TCSPC

Lifetime imaging can be implemented both in wide field and in scanning microscopes such as confocal microscopes and two-photon excitation microscopes. The most common implementations in time-domain fluorescence lifetime imaging microscopy (FLIM) are based on TCSPC [8, 9] and time-gating (TG) [2, 10],... 

Fig. 3.1. Schematic diagram of a TCSPC setup. Using a fast timing device (e.g., time-to-amplitude-converter) the time is measured between the excitation pulse and the detection of a photon. By repeating this procedure many times a decay curve is measured. TAC time to amplitude convertor, MCA multi channel analyzer, PC personal computer. 

In Fig. 3.5A a comparison between time-gated detection and TCSPC is shown. The time-gated detection system was based on four 2 ns wide gates. The first gate opened about 0.5 ns after the peak of the excitation pulse from a pulsed diode laser. The TCSPC trace was recorded using 1024 channels of 34.5 ps width. The specimen consisted of a piece of fluorescent plastic with a lifetime of about 3.8 ns. In order to compare the results, approximately 1700-1800 counts were recorded in both experiments. The lifetimes obtained with TG and TCSPC amounted to 3.85 0.2 ns and 3.80 0.2 ns respectively, see Fig. 3.5B. Both techniques yield comparable lifetime estimations and statistical errors. 

Fig. 3.5. A comparison between TG and TCSPC using the same number of detected photons. (A) The distribution of photons over the time bins. (B) Bar plot of the lifetimes including errors (n — 4).

Conventional TCSPC equipment has been successfully employed in LSM for fluorescence spectroscopy on discrete microscopic volumes [18, 19] and for fluorescence lifetime imaging at a low acquisition speed [1], The use of conventional TCSPC equipment for imaging results in very long acquisition times, several to many minutes per (time-resolved) image. Importantly, operating the TCSPC detection system at too high detection rates, above 5% of the excitation frequency, results in distortion of the recorded decay curve [20],... 

At present dedicated TCSPC FLIM boards are commercially available. They are compatible with most LSMS and are easily synchronized with the scanning microscope and pulsed laser. These boards, often plug-in cards for PCs, have a lower deadtime than do the conventional TCSPC electronics intended for use in spectroscopy and the memory bottle neck of the histogram-ming memory has been removed [21, 22], Consequently, these dedicated boards provide higher acquisition speeds. 

Dedicated TCSPC electronics is used in all practical TCSPC-FLIM implementations [21, 22]. There are several issues that should be noted. First of all, the lifetime acquisition has to be synchronized with the scanning of the confocal or multiphoton microscope. To this end, the pixel clock and often the line and frame synchronization signals of the scanning microscope are used. 

In TCSPC imaging, the number of time channels is usually restricted to 32-128. In general, a higher number of channels do not provide additional information because only a limited number of detected photons are accumulated per pixel, often several hundred to a few thousand. 

In time-gated photon counting, comparatively high photon count rates can be employed count rates as high as 10 MHz are often used. TG has the advantage of virtually no dead-time of the detection electronics ( 1 ns), whereas the dead-time of the TCSPC electronics is usually on the order of 125-350 ns. This causes loss of detected photons, and a reduced actual photon economy of TCSPC at high count rates. 

Not only PMTs and other detectors such as avalanche photodiodes suffer from dead-time effects also the detection electronics may have significant dead-times. Typical dead-times of TCSPC electronics are in the range 125-350 ns. This may seriously impair the efficiency of detection at high count rates. The dead-time effects of the electronics in time-gated single photon detection are usually negligible. 

Therefore, the throughput of current systems based on time-gated SPC is somewhat higher than in TCSPC-based systems. 

Both TCSPC and TG benefit from operation in SPC mode. SPC results in little or no noise and a high photon-economy [10]. Therefore, TCSPC and TG are ideal for high spatial and lifetime resolution imaging [24], Both techniques offer high image contrast also on dim samples. However, the dead-time of the detectors and the point scanning character limit the throughput of these systems. 

In wide field microscopy, spatial information of the entire image is acquired simultaneously thus providing comparatively short acquisition times compared with scanning microscopy implementations. Combining TCSPC with wide field microscopy is not straightforward. However, a four quadrant anode multichannel plate (MCP) has been used for time- and space-correlated SPC experiments [25, 26]. This detector has excellent timing properties that make it very suitable for FLIM. Unfortunately, it can be operated only at low count-rates ( 105-106 Hz) therefore, it requires comparatively long acquisition times (minutes). 

The optimum gate width AT for a specific lifetime amounts to 2.5t. In Fig. 3.10 a typical F-x curve is shown for time domain lifetime detection with a variable number of time bins and a total detection window (sum of all the time bins/gate widths) of 10 ns (de Grauw and Gerritsen, 2001). The curves are representative for both TCSPC and TG operating in a high excitation frequency mode of... 

TCSPC is inherently self-referenced and therefore, with the exception of the regular recording of the IRF, a TCSPC system requires practically no day-to-day calibration. In TCSPC the recording... 

Because of the low timing-jitter (down to 25 ps) TCSPC-based systems are often equipped with a MCP-PMT at detriment of acquisition speed (<106 counts per second). On the other hand, a TG-SPC system equipped with four gates and a fast PMT (10 MHz) could be slower than a TCSPC at low count-rates (<100 kHz), because of a lower photon-economy. However, already at 1 MHz, the former would be almost three times faster and more the one order of magnitude faster at 10 MHz. 

The above limitations are implementation dependent and no intrinsic limitations. The throughput of TCSPC can for instance be improved by the use of multiple detectors, and multiple TCSPC boards [42]. The photon-economy of TGSPC could be optimized somewhat by increasing the number of gates. 

Becker, W., Bergmann, A., Biskup, C., Kelbauskas, L., Zimmer, T., Klocker, N. and Benndorf, K. (2003). High resolution TCSPC lifetime imaging. Proc. SPIE 4963, 175-84. 

Fig. 4.1. Multiphoton fluorescence intensity (A-C) and TCSPC fluorescence lifetime images (D-F) of fresh unstained sections of human cervical biopsy excited at 740 nm and imaged between 385 and 600 nm. The individual acquisition times were 600 s. Adapted from Fig. 22.11 of Ref. [8].

The introduction and diversification of genetically encoded fluorescent proteins (FPs) [1] and the expansion of available biological fluorophores have propelled biomedical fluorescent imaging forward into new era of development [2], Particular excitement surrounds the advances in microscopy, for example, inexpensive time-correlated single photon counting (TCSPC) cards for desktop computers that do away with the need for expensive and complex racks of equipment and compact infrared femtosecond pulse length semiconductor lasers, like the Mai Tai, mode locked titanium sapphire laser from Spectra physics, or the similar Chameleon manufactured by Coherent, Inc., that enable multiphoton excitation. 

In contrast, time domain instruments attempt to directly measure the decay characteristics of a fluorophore of interest by excitation with ultrashort light pulses and monitoring the decay using either TCSPC [5] or a time gated image intensifier [8],... 

---