MIET analysis software

After the first applications of MIET microscopy had been published, the further development and establishment of the technique quickly required a well-maintained software package implementing the latest analysis methods. Thus, a major part of this work was to improve, extend and re-implement previous proof-of-principle methods based on the theoretical concepts introduced above into a reliable and consistent toolchain. To simplify and document the data analysis steps for general users, the software was equipped with a graphical user interface (GUI). Since this is the part of the program which primarily determines the user experience, we call the whole software package MIET GUI. It has been described in detail in [7], and a free version is available online from the institute’s homepage.

In general, the data analysis of a MIET experiment consists of three steps. Firstly, lifetime values are extracted from the raw data. Secondly, a MIET lifetime-versus-height calibration curve is calculated based on the sample parameters. Thirdly, the lifetimes are converted to height values based on the MIET curve. The implementation of the first step varies strongly based on the FLIM setup used to acquire the data, starting from the fundamental differences between frequency- or time-domain measurements, but also including the very specific data formats used by different manufacturers. Commercial FLIM setups usually include a proprietary software which performs this analysis step, while users who tailor a FLIM setup to a specific purpose often like to have full control over the employed fitting algorithms. In an internal version of the software package, we included the full lifetime fitting process: Starting with the extraction of photon arrival times from the binary data produced by the counting electronics, TCSPC histograms are compiled and dead-time corrected as described in section 6.3.2 in the appendix, and finally fitted using IRF- or tail-fits as described in section 3.5. However, this module was not included in the published version for the aforementioned reasons.

air

PVA

gold

glass 40 nm

12 nm

``infinite"

``infinite"

Figure 4.1: Screenshots of the MIET graphical user interface. Top: Main window, choice between conversion of lifetimes to height values (Complete analysis) or only calculation of MIET curve; specification of lifetime data. Bottom: Parameters window, opened

A screenshot of the published version of the MIET GUI is shown in figure 4.1. It allows the user to specify the two important sets of sample parameters – the refractive indices and thicknesses of the materials making up the stratified system, as well as the optical properties of the fluorophore – which are needed to calculate a MIET calibration curve, and to specify the lifetime values that are to be converted into heights.

The software simplifies the workflow in several ways. To name just two examples, the GUI offers the option to only output the MIET curve¹, or to process an arbitrary number of data sets from the same sample type (i.e. with the same MIET curve) in bulk. The representation of the stratified system as three stacks of layers – one layer containing the fluorophore, with one stack above and one below it, see lower half of figure 4.1 – is capable of mapping virtually any sample that has a planar geometry. This is in accordance with our aim of creating a general-purpose MIET analysis software that is not restricted to a specific sample type. The only assumption made by the algorithm is that the bottommost medium of the lower stack (usually the glass cover slip) and the uppermost medium of the upper stack (usually buffer in a cell sample, or air when molecules are deposited on a silica spacer) are assumed to be infinitely thick.

In the simplest case, a MIET curve is calculated for one single wavelength λ, usually the maximum emission wavelength of the fluorophore. In reality, however, emitters fluoresce over a range of wavelengths, which all have slightly different MIET curves.

This is taken into account if the free space fluorescence spectrum F(λ) is specified:

hτ(z)i_λ = Z λmax

λmin

τ(z, λ)·F(λ) dλ

Z λmax

λmin

F(λ) dλ. (4.1)

When the actual evaluation is initiated, the program first calculates the MIET curve.

In principle, the total energy S_tot emitted per time by one emitter depends not only on the z-position of the emitter but also on its orientation θ with respect to the optical axis. When denoting the values for fluorophores with a vertical or a horizontal dipole moment as Stot,⊥(z) and Stot,k(z), respectively, the linearity of the relevant equations leads to the simple relation:

S_tot(z, θ) = Stot,⊥(z)·cos²θ+Stot,k(z)·sin²θ. (4.2) In our evaluations, we assume an ensemble of randomly oriented fluorophores, and thus set S_tot(z) = [S_tot,⊥(z) + 2S_tot,k(z)]/3². This value is then used to determine the MIET curve according to equation (2.222): τ(z) =τ₀·S_tot(z)/S₀. Since this function oscillates with decreasing amplitude at largez, an unambiguous relation betweenτ andz can only be found up to the z-value of the first minimum, z_min. If some prior knowledge makes it unlikely that fluorophores are at high z-values, the user can alternatively choose to use the curve up to the z-value of the first maximum, z_max. All lifetimes higher than τ(z_min/max) are marked as “not a number” when transforming lifetime values into heights.

Finally, the resulting height images are automatically plotted and the raw data is saved.

1 Which can be used to estimate the impact of certain sample parameters on the obtainable height resolution when designing an experiment.

2 If it is known that the fluorophores have a specific orientation, the code can be adapted accordingly.

4.1.2 Cell-substrate dynamics of the epithelial-to-mesenchymal