2002) 85-95
[72a] Wolfgang Becker, Axel Bergmann, Christoph Biskup, Thomas Zimmer, Nikolaj Klöcker, Klaus Benndorf, Multiwavelength TCSPC lifetime imaging. Proc. SPIE 4620 (2002) 79-84
[72b] Wolfgang Becker, Axel Bergmann, Christoph Biskup, Laimonas Kelbauskas, Thomas Zimmer, Nikolaj Klöcker,
Klaus Benndorf, High resolution TCSPC lifetime imaging. Proc. SPIE 4963-30 (2003)
[73] A. Schönle, M. Glatz, S. W. Hell, Four-dimensional multiphoton microscopy with time-correlated single photon
counting. Appl. Optics 39 (2000) 6306-6311
[74] S. Jakobs, V. Subramaniam, A. Schönle, Th. Jovin, S.W. Hell, EGFP and DsRed expressing cultures of escherichia
coli imaged by confocal, two-photon and fluorescence lifetime microscopy. FEBS Letters 479 (2000) 131-135
[75] D. Schweitzer, A. Kolb, M. Hammer, E. Thamm, Tau-mapping of the autofluorescence of the human ocular fundus.
Proc. SPIE Vol. 4164, 79-89
[75b] D. Schweitzer, A. Kolb, M. Hammer, E. Thamm, Basic investigations for 2-dimensional time-resolved
fluorescence measurements at the fundus. International Ophthalmology 23 (2001) 399-404
[76] HRT-41, HRT-81, HRT-82 Routing Devices, Operating Manual. Becker & Hickl GmbH, www.becker-hickl.com
[77] PML-16 16 Channel Detector Head, Operating Manual. Becker & Hickl GmbH, www.becker-hickl.com
[78] E-S. Kwak, T.J.Kang, A.A. Vanden Bout, Fluorescence lifetime imaging with near-field scanning optical
microscopy. Anal. Chem. 73 (2001) 3257-3262
[79] W. Hartmann, F. Bernhard, Fotovervielfacher und ihre Anwendung in der Kernphysik. Akademie-Verlag Berlin
1957
[80] M. Köllner, J. Wolfrum, How many photons are necessary for fluorescence-lifetime measurements? Phys. Chem.
Lett. 200 (1992) 199-204
[81] Gerritsen, H.C., Asselbergs, M.A.H., Agronskaia, A.V., van Sark, W.G.J.H.M., Fluorescence lifetime imaging in
scanning microscopes: acquistition speed, photon economy and lifetime resolution. J. Microsc. 206 (2002) 218-224
[82] L. Kelbauskas, W. Dietel, Internalization of aggregated photosensitizers by tumor cells: Subcellular time-resolved
fluorescence spectroscopy on derivates of pyropheophorbide-a ethers and chlorin e6 under femtosecond one- and
two-photon excitation. Photochem. Photobiol. 76 (2002) 686-694
41
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

advertising