Differences between FLIM phasor analyses for data collected with the Becker and Hickl SPC830 card and with the FLIMbox card

Abstract

The phasor approach to FLIM (Fluorescence Lifetime Imaging Microscopy) is becoming popular due to the powerful fit free analysis and the visualization of the decay at each point in images of cells and tissues. However, although several implementation of the method are offered by manufactures of FLIM accessories for microscopes, the details of the conversion of the decay to phasors at each point in an image requires some consideration. Here, we show that if the decay is not properly acquired, the apparently simple phasor transformation can provide incorrect phasor plots and the results may be misinterpreted. In particular, we show the disagreement in experimental data acquired on the same samples using the two cards (FLIMbox, frequency domain and Becker & Hickl BH 830, time domain) and the effect produced by using the BH 830 card with different settings. This difference in data acquisition translates to the assignment of phasor components calculated using different acquisition parameters. This effect is already present in the original data that are not acquired with the proper parameters for the phasor conversion. We also show that the difference in the resolution of components already exists in the data acquired in the time domain when used with settings that do not allow acquisition of the fluorescence decay on a sufficient large time scale. RESEARCH HIGHLIGHTS: This paper is intended to made researchers aware of some simple requirements for the conversion of time-domain data (typically TCSPC) to phasors. The use of phasors for FLIM analysis has seen a surge of popularity. Since the phasor approach is a fit free method and has a powerful visualization of the data, it appears very simple to use. This paper shows that when the original data in the time domain is not acquired with the proper time range to cover the lifetimes in a sample, the conversion to phasors can produce very erroneous results. These results are appearing more frequently in the literature since many of the manufacturers of FLIM accessories for microscopes are now offering the phasor analysis in their software. Here, we show that the phasor transformation per se cannot correct for the problems with data acquisition and that one is misled to think that the "phasor approach" is a universal fix for the lack of the proper time range for data acquisition.

Keywords: Becker & Hickl; FLIM; FLIMBox; fluorescence; lifetime; phasor analysis.

Figure 1

BH 830 card acquisition of the decay of a solution of Rhodamine 110 excited with a high rep laser at 80 MHz with two different settings of the card. (A) Gain of 4 gives a TAC time range of 12.5 ns. Due to the non-linearity of the TAC at early and later times, only one portion of the decay, indicated by the red and blue lines, is valid giving a total range of about 9 ns, which is insufficient to cover an entire period of the laser. (B) Setting the gain at 2 gives a total larger range (25 ns) but only one part corresponding to 12.5 ns is used (indicated by the red and blue lines) which covers an entire period of the excitation laser. The grayed parts of the decay are not used.

Figure 2

Data acquired with the FLIMbox card in the frequency domain, directly as phasors. Data were obtained for A) solution of Rhodamine 110 (red circle). B) Cy3 (green circle) and C) 9-10 H Acridanone (blue circle) and their mixtures. D) Mixture of Rho110 and 9-10 H Acridanone. E) Rho 110 and Cy3 Mix 1. F) Rho 110 and Cy3 Mix 2. G) 9-10 H Acridanone and Cy3 (mix1). H) 9-10 H Acridanone and Cy3 (mix2). I) Mixture of 9-10 H Acridanone, Rho 110 and Cy3. The graphical solution for the fraction of the components for the mixtures in Figure 2 are: (D) F(9-10 H Acridanone)=0.705, (E) F(Cy3)=0.228, (F) F(Cy3)=0.556, (G) F(Cy3)=0.405, (H) F(Cy3)=0.598, (I)-F9Rhodamine)=0.177, F(9-10 H Acridanone)=0.519, F(Cy3)=0.304.

Figure 3

Effect of the number of photons/pixel on the width of the phasor distribution for data acquired with the FLIMbox. Only the coordinate G is shown since the phasor distribution are symmetrical. Note that width of the phasor distribution is independent on the position of the phasor in the phasor plot but only depends on the number of photon acquired for a given pixel. For this figure we use the mixture of 3 components to show that the composition of the phasor does not affect the cluster distribution.

Figure 4

Effect of gain of the BH 830 card on the location of the phasors of single exponential decay and of mixtures. The Calibration of the phasor plot is done using the Coumarin 6 standard with lifetime of 2.5ns. The labels of the phasor clusters is the same used in Figure 2. A) Solution of Rhodamine 110. B) Cy3. C) 9-10 (H) Acridanone. D) Mixture of Rho110 and 9-10 (H) Acridanone. E) Rho 110 and Cy3 Mix 1. F) Rho 110 and Cy3 Mix 2. G) 9-10 H Acridanone and Cy3 (mix1). H) 9-10 (H) Acridanone and Cy3 (mix2). I) Mixture of 9-10 (H) Acridanone, Rho 110 and Cy3.

Figure 5

Convallaria sample and phasor distribution. (A) Image obtained with FLIMbox and (B) using BH SPC830 card with a gain of 4 and (C) using a gain of 2. (D)Phasor distribution for the corresponding instrument settings. G-I) Corresponding color mapped FLIM images. The color map for the phasor distribution is shown in (D-F). The results show that FLIM acquisition with FLIMbox and BH SPC830 card with a gain of 2 give similar phasor plot. The phasor distribution and hence, the color map is completely different for gain of 4.

Figure 6

Graphical solution of two-exponential components mixture using the phasor plot of the first (A) and (B) second harmonics at 80 MHz and 160 MHz, respectively. An experimental phasor point P is shown in the first and second harmonic plots. The graphical algorithm searches for the pair of lifetimes T1 and T2 that determine the line that simultaneously passes through P in both plots. The inset in (A) and (B) is the zoom of the data points in P and the line (red line) passing through the center of P for the first and second harmonics, respectively.

Figure 7

Effect of the number of counts on the width of the phasors cluster and the action of the median filter in reducing the spread of the cluster. Data is shown here for Rhodamine 110, single exponential collected with the BH SPC830 card, gain 2. The expected value for this dye measured at 80 MHz is Phase= 63.9 Mod= 0.448 TP= 4.056 ns TM= 3.969 ns. These values are obtained at the 3 counting rates, whether the median filter is applied or not. Panels (A), (B) and (C) show the effect of the number of photons collected per pixel on the phasor cluster. Data are summarized in graph (J), where FWHM (full width at half maxima) of the distribution in S and G co-ordinates are plotted. The results of the application of a median filter 3×3 and 5×5 are shown in panels (D), (E), (F) and (G), (H) and (I), respectively.