# GUI Documentation¶

## Fitting options¶

Global Fitting Specify the scope of the global variables.

• Pixel-wise the decays are fitted on a pixel-by-pixel basis
• Image-wise global fitting is performed across each image individually, i.e. lifetimes are constant across each image
• Global global fitting is performed across the entire dataset, i.e. lifetimes are constant across all images.

Global Variable combo-box to confine the scope of the global fit to a particular multi-well plate metadata parameter (e.g. ‘Row’ or ‘Column’). If a parameter is selected all images with the same parameter value will be globally analysed together.

Global Mode combo-box to set up the method of global analysis. The following choices are available:

• Global binning All pixels in the global fit are binned into a single decay. This decay is then fitted to determine the global parameters. Each pixel is then fitted individually with the global parameters fixed (using the method of linear least square minimisation) to determine the local variables.
• Global analysis All the pixels in the global fit are simultaneously analysed using Variable Projection algorithm (see in Advanced tab)

No. Exp combo-box to set up the number of exponential decay components $$n_\tau$$

No. Fixed The number of fixed fluorescence lifetimes; the first No. Fixed values in the Initial Tau list are fixed.

Fit contributions combo-box to set up the fitting mode for the amplitudes Ak (formula (3) and (4)). The following choices are available:

• Fixed the fractions $$\beta_i$$ are fixed for corresponding amplitudes (set up by the textboxes appearing to the right)
• Fit locally amplitudes are fitted locally, i.e. on a pixel-by-pixel basis
• Fit globally amplitudes are fitted globally, i.e. they are invariant across the image. Use this option when fitting a complex donor FRET model or polarisation resolved data.
• Fit globally (Grouped): use this option to fit a model where we have a mixture of several fluorophores with complex decays. The lifetimes and contributions are specified in groups, indexed from zero. Within each group the lifetime and fractional contributions are invariant across the image but the fractional population of each group varies locally. The groups are specified using the Initial tau list.

Fit Reference combo-box to set up the fitting mode for the variable $$\tau_R$$ (formula (4)), as Fixed or Fitted. The reference lifetime is fitted globally.

Initial Tau combo-box to choose the initial values of lifetimes. These are estimated automatically unless ‘Automatically estimate initial guesses’ is deselected.

### Stray Light tab¶

You can select the fitting scope of the “stray light” variables entering the model expressions (3-5), by choosing either Fixed, or Fitted Locally, or Fitted Globally. The scope is set for the following variables:

• Offset A constant background value, Z in the formula (5)
• Scatter A fast-light scatter contribution, S in the formulas (3) and (4)
• TV Background A time-varying background contribution that is scaled by the coefficient V, see formula (5). This option can only be used if a time varying background has been loaded. The V coefficient can be fitted globally or fixed by setting the TVB Background to ‘Fixed’ and TVB to 1.

### Anisotropy tab¶

These options are used to set up an anisotropy decay model if polarisation resolved data has been loaded.

No Decays combo-box to specify the number of rotational correlation times to fit.

No Fixed combo-box to specify the number of rotational correlation times that are fixed. The first No Fixed are fixed to the value specified in the Phi list.

Use Phi textbox to specify the initial guesses and fixed values for the anisotropy correlation times.

### FRET tab¶

These options are used to fit a complex-donor FRET decay model. In this model we assume that the donor fluorophore has two or more conformations with associated lifetimes whose FRET efficiencies are linked by their relative quantum yields. To use this mode Fit Contributions should be set to Fixed or Globally in the Lifetime tab, as we assume that the fractional contribution of the various donor species are invariant across the image.

No. FRET Species combo-box to specify the number of different FRET conformations present. For example, for an intra-molecular FRET probe where there are likely to be two different conformations with different (but non-zero) FRET efficiencies, set this to two and set Include Donor Only to No. For an inter-molecular FRET probe where the unbound state will have a zero FRET efficiency, set this to one and set Include Donor Only to Yes.

No. Fixed combo-box to set the number of FRET conformations which have fixed efficiencies. The first No. Fixed FRET efficiencies in the E list will be fixed.

Include Donor Only combo-box to specify whether a donor-only contribution should be included, that is a contribution from a species which is not FRETing.

No. Threads Set up the number of threads used in multi-threading processing. This is automatically set to the number of cores present on the local machine which in general is optimal.

Algorithm combo-box to choose the method and the type of error function minimization. The following methods are available:

• Variable Projection: Use separable fitting to minimise the mean-square error function.
• Maximum Likelihood: Use the maximum likelihood fitting metric, only applicable to pixel-wise fitting. Use this option for fitting data with low photon count numbers, particularly TCSPC data.
Weighting Mode combo-box to select the method of weighting the residuals entering the mean-square error function when using Variable Projection.
Error function =

where ti are the bin (delay) times, σ(ti) is weighting coefficient, and Dmodel(ti) and y(ti) are the model and measured intensity values, respectively.

• Average Data Weight according to the average measured decay across the global scope. This is the default, and advised option.
• Pixelwise Data Weight according the measured decay in each pixel $$chi^2(ti)=y(ti)$$. Can display significant bias at low photon counts.
• Model Weight according to the model decay $$\chi^2(ti)=D_{model}(ti)$$. May lead to problems with convergence if initial estimates are significantly different from the true values.

Pulse Train Correction combo-box to specify whether incomplete decays should be included in the fit. Unless you have a very low repetition rate laser source or are fitting very fast decays, this option should only be disabled for testing and when fitting data simulated without incomplete decays.

Auto Resampling combo-box to dynamically combine bins to ensure that there are at least 10 counts in each bin. This option is superseded by Maximum Likelihood fitting.

IRF combo-box to set up the type of IRF. The following choices are available:

• Single Point Use the same IRF for all pixels
• SV IRF Use a spatially varying IRF that has been loaded from the IRF menu. Each pixel is fitted using the IRF from the corresponding pixel in the spatially varying IRF.

IRF shift map Use an IRF shift map that has been loaded from the IRF menu. The same IRF is used in each pixel, time shifted according to the shift map loaded.

Live Fit checkbox to dynamically refit the currently selected decay as fitted parameters are changed or as the selected region is changed. This option does not automatically fit the whole dataset.

Calculate Errors checkbox to estimate confidence intervals on the global parameters based on the F statistic. This option should currently be considered experimental and significantly increases the fitting time.

## Instrument and preprocessing options¶

### Data tab¶

These settings are used to specify data acquisition parameters and set any required pre-processing.

Smoothing The kernel size used for spatially smoothing the data to reduce the noise. The equivalent binning setting in Becker and Hickl software is provided for reference.

Note

To ensure that the smoothing does not affect the choice of background or threshold values, data is averaged rather than summed over the kernel. The $$\chi^2$$ value correctly accounts for the level of smoothing applied.

Integrated Min. Pixels which have an integrated intensity of less than this value will be excluded from the fit. This may be used to exclude pixels which are too dim to give a good fit.

Note

The integrated intensity is calculated after accounting for the specified background and is not smoothed.

Time Min. and Time Max. The limits of the valid FLIM signal. Time bins or gates outside of these values will be excluded from the fit. Use this to remove, e.g. TCSPC data which falls outside of the linear range of the TAC.

Counts/Photon The average number of digital counts recorded per photon. This is used to correct the $$\chi^2$$ value to account for the gain provided by the intensifier and camera.

Tip

For TCSPC data this should be left at 1.

Rep. Rate The repetition rate of the laser used to acquire the data, in MHz. This is used to correct for incomplete decays.

Warning

This value must be set correctly; if it is set incorrectly the fitting may fail or produce a very poor fit.

Gate Max. Saturation limit of the data. Pixels which have a time bin or gate above this limit will be excluded from the fit. This can be used to exclude saturated pixels which will bias the fit.

Tip

For 16-bit TCSPC data (e.g. B&H, LaVision) use 65,536.
For 12-bit time gated data use 4,096.

### Background tab¶

The settings in this tab are used to set the background subtraction. These options subtract a background from the data before fitting.

Tip

A background due to detector dark noise or stray room light should be added to the model rather than being subtracted, otherwise the fit will be weighted incorrectly. See Dealing with backgrounds for more information.

Background The type of background to subtract:

 None No background value Single Value A constant value, e.g. camera offset Image A background image, subtracted uniformly from every time bin or gate. Use to correct for a non-uniform background due to, e.g. camera non-uniformities TV Image A time varying background (TVB) image. Use to account for time and spatially varying background fluorescence, e.g. due to plate fluorescence with non-uniform illumination

Background Value The background value subtracted if Single Value is selected in Background.

### IRF tab¶

The settings in this tab relate to the Instrument Response Function (IRF) and how it is processed. See Instrument Response Function for more information.

IRF Type The type of IRF loaded

 Scatter An IRF recorded with a scattering sample or raman signal. Reference A mono-exponential decay recorded with a reference dye.

Reference Lifetime The lifetime of the reference decay. This value will only be used if IRF type is set to Reference.

Background The level background present in the IRF. The background signal level is estimated automatically after loading, see Estimate IRF Background option in the IRF menu.

BG is Afterpulsing. Whether the background in the IRF is due to afterpulsing. If so, the background value is used to compensate for the afterpulsing, otherwise the background value is subtracted from the IRF.

Time Min. and Time Max.` The limits of the valid FLIM signal. Time bins or gates outside of these values will be removed from the IRF.

IRF Shift Shift in picoseconds to apply to the IRF.

G Factor The relative sensitivities of the parallel and perpendicular detector channels for polarisation resolved data.