DE102021107759A1 - Method for determining luminescence lifetimes of a luminescent sample - Google Patents

Abstract

The invention relates to a method for determining luminescence lifetimes of a luminescent sample, the method comprising at least a first and a second method section, with at least the following steps being carried out in the first method section: - detecting a first time-correlated luminescence measurement on a first sample, the first sample has a plurality of luminescence lifetimes;- determining a first decay behavior of the luminescence from the first measurement of the first sample;- determining a first model which is adapted to the first decay behavior with a first fitting method; wherein the second method section comprises at least the following steps:- Acquiring a second time-correlated luminescence measurement on a second sample, which can include the first sample or the first samples;- Determining a second decay behavior of the luminescence from the second measurement of the second sample;- Adjusting a Overall model to the second decay behavior, wherein the overall model includes a linear combination of the first model and a second model, wherein the linear combination for the first model includes a first model associated with the first coefficient of the linear combination and a second coefficient of the linear combination includes the second model is assigned, wherein for the adjustment, the first coefficient and the second coefficient are varied with a second adjustment method in such a way that the overall model is adjusted to the second decay behavior.

Description

The invention relates to a method for determining luminescence lifetimes of a luminescent sample. In the context of this specification, fluorescence decay times in particular are also referred to more generally as luminescence lifetimes below. The method provides for determining the luminescence lifetime of a luminescent species and its proportion in a sample which has at least one other species with at least one other, different luminescence lifetime.

For example, in the case of a FRET (Förster Resonance Energy Transfer) study, a sample may contain different proportions of a first species that does not participate in the FRET process and a second species whose luminescence lifetime is different at different locations in the sample is changed by the FRET process. In this sample, one would like to determine the respective proportions of the two species for the different locations (in order to derive the so-called FRET binding) and for the second species to determine the luminescence lifetimes changed by the FRET process, which takes place at different rates, in particular to derive the so-called FRET - Derive efficiency.

To determine these fractions and the unknown luminescence lifetimes, experiments with pulsed laser excitation and time-correlated single photon detection are performed. A histogram of the photon arrival times is obtained as a result of such a measurement. The decreasing frequency of photons with increasing arrival time is called the decay behavior of the examined sample.

Mathematically, attempts are often made to describe this decay behavior using exponential functions, with the luminescence lifetime in the exponent as a characteristic measure of the decay of the luminescence over time. Phenomenologically or causally, more than one exponential function is often necessary to describe the decay behavior of just a single species with sufficient accuracy. In order to describe the decay behavior of a mixture of different species with a mathematical model, one often needs more than two exponential functions. In addition to the different species, e.g. scattered excitation light and/or the self-luminescence of the sample environment can also contribute to the measured signal and thus make the mathematical description of the measured decay behavior more complex.

Another difficulty in fitting the mathematical model to the decay behavior is the signal-to-noise ratio in the histogram. This can result in small fractions and small differences in the luminescence lifetime of a given species no longer being able to be reliably detected due to the noise in the sample.

The aim of the invention is to provide a method which improves the accuracy for the determination of the unknown proportions and the unknown luminescence lifetimes of a species in the presence of other species, the decay behavior of the latter species being characterized by more than one luminescence lifetime.

This aim is solved by a method according to claim 1, a system according to claim 18 and a computer program according to claim 19.

Advantageous embodiments are disclosed in the dependent claims.

Some terms of the claims are explained below by way of example.

Luminescence is understood to mean in particular the emission of electromagnetic radiation, in particular in the ultraviolet, visible and/or infrared spectral range, i.e. e.g. between 200 nm and 1200 nm wavelength, as a reaction to an in particular optical excitation. Luminescence includes, for example, processes such as fluorescence and phosphorescence, but also the spontaneous emission of photons from colloidal luminescent dyes, such as from quantum dots or nanotubes.

The luminescence lifetime is a characteristic time constant that describes the average time between excitation (e.g. by means of light) of the sample and subsequent emission of electromagnetic radiation.

In the context of the present specification, the term lifetime is also simply used instead of the term luminescence lifetime in order to simplify readability. In the literature, the luminescence lifetime is also referred to as the excited state decay time.

Fluorescence lifetimes typically range from a few hundred picoseconds to a few nanoseconds.

A sample has at least one luminescent species. A luminescent species can include or consist of a luminescent quantum system, for example. There are different types of quantum systems, such as luminescent proteins, organic dyes, colloidal dyes, or other microscopic, luminescent atomic or molecular compounds. Furthermore, the sample can have one or more such luminescent species, in which case the luminescent species can be of the same type or different types and can differ in particular with regard to an excitation and/or emission wavelength.

In particular, a species can also include a FRET pair with a donor and acceptor dye.

The sample may be a non-homogeneous sample in which the proportions of species contained therein may vary over time and/or locally.

Acquiring a time-correlated luminescence measurement refers to the concrete implementation of a luminescence measurement with a time-correlated acquisition of the detected photons or generally the emission intensity or the reading of measurement data from an electronic data storage device containing data comprising such a measurement or simulation. A time-correlated luminescence measurement is to be understood, for example, as a measurement in which an arrival time or a detection time of one or more photons at a detector is measured at least relative to a luminescence excitation event, such as a light pulse, in particular repeatedly. This direct measurement of arrival times is also referred to as a time-correlated measurement in the time domain. Alternatively, a time-correlated measurement can be understood as a measurement in the so-called frequency domain, in which a phase and amplitude difference between a periodically modulated luminescence excitation source and a periodically varying detector signal of the detected luminescence is detected.

The decay behavior of the sample can be determined from such a time-correlated measurement - regardless of the domain in which the measurement is made. If the measurement is in the time domain, the decay behavior is usually presented in the form of a histogram of photon arrival times. With a modulated excitation of the sample, the decay behavior is determined in a different way from the measurement data obtained from it. The method according to the invention can be applied both to time-correlated measurements in the time domain and in the frequency domain.

The decay behavior of a sample can be recorded in different, in particular spectrally separated, channels. The channels can, for example, be set up both for excitation of the sample, e.g. by light of different laser wavelengths, and for detection, e.g. by separating the collected luminescence light into different spectral ranges. In addition, the decay behavior can also be recorded in other optical channels, in which the decay behavior is recorded, e.g. as a function of the polarization of the excitation light used or the polarization of the collected luminescence light. The decay behavior can be recorded and evaluated for one or more such channels.

A model in the context of the specification can be a representation or a description of the decay behavior of the sample in which various physical and/or empirical assumptions are combined. For this purpose, the model can be present as a parametric model or as an empirically determined or measured pattern. Ideally, the model describes the decay behavior determined from the measurement as precisely as possible, at least in terms of its appearance.

In particular, when the model is a parametric model, the model can be adjusted by an adjustment method by varying the adjustment parameters of the model. A large number of such fit algorithms are known from the literature, with the adaptation typically being carried out by minimizing or maximizing at least one quality parameter, so that a deviation of the adapted th model is minimized by the determined decay behavior or a course of the adapted model is brought into the greatest possible agreement with the measured decay behavior.

Alternatively or additionally, the model can be determined as a pattern from the decay behavior for an individual species and in particular correspond directly to the detected decay behavior, so that an adjustment of the model with regard to adjustment parameters is not necessary.

In the case of describing the decay behavior with a pattern, the underlying luminescence lifetimes are not necessarily known.

Determining the model corresponds in particular to adapting the model to the determined decay behavior or creating the pattern which corresponds to the determined decay behavior. A user of the method can specify whether the model should be a parametric model or a pattern. If the model is a pattern, provision can be made for eliminating a background signal or a background signal from the pattern, e.g. by suitable subtraction of the background signal from the decay behavior.

• - Erfassen einer ersten zeitkorrelierten Lumineszenzmessung an einer ersten Probe, wobei die erste Probe eine Vielzahl von Lumineszenzlebensdauern aufweist;
• - Ermitteln eines ersten Abklingverhaltens der Lumineszenz aus der ersten Lumineszenzmessung der ersten Probe in mindestens einem spektralen und/oder optischen Kanal;
• - Ermitteln eines ersten Modells, welches dem ersten Abklingverhalten insbesondere in allen spektralen und/oder optischen Kanälen mit einem ersten Anpassungsverfahren angepasst wird oder welches dem ersten Abklingverhalten in Form eines ersten Musters entspricht;

wobei, wenn der erste Verfahrensabschnitt mehrfach durchgeführt wird, der erste Verfahrensabschnitt jeweils mit einer weiteren ersten Probe durchgeführt wird, die sich insbesondere von der einen ersten Probe oder den mehreren ersten Proben aus den vorangegangenen Durchläufen unterscheidet, insbesondere so, dass für jede weitere erste Probe ein weiteres erstes Abklingverhalten und ein weiteres erstes Modell ermittelt wird und mit dem ersten Anpassungsverfahren an das jeweilige weitere erste Abklingverhalten angepasst wird und/oder dem weiteren ersten Abklingverhalten in Form eines weiteren ersten Musters entspricht,
wobei der zweite Verfahrensabschnitt zumindest die folgenden Schritte umfasst:

• - Erfassen einer zweiten zeitkorrelierten Lumineszenzmessung an einer zweiten Probe, welche die erste Probe oder die ersten Proben umfassen kann und insbesondere umfasst;
• - Ermitteln eines zweiten Abklingverhaltens der Lumineszenz aus der zweiten Lumineszenzmessung der zweiten Probe in dem mindestens einen spektralen und/oder optischen Kanal;
• - Anpassen eines Gesamtmodells an das zweite Abklingverhalten insbesondere in allen Kanälen, wobei das Gesamtmodell eine Linearkombination aus jedem ermittelten ersten Modell und einem zweiten Modell umfasst bzw. aus der Linearkombination besteht, wobei die Linearkombination für jedes ermittelte erste Modell einen jeweils diesem ersten Modell zugeordneten ersten Koeffizienten der Linearkombination und einen zweiten Koeffizienten der Linearkombination umfasst, der dem zweiten Modell zugeordnet ist, wobei zum Anpassen des Gesamtmodells an das zweite Abklingverhalten jeder erste Koeffizient und der zweite Koeffizient mit einem zweiten Anpassungsverfahren so variiert werden, dass das Gesamtmodell an das zweite Abklingverhalten angepasst wird;

wobei bei dem Anpassen mit dem zweiten Anpassungsverfahren zusätzlich noch mindestens ein Anpassungsparameter des zweiten Modells anpasst wird, wobei mindestens eine oder mehrere der zweiten Probe zugeordnete(n) Lumineszenzlebensdauer(n) und insbesondere dazu assoziierte Amplitude(n) aus dem mindestens einen Anpassungsparameter des angepassten zweiten Modells ermittelt werden;

• - Insbesondere ermitteln eines Anteils mindestens einer ersten Probe oder von der ersten Probe umfassten Spezies der mindestens einen ersten Probe, der in der zweiten Probe enthalten ist, wobei der Anteil anhand der jeweiligen zugeordneten Koeffizienten der Linearkombination bestimmt wird und insbesondere ermitteln eines Anteils mindestens einer weiteren Spezies, die nur in der zweiten Probe enthalten ist, wobei auch dieser Anteil anhand der jeweiligen zugeordneten Koeffizienten der Linearkombination bestimmt wird.

According to the invention, a method is now provided which is configured for determining luminescence lifetimes of a luminescent sample, the method comprising at least a first and a second method section. The first stage of the process is carried out at least once or several times, with at least the following steps being carried out for each run through of the first stage of the process:

• - acquiring a first time-correlated luminescence measurement on a first sample, the first sample having a plurality of luminescence lifetimes;
• - determining a first decay behavior of the luminescence from the first luminescence measurement of the first sample in at least one spectral and/or optical channel;
• - determining a first model which is adapted to the first decay behavior in particular in all spectral and/or optical channels using a first adaptation method or which corresponds to the first decay behavior in the form of a first pattern;

where, if the first method step is carried out several times, the first method step is carried out in each case with a further first sample, which differs in particular from the one first sample or the several first samples from the previous runs, in particular in such a way that for each further first sample a further first decay behavior and a further first model is determined and is adapted to the respective further first decay behavior with the first adaptation method and/or corresponds to the further first decay behavior in the form of a further first pattern,
wherein the second method section comprises at least the following steps:

• - detecting a second time-correlated luminescence measurement on a second sample, which can include and in particular includes the first sample or the first samples;
• - determining a second decay behavior of the luminescence from the second luminescence measurement of the second sample in the at least one spectral and/or optical channel;
• - Adaptation of an overall model to the second decay behavior, in particular in all channels, the overall model comprising a linear combination of each determined first model and a second model or consisting of the linear combination, the linear combination for each determined first model having a first associated with this first model Coefficients of the linear combination and a second coefficient of the linear combination, which is assigned to the second model, wherein to adapt the overall model to the second decay behavior, each first coefficient and the second coefficient are varied with a second adaptation method in such a way that the overall model is adapted to the second decay behavior becomes;

wherein at least one adjustment parameter of the second model is additionally adjusted during the adjustment with the second adjustment method, with at least one or more luminescence lifetime(s) assigned to the second sample and in particular amplitude(s) associated thereto from the at least one adjustment parameter of the adjusted one second model are determined;

• - In particular, determining a proportion of at least one first sample or of the species included in the first sample of the at least one first sample that is contained in the second sample, the proportion being determined on the basis of the respective assigned coefficients of the linear combination and in particular they determine a proportion of at least one other species that is only contained in the second sample, this proportion also being determined using the respective assigned coefficients of the linear combination.

The invention therefore provides that, in order to increase the reproducibility and accuracy in the evaluation of luminescence lifetime measurements of a second sample, which is expected to have a proportion of a first sample, the first sample is first measured in the absence of the second sample and from this first measurement, for example the luminescence lifetimes as well as the associated amplitudes of the decay components of the first sample can be determined or at least a pattern of the decay behavior of the first sample is created. In the second method section, the second sample can then be measured, which, as expected, has a proportion of the first sample. When evaluating the second measurement, the second model can now be adjusted using the second adjustment method, so that compared to the first sample, newly added luminescence lifetimes of the second sample, in particular due to at least one species only contained in the second sample, can be determined and at the same time by "recording “ From the already determined characteristics of the first sample, e.g. in the form of determined luminescence lifetimes and their amplitude ratios or a pattern, the proportion of these characteristics in the second sample can be determined exactly. This procedure significantly reduces the number of adjustment parameters when adjusting the overall model and thus allows a reliable statement to be made both about the proportion of the first sample in the second sample and, above all, about the proportion and the luminescence lifetimes of the added species in the second sample, which in was not included in the first sample.

Depending on the complexity of the second sample, several first samples can also be measured and these can then be taken into account accordingly via coefficients that have to be adjusted.

In particular, the added species of the second sample are present together with the species or species comprised by the at least first sample.

The method according to the invention allows the use of luminescence dyes which have more than one luminescence lifetime now also for complex samples in which, for example, the decay behavior of at least one species changes as a result of the interaction between the dyes involved.

This enables a significantly improved and more robust evaluation of time-correlated measurements in connection with complex samples.

The term "complex sample" refers in particular to a second sample that is expected to have one or more luminescent components, which in turn can each have different species that can be detected separately in the first step of the method.

This means that the first and the second sample differ in particular in their respective decay behavior. In particular, the first sample and the second sample can be present in the same measurement object, but e.g. at different sample material locations or at different times, so that the first sample can be recorded separately from the second sample.

A luminescent component of the second sample may contain a background signal, such as autofluorescence from a buffer or tissue. Such a component can also have a first species, which is initially measured separately—also in the sense of being spatially separated, but in the same measurement object—as the first sample and already shows a large number of luminescence lifetimes there. In the case of a FRET scenario, this first species of the first sample could comprise FRET donor molecules that do not interact with a FRET acceptor. An additional second species in the second sample can include components of the first species, ie the first sample, which, however, show a change in their decay behavior compared to the first species in the first sample due to an additional interaction. In the case of the FRET scenario, this second species in the second sample comprises donor molecules that exhibit shorter luminescence lifetimes through interaction with FRET acceptors.

The method according to the invention makes it possible to reliably determine the proportions of species of the first sample in the second sample that are present unchanged in the second sample and at the same time to determine proportions of at least one other species in the second sample via the adaptation parameters of the second model that are present in the second sample compared to the first sample.

As already mentioned, the method according to the invention allows a significant reduction in the parameters to be adjusted when adjusting the overall model of the second sample by introducing the linear combination of known and unknown decay behavior, so that a robust and repeatable determination of sample proportions can be carried out.

According to one embodiment of the invention it is provided that at least a first sample of the at least one first sample is contained in the second sample.

In particular, the second sample comprises at least two different species, wherein a first species of the at least two different species is also included in the first sample and a second species of the at least two different species is exclusively included in the second sample and therefore not included in the first sample.

According to the invention, it is possible to carry out the first method step several times and in each case to record a further first sample, which in particular has a further first species, with regard to the decay behavior. For each implementation of the first method section, the method steps listed there are carried out and, in particular, a first model is determined, which is adapted using the adaptation method or which is a pattern corresponding to the decay behavior.

In particular, at least one decay behavior of a first sample has a large number of luminescence lifetimes. Furthermore, it may be the case that when a large number of first samples is detected, at least one decay behavior of a first sample has only one luminescence lifetime.

By performing the first method section multiple times, a large number of first samples can be recorded and a large number of first models can be determined, in particular not all of these first models necessarily having to be either all adapted to the decay behavior by the adaptation method or having to be available as samples. A mixture of patterns and parametric models is quite possible and can have advantages.

In order to increase the comprehensibility and readability in the following description and the claims, instead of the expression one or more first sample(s) only the first sample is referred to, whereby both possible expressions are to be included, viz the first sample, but also the plurality of first samples, in particular comprising the first sample and at least one further first sample. Furthermore and logically, a first model is assigned to each first sample, and therefore in the following and in the claims only the first model can or will be referred to instead of the respective first model or the like.

The previous paragraph applies analogously with regard to further expressions which relate to the first sample and are determined or assigned for each first sample. However, for reasons of clarity, reference may also be made to the one or more first sample(s).

According to the invention, in the second method section in the overall model, all the first models determined are combined with the second model in a linear combination and scaled individually via the respectively assigned coefficients of the linear combination in order to be adapted to the second decay behavior determined. In addition, at least one fitting parameter of the second model is also varied, in particular where this fitting parameter can be a luminescence lifetime of the additional species in the second sample. This additional species is also referred to as the second species in the specification. If only one further luminescence lifetime is determined, an amplitude associated with this luminescence lifetime can correspond to the coefficient of the linear combination. In the case of multiple luminescence lifetimes for the additional species in the second sample, an associated amplitude can be determined for each luminescence lifetime of the second model, overestimating the fitting of the second model due to the coefficient of linear combination and the amplitudes of the luminescence lifetimes should be avoided, e.g. by appropriate Reduction of freely variable amplitudes.

This distinguishes the method according to the invention, for example, from pure so-called “linear decomposition” methods in which only the contributions of individual species are determined using a linear decomposition.

The method according to the invention enables an improved and more reliable determination of the luminescence lifetimes, in particular of the newly added species in the second sample, since the number of adjustment parameters to be varied, e.g. the coefficients of the linear combination and the at least one adjustment parameter of the second model, are used to adjust the overall model the second decay behavior is reduced compared to an adjustment that would be based on fitting all degrees of freedom of a sufficient model to the second decay behavior.

This is achieved in particular by the fact that the first sample - or the large number of first samples - which can be contained in the second sample and which is or are expected to be present to a certain extent in the second sample, are recorded separately beforehand and as a result, their contribution in the second sample can be determined robustly, since for each first sample only the coefficient in the linear combination has to be adjusted - i.e. only one degree of freedom in the overall model, instead of, for example, several parameters for describing the luminescence lifetimes and associated amplitudes alone for each first rehearsal.

This makes it possible to more reliably determine the luminescence lifetimes and variables associated therewith, even in the case of complex samples which, for example, have several dyes and/or in which an autofluorescent background is present.

According to one embodiment of the invention, it is provided that the decay behavior of the first sample and the second sample is determined for more than one spectral and/or one optical channel.

This embodiment leads to a particularly robust variant of the method.

According to a further embodiment of the invention, it is provided that the adaptation of the first model, in particular each first model that is not a pattern, to the respective determined first decay behavior is determined with the first adaptation method in such a way that a deviation between the first model and the respective first Decay behavior is minimized, in particular wherein the first adaptation method comprises an iterative optimization method, in particular wherein the first adaptation method comprises a maximum likelihood (MLE) method or a least squares (least squares) method.

In this way, the first model can be matched as correctly as possible to the determined decay behavior. This applies in particular if the model is not a pattern but has at least one adjustment parameter that can be varied using the adjustment method.

A distance or a distance metric between the detected decay behavior and the model can be used as the deviation.

According to a further embodiment of the invention, it is provided that the adjustment of the overall model is determined with the second adjustment method in such a way that a deviation between the overall model and the second decay behavior is minimized, in particular with the second adjustment method comprising an iterative optimization method, in particular with the second adjustment method a maximum likelihood (MLE) method or a least squares (least squares) method.

According to one embodiment of the invention, it is provided that the adaptation of the overall model is determined using the second adaptation method by varying all adaptation parameters of the second model and the coefficients of the linear combination, so that a deviation between the overall model and the second decay behavior is minimized.

According to one embodiment of the invention, it is provided that the first and the second adaptation method is the same adaptation method.

According to one embodiment of the invention, it is provided that for each first sample, a multiplicity of luminescence lifetimes and associated amplitudes assigned to the respective first sample are determined from the respective first, in particular adapted model.

According to one embodiment of the invention, it is provided that the first model or at least one of the first models comprises a first parametric model which has a large number of luminescence lifetimes and amplitudes associated therewith as adjustment parameters, with the luminescence lifetimes and the associated associated amplitudes of the respective first model are varied such that the respective first model is adapted to the respective first decay behavior, in particular wherein in the second method step the determined luminescence lifetimes and the associated amplitudes of the respective first model are kept constant and the associated amplitudes are kept constant by the second Adaptation method in the second method step with a common scaling factor, in particular with the respective associated coefficient of the linear combination, multiplied.

In this way, the luminescence lifetimes of the first sample and their associated amplitudes can be determined, so that further information can be obtained, e.g. from comparison with the luminescence lifetimes of the second sample. Because the luminescence lifetimes and their associated amplitudes are kept constant in the second method step, the method becomes more robust compared to conventional methods in which the luminescence lifetimes and their associated amplitudes are not kept constant during the adjustment.

It is important here that, according to this embodiment, the amplitude ratios of the decay components of the first sample are also kept constant and not just the determined luminescence lifetimes. In this way, a proportion of the first sample in the second sample can be clearly determined, even if this is possibly zero. This also gives the method an additional increase in reliability and robustness compared to conventional methods.

According to one embodiment of the invention, it is provided that the second model comprises a second parametric model which has one or more luminescence lifetimes and amplitudes associated therewith as fitting parameters, with the second fitting method in addition to each first coefficient for each first sample and the second coefficient for the second model, the luminescence lifetimes and the associated amplitudes of the second model are varied such that the overall model is adapted to the second decay behavior. If the second decay behavior has only one luminescence lifetime, the second coefficient corresponds to the amplitude associated with that luminescence lifetime. In the case of multiple luminescence lifetimes and associated amplitudes, over-determination due to the dependency between the second coefficient and the amplitudes of the second model should be avoided. This can be done in a variety of ways familiar to those skilled in the art.

In this way, further luminescence lifetimes and their amplitudes (or amplitude ratios) of the second sample can be recorded, as a result of which an improved comparison and a more precise evaluation of the information obtained by the method can take place. This embodiment is of great advantage in particular in the case of FRET measurements in which a change in the donor lifetime is expected and this change is included in the calculation of the FRET efficiency. The parametric models allow certain characteristic quantities, such as the luminescence lifetime and its associated amplitude, to be determined with high accuracy.

According to one embodiment of the invention, it is provided that the first model or at least one of the first models and/or the second parametric model comprises a linear combination of exponential functions, the adaptation parameters of the luminescence lifetimes being in the exponents of the exponential functions and the associated amplitudes being adaptation parameters in the respective ones Prefactors of the exponential functions are.

The exponential function of the first model can assume the following simplified form (a convolution of the measurement signal with an instrument response function is not taken into account in this model - however, an extension to this situation is familiar to the person skilled in the art): $$\text{l}{1}_n\left(t\right)={\displaystyle \sum _{m=1}^{M\left(n\right)}A{1}_{n,m}}\text{ex}\left(\frac{-t}{\tau 1_{n,m}}\right)$$

Where I1 _n (t) is the function value of the first model, t represents the time variable, A1 _n,m is the amplitude of the exponential function with the characteristic luminescence lifetime τ1 _n,m , m and n are respectively indices, where the index n refers to the n -te of, for example, N first measurements and m the index of the exponential function when the first model with index n, has M(n) luminescence lifetimes. The characteristic luminescence lifetimes τ1 _n,m and the amplitudes A1 _n,m are the adaptation parameters of the respective first model.

In the same way, the second model can be represented as follows: $$\text{l}2\left(t\right)={\displaystyle \sum _{k=1}^{K}A{2}_k}\text{ex}\left(\frac{-t}{\tau 2_k}\right)$$

Where 12(t) is the functional value of the second model, t represents the time variable, A2 _k denotes the amplitude of the exponential function with the characteristic luminescence lifetime τ2 _k , and k is the index of the coefficient of the linear combination of the exponential function. The characteristic luminescence lifetimes τ2 _k and the amplitudes A2 _k are the adjustment parameters of the second model. In the simplest case of an assumed mono-exponential decay behavior, K = 1.

The overall model can then take the following form: $$G\left(t\right)=\left({\displaystyle \sum _{n=1}^{N}G{1}_n}\cdot \text{I}{1}_n\left(t\right)\right)+G2\cdot \text{I}2\left(t\right)+\text{bg}$$

Where g1 _n and g2 are the coefficients of the linear combination of the overall model. BG can be a time-independent or only slowly changing background contribution to the measurement signal.

It is pointed out again that a possible overdetermination of the model G(t) by the coefficients g1 _n and g2 in connection with the amplitudes A1 _n,m and A2 _k should be avoided. The formula given for the overall model does not take this overdetermination into account. However, the person skilled in the art is familiar with measures to prevent such an overdetermination—eg by reducing the number of freely variable amplitudes or by setting amplitude ratios for the first model and the second model. Other mathematically equivalent formulations, such as the introduction of amplitude ratios within the models, are obvious and unambiguously disclosed to those skilled in the art. The representation of the models chosen in this disclosure primarily serves to illustrate possible embodiments. Actual implementation in a computer may vary and is not intended to be precluded by the mathematical representation chosen here. This applies in particular with regard to additional convolutions with an Instrument Response Function (IRF), which are not mentioned in the mathematical model presented for didactic reasons.

According to one embodiment of the invention, it is provided that the sum of all amplitudes of the first model, in particular of each first model, is normalized to a normalization value, in particular the normalization value being equal to one.

According to one embodiment of the invention, it is provided that the sum of all amplitudes of the second model is normalized to a normalization value, in particular the normalization value being equal to one.

Alternatively, one can also normalize the first and the second model to their respective area integral.

According to one embodiment of the invention, it is provided that the first model or at least one first model of the first models comprises a first pattern, which is determined in particular from the respective first decay behavior, in particular the first pattern corresponds to the first decay behavior, in particular the first pattern corresponds to the first decay behavior minus a time-independent measurement background, so that the measurement background is not scaled when the first pattern is multiplied by the respective first coefficient, in particular the first pattern includes all spectral and/or optical channels of the first decay behavior.

This variant makes it possible to use a pattern instead of a parametric model, so that the determination of the luminescence lifetimes is not necessary. This can be particularly advantageous if no physical assumptions are to be made for the expected decay behavior of the first sample, but instead an empirical approach is desired, for example.

According to one embodiment of the invention, it is provided that the second model comprises a second pattern, which is determined in particular on the basis of the second decay behavior, in particular the second pattern corresponds to the second decay behavior minus the first model, in particular the second pattern contains all spectral and/or optical channels of the second decay behavior.

In this way, a purely empirical evaluation of the data can be carried out.

According to one embodiment of the invention, it is provided that the first and the second decay behavior each comprise at least two spectral and/or optical channels, in particular the channels comprising different wavelength ranges of a detected light and/or an excitation light, or different polarizations of the detected light and/or or the excitation light, in particular the first and the second model being adapted to all channels of the respective decay behavior or the corresponding patterns being determined.

According to one embodiment of the invention, it is provided that in the second method step, when adapting the overall model to the second decay behavior, the second model is assigned a pattern which is part of a pattern family consisting of a large number of different predetermined patterns, and the adaptation to the overall model for each of the patterns of the pattern family is carried out and that pattern is selected as the second model which, together with the first model, achieves the best adaptation to the second decay behavior, in particular wherein at least one luminescence lifetime is assigned to each pattern from the pattern family.

This embodiment allows a particularly fast and efficient adaptation, since the best adaptation to the overall model is determined from a family of predetermined patterns. The number of adjustment parameters is thus minimized. Such an approach allows reliable determination of complex samples in real time. Regardless of this adjustment, each pattern of the family may have associated one or more luminescence lifetime(s) and associated amplitude(s).

According to one embodiment of the invention, it is provided that the first and the second measurement comprise an imaging fluorescence lifetime imaging microscopy (FLIM) measurement, in particular wherein the respective first and the second decay behavior are each derived from at least one sample region of interest (region of interest: ROI) is determined, in particular with the sample region of interest also being able to consist of a single pixel, in particular with the first and/or the second decay behavior being determined in each case from a multiplicity of sample regions of interest.

This embodiment based on ROIs allows a reduction in the signal-to-noise ratio in the determined decay behavior.

According to an embodiment of the invention it is provided that the first and the second measurement are a point measurement, in particular the first and the second measurement being recorded in a fluorescence lifetime spectrometer or by a point measurement with a microscope.

In the context of the specification, a point measurement is understood to mean, in particular, a measurement in which a measurement location may be known, but the detected measurement signal does not carry or imprint or receive any location-resolving information. This means that with a point measurement, information is obtained in particular from a narrowly limited, in particular essentially punctiform or diffraction-limited area.

According to one embodiment of the invention, it is provided that in the second method section the first and the second model are temporally shifted relative to one another during adaptation with the second adaptation method on the time axis, beginning with each new excitation cycle. This embodiment allows for improved matching that can compensate for drift in the excitation signals over time. This can be particularly useful when evaluating measurements that were taken on different days, for example.

According to one embodiment of the invention, it is provided that the first and the second measurement are time-correlated single-photon measurements or time-correlated photon number-resolved measurements.

According to one embodiment of the invention, it is provided that each first and second decay behavior is a histogram or can be represented as a histogram, in particular with each histogram comprising one or more of the spectral and/or optical channels.

• - die zweite Probe eine FRET-Probe umfasst, aufweisend mindestens ein FRET-Paar mit einem Donor und einem Akzeptor, die an einem FRET Prozess teilnehmen, und die erste Probe dem FRET-Paar entspricht, aber der Akzeptor fehlt oder nicht am FRET Prozess teilnimmt,
• - die zweite Probe zumindest eine lumineszierende Komponente umfasst, und die erste Probe einen lumineszierenden oder einen autofluoreszenten Hintergrund umfasst, und/oder
• - die zweite Probe zumindest eine lumineszierende Komponente umfasst, und die erste Probe einen lumineszierenden Hintergrund eines Messvorrichtungsbestandteils umfasst.

According to one embodiment of the invention, it is provided that in the method individually and in combination

• - the second sample comprises a FRET sample comprising at least one FRET pair with a donor and an acceptor participating in a FRET process, and the first sample corresponds to the FRET pair but the acceptor is absent or does not participate in the FRET process ,
• - the second sample comprises at least one luminescent component, and the first sample comprises a luminescent or an autofluorescent background, and/or
• - the second sample comprises at least one luminescent component, and the first sample comprises a luminescent background of a measuring device component.

• - Mindestens eine Lichtquelle zur gepulsten oder modulierten optischen Anregung einer lumineszenten Probe;
• - Mindestens ein Detektor, der dazu eingerichtet ist, die Lumineszenz zeitkorreliert zur Anregung und zeitaufgelöst zu detektieren und ein Detektorsignal auszugeben;
• - Eine Aufnahmeeinrichtung, die aus dem Detektorsignal und einem Synchronisationssignal, das zumindest eine Wiederholrate oder Modulationsfrequenz der Lichtquelle umfasst, eine digitale elektronische Datenstruktur zu generieren, aus der ein Abklingverhalten der detektierten Lumineszenz bestimmt werden kann;
• - Eine Auswerteeinrichtung, die dazu eingerichtet ist, das Abklingverhalten aus der Datenstruktur zu erzeugen und eine Anpassung eines Modells mit einem Anpassungsverfahren an das Abklingverhalten auszuführen.

According to a further aspect of the invention, a system for the time-resolved and time-correlated measurement of luminescence is provided, having at least the following components:

• - At least one light source for pulsed or modulated optical excitation of a luminescent sample;
• - At least one detector which is set up to detect the luminescence in a time-correlated manner to the excitation and in a time-resolved manner and to output a detector signal;
• - A recording device, which generates a digital electronic data structure from the detector signal and a synchronization signal, which comprises at least one repetition rate or modulation frequency of the light source, from which a decay behavior of the detected luminescence can be determined;
• - An evaluation device that is set up to generate the decay behavior from the data structure and to carry out an adaptation of a model with an adaptation method to the decay behavior.

A computer program or a computer program product is also provided according to the invention, comprising instructions which cause the system to carry out the method steps according to one of the embodiments.

Furthermore, a computer-readable storage medium is provided, on which the computer program is stored.

The invention is illustrated below using an exemplary description. The embodiments mentioned there are only disclosed for illustrative purposes. Features or feature combinations contained therein can serve as the basis for further claim features.

In the following, a TCSPC (time-correlated single photon counting) system is assumed, which determines the decay behavior of a luminescent sample in the time domain.

The combination of models that contain patterns and models that are parametric models that are fitted to the decay behavior by varying the fitting parameters can offer advantages in several application areas.

The complete consideration and prior determination of the decay behavior of a donor dye in time-resolved FRET measurements can allow an improved, i.e. more accurate and more robust determination of the FRET efficiency, especially in the case of donor dyes with multiple exponential decay.

The term time-resolved in the context of the specification refers to time-correlated measurements that allow the determination of a decay behavior, in the case of a TCSPC measurement based on the creation of a histogram.

A combination of a model in the form of a pattern for the donor dye, where the pattern is determined from the first measurement on a donor-only sample, i.e. the first sample that has no interacting acceptor dye, and a mono- or multi-exponential Fit, so one two The second model, in the form of a parametric model, is used to determine a FRET efficiency and a binding (relative fraction of functional FRET species in the second sample) for the species newly introduced in the second sample.

Another application of the invention can be the consideration of autofluorescence, so that the "actual" sample signal, ie the sample signal of interest of the second sample, which is superimposed by the autofluorescence, can be determined.

An autofluorescence pattern determined in the first measurement of the first sample as a first model is then used in the analysis of the second measurement and allows the determination of the luminescence lifetime of the second model, which corresponds to the signal of the actual species. In this case, the number of luminescence lifetimes and their amplitudes of autofluorescence are of little interest and it is advantageous to take an empirical approach here in the form of a template for the first model describing autofluorescence.

Both applications can also be combined with each other in order to consider both the autofluorescence and the donor-only sample in a time-resolved FRET measurement when evaluating the measurement data. To do this, a first pattern would be determined from the measured decay behavior in a first measurement on a first sample that only has autofluorescence, and another first pattern would be determined by means of a further first measurement on a further first sample that has the donor-only sample , parametric model, so that the luminescence lifetimes and their amplitudes are determined for this sample. In the second measurement, the FRET portion can then be determined based on the decay behavior of the second sample, which, in addition to active FRET pairs, has both autofluorescence and a portion of donor-only sample, by varying the coefficients of the linear combination of the first model and the second Model including its adjustment parameters are determined.

The evaluations can be imaging or only refer to certain areas of the sample (regions of interest: ROI).

When adapting the overall model, the pattern can be scaled with regard to its amplitude by means of a linear transformation and, in particular, can also be shifted in time with regard to the excitation pulse.

In FRET experiments with donor dyes, whose decay behavior is multi-exponential even in the absence of an acceptor ("donor-only"), i.e. has a large number of luminescence lifetimes, the approaches typically used up to now do not calculate correct FRET efficiencies and fractions of intact FRET -pairs (also referred to as binding) in the examined sample material. A species that has at least one donor and at least one acceptor dye is referred to as a FRET pair.

In the prior art, in the simplest case, a binding of 100% is typically assumed as the initial situation, but this is often not the case.

In most FRET experiments, there is a proportion of non-functioning FRET pairs in the sample material. In this proportion, the donor-only species should be mentioned in particular, whose FRET donor fluoresces but is not quenched by the FRET acceptor. For example, the corresponding molecule which has a donor dye can be regarded as a species.

If these donor-only species are not taken into account when determining the FRET efficiency, as is predominantly the case in the prior art, this FRET efficiency is often significantly underestimated.

If the donor-only species is already taken into account, it is currently only possible with the approaches used to correctly determine the binding for mono-exponential decay behavior of the donor-only species. In most cases, however, this requirement is not met due to the choice of dye or the direct environment of the dye.

Being able to use a multi-exponential donor dye in FRET experiments and correctly calculate both binding and FRET efficiency would be a major advance.

An embodiment of the invention is described below using a FRET sample as an example, but it can also be used for other, complex samples.

A problem in the evaluation of FRET experiments is the large number of adjustment parameters that would be necessary to determine the decay behavior of a portion of a multi-exponential donor-only sample and a portion of a FRET sample, i.e. a sample that has a donor dye , which interacts with the acceptor dye, can be described reliably.

The invention now makes it possible to reduce the number of adaptation parameters in that the complete information on the decay behavior of the first sample, in this example the donor-only sample, is recorded and determined separately, ie in the absence of the FRET species.

In the following, it is assumed for purposes of illustration that the donor-only sample has a bi-exponential decay behavior, i.e. two characteristic luminescence lifetimes.

In a first variant of the invention, the first decay behavior of the first sample, ie the donor-only sample, is determined in the first measurement. After the measurement, this first decay behavior is available, for example, in the form of a histogram of photon arrival times, which shows the time course of the decay behavior of the first sample after pulsed excitation.

By fitting a first parametric exponential function model to the first decay behavior, two luminescence lifetimes τ1 ₁ and τ1 ₂ and two amplitudes of the exponential decays A1 ₁ and A1 ₂ associated with the luminescence lifetimes are determined as fitting parameters of the first model. In addition, the statistical uncertainties of the respective adjustment parameters can also be determined.

The first parametric model M1 (see also 1 ) can be represented as follows: $$\text{I}1\left(t\right)={\displaystyle \sum _{m=1}^{2}A{1}_m}\text{ex}\left(\frac{-t}{\tau 1_m}\right)=A{1}_1\text{ex}\left(\frac{-t}{\tau 1_1}\right)+A{1}_2\text{ex}\left(\frac{-t}{\tau 1_2}\right)$$

In a second measurement, the decay behavior of a FRET sample material is measured that has an unknown proportion of the donor-only sample and an unknown proportion of another FRET species. The FRET sample material with donor-only sample and the unknown proportion of FRET species thus represents the second sample in this context.

Assume now that the FRET sample exhibits a mono-exponential decay behavior that can be characterized by a further luminescence lifetime. The decay behavior of the second sample is evaluated as follows.

An overall model is provided consisting of a linear combination of the first model for the donor-only sample and a second model for the FRET species. The coefficients of the linear combination are g1 for the coefficient associated with the first model and g2 for the coefficient associated with the second model.

In the overall model, the adjustment parameters τ1 ₁ and τ1 ₂ as well as A1 ₁ and A1 ₂ of the first model are fixed and are not varied.

As mentioned, the second model in this example is a mono-exponential parametric model with the fitting parameters τ2 for the luminescence lifetime and A2 for the associated amplitude. The second model describes the FRET species with interacting donor and acceptor dye. It should be noted that the coefficient g2 has a dependence on the amplitude of the second model. The amplitude A2 is therefore not explicitly taken into account in the following and is in particular set equal to one.

The overall model MG (see also 1 ) can therefore be written as follows: $$\begin{array}{l}G\left(t\right)=G1\cdot \text{I}1\left(t\right)+G2\cdot \text{ex}\left(\frac{-t}{\tau 2}\right)\\ \text{=}G\text{1}\left(A{1}_1\cdot \text{ex}\left(\frac{-t}{\tau 1_1}\right)+A{1}_2\cdot \text{ex}\left(\frac{-t}{\tau 1_2}\right)\right)+G2\cdot \text{ex}\left(\frac{-t}{\tau 2}\right)\end{array}$$

Instead of the fitting parameters τ1 ₁ and τ1 ₂ as well as A1 ₁ and A1 ₂ , only one parameter is varied with respect to the first model when fitting the overall model MG to the decay behavior of the second measurement, namely the coefficient g1, which determines the proportion of the donor-only sample im FRET sample material describes, i.e. the second sample. The parameters that are adjusted in the overall model are g1, g2, τ2.

In this way, the proportion of the first sample can be described with a single fitting parameter, g1, when fitting the overall model.

In the overall model, the coefficients g1 and g2 as well as the fitting parameter of the second model, τ2, which describes the decay behavior of the FRET species, are adapted to the decay behavior from the second measurement. The coefficients give information about the proportions of the donor-only species and the FRET species in the FRET sample material as the second sample.

sowie die FRET Effizienz von ausschließlich der funktionsfähigen FRET-Spezies, bestimmt aus τ2 sowie der mittleren Lebensdauer der ersten Probe.This allows the binding B to be calculated with $$B=\frac{G2}{G1+G2}$$

and the FRET efficiency of only the functional FRET species, determined from τ2 and the mean lifetime of the first sample.

In a second variant of the method, instead of a parametric first model, with the adaptation parameters τ1 ₁ and τ1 ₂ as well as A1 ₁ and A1 ₂ , a pattern is used that is determined from the decay behavior of the first measurement. The pattern can be identical to the first decay behavior or it can be a processed version of the decay behavior. A processing step can be, for example, a signal background elimination, for example of components in the decay behavior that have no features correlated with the excitation pulse. Like the first decay behavior, the pattern can be an averaged pattern/decay behavior that is determined from several ROIs or first samples.

In the second part of the method, only the coefficient g1 for the first model, ie the sample, is then determined for the first sample.

An optional adjustment parameter "Shift" in the overall model can correct a time offset between the decay behavior from the second measurement and the sample. As with the first variant, this second variant also has only one additional fit parameter, g1, which can be used to determine the proportion of the donor-only sample, i.e. the first sample, in the FRET sample material, i.e. the second sample to determine.

The binding can be calculated from the coefficients of the linear combination g1 and g2, as shown in the first variant.

In general, a background BG can also be included in the overall model as an optional further adjustment parameter.

The number of free adjustment parameters is then no higher than in the known approach to FRET evaluation, in which the donor is always assumed to be monoexponential.

If a bi-exponential parametric model is used for the FRET species in the second model, there are two more free fitting parameters in the overall model, namely a second luminescence lifetime and a parameter that describes the associated amplitudes or the amplitude ratios of the two exponential functions of the second model considered. Here, too, the background BG can optionally be included in the overall model.

Starting from the coefficients of the linear combination, the binding can then in turn be calculated, and starting from the luminescence lifetimes of the first and second model, the FRET efficiency.

In the case of an imaging, i.e. spatially resolved, recording of the decay behavior, the binding and/or the FRET efficiency calculated in this way can be used as a gray scale and/or color value, so that a spatially resolved representation of the FRET efficiency and/or the binding is achieved.

The shift parameter of the donor-only pattern, an offset of the instrument response function (IRF) and a background parameter BG in the overall model can in turn be taken into account as further adjustment parameters.

The method according to the invention can be expanded by adding a large number of first models. The first models to consider are, for example: an autofluorescence pattern or other fluorescent dyes. In this way, the fluorescence parameters of other fluorescent species in the second sample can be included or taken into account in the evaluation.

A third variant of the method provides that the second model is a pattern that is selected from a predefined pattern family MF. The patterns of the pattern family MF can be assigned different decay behaviors and/or luminescence lifetimes. When adapting the overall model, the patterns of the pattern family are assigned to the second model individually (or also in combination) and the coefficients of the linear combination are varied. The luminescence lifetime(s) of the pattern(s) of the pattern family that leads to the best fit of the overall model are used for the further calculations. The best fit can be determined, for example, by evaluating the residuals and/or a quality parameter X ² .

This approach has the advantage that the parameter space is greatly reduced, which leads to a reduction in computing time and computing load and is therefore particularly advantageous for real-time applications. The selection from predetermined patterns of the pattern family thus saves the explicit calculation of the model function when adjusting the overall model during the adjustment process.

So, as an example, a pattern family with 11 patterns could now be used for the FRET probe, whose decay behavior has a mean decay time between 2 ns and 3 ns with intervals of 0.1 ns.

When adapting the overall model to the decay behavior of the FRET sample material, the coefficients g2, g1 would be varied for each pattern in the pattern family.

The overall model for the jth pattern Pj of the pattern family could therefore be described as follows: $$\text{G}\left(t\right)=G1\cdot \text{I}1\left(t\right)+G2\cdot \text{P}j$$

It only plays a minor role whether the first model is also a pattern or a parametric model.

The fit parameter of the second model can thus refer to an index indexing the family members of the model family.

The third variant is particularly advantageous if the decay behavior of the first and the second sample extends over a large number of spectrally different channels.

• 1 zeigt schematisch eine erste Ausführungsform des erfindungsgemäßen Verfahrens;
• 2 zeigt schematisch eine zweite Ausführungsform des erfindungsgemäßen Verfahrens;
• 3 zeigt eine schematische Darstellung eines mehrdimensionalen Abklingverhaltens; und
• 4 zeigt eine schematische Darstellung einer Musterfamilie

An exemplary description of some embodiments of the method according to the invention is given below using a description of the figures.

• 1 shows schematically a first embodiment of the method according to the invention;
• 2 shows schematically a second embodiment of the method according to the invention;
• 3 shows a schematic representation of a multi-dimensional decay behavior; and
• 4 shows a schematic representation of a pattern family

In 1 a schematic sequence of the method according to the invention is shown according to an exemplary embodiment. In a first method section V1 (upper panel), a first decay behavior 10 is determined in a first time-resolved luminescence measurement, for example a TCSPC measurement, on a first luminescent sample P1. The first decay behavior 10 is in 1 in the form of a schematic histogram with a logarithmic I(t) axis and a linear t-axis. Depending on the quality of the measurement, the decay behavior 10 can be noisy to varying degrees.

In a next step, a first model M1 is then adapted to the decay behavior 10 using a first adaptation method. As already explained, this can be done, for example, using a parametric exponential model, in which the luminescence lifetimes and their associated amplitudes are varied in the form of adjustment parameters until the adjustment process is complete, e.g. because a quality indicator has been optimized with regard to the adjustment to the decay behavior. According to the invention, the first sample P1 has a decay behavior with more than one luminescence lifetime.

In a second method section V2 (second panel), a second time-resolved measurement is carried out on a second luminescent sample P2, which, as expected, includes an unknown proportion of the first sample P1. A second decay behavior 20 is determined from this measurement.

In a next step, an overall model MG is adapted to second decay behavior 20 . The overall model includes the first model M1 and a second model M2, which are combined in the form of a linear combination to form the overall model MG.

The overall model MG is adapted by varying the coefficients g1 and g2 of the linear combination and by varying at least one further adaptation parameter of the second model M2, which can be the luminescence lifetime of a mono-exponential luminescence decay, for example.

It is important that the adaptation parameters of the first model are not further varied in the second method section, but rather are kept at the values that were previously determined in the first method section by the first adaptation method.

In this way, the method according to the invention allows proportions of a first sample in a second sample to be reliably determined by two-stage measurement with a small number of adjustment parameters.

In 2 is an extension of the procedure 1 shown. In this example, two first samples P11 and P12 are measured, which are contained in a second sample P2 of unknown composition.

At least one of the first samples P11, P12 has a decay behavior 11, 12 that includes a large number of luminescence lifetimes.

In the example in 2 the first model M11 for the first sample P11 is a parametric model, which is adapted to the first decay behavior 11 using the first adaptation method. The additional first model M12 of an additional first sample P12 is a pattern that essentially corresponds to the decay behavior 12 of the additional first sample P12. This can be advantageous if the determined decay behavior is, for example, autofluorescence, which experience has shown has a complex decay behavior.

In the second method section V2, the second sample P2 is measured, which has unknown proportions of the two first samples P11, P12. Correspondingly, the overall model MG contains a linear combination with three summands, namely the first two models M11, M12 and the second model M2. In the second adjustment process, the three coefficients assigned to the respective models are now varied, as well as at least one adjustment parameter of the second model M2, until the second adjustment process has ended, e.g. because a quality indicator reflects an optimum of the adjustment.

In this way, proportions of a large number of first samples P11, P12 with different dyes and/or decay behavior in a second sample P2 can be determined without the second adjustment method delivering unreliable results due to too many free adjustment parameters, since the Number of parameters to be adjusted has already been reduced by carrying out the first method step V1.

The in the 1 and 2 The decay behaviors shown schematically are simple histograms representing, for example, a single spectral channel of the decay behavior. However, within the meaning of the invention, it is also provided that the decay behavior can relate to a large number of spectral and/or optical channels.

The temporal distribution or assignment of the detected luminescence relative to excitation pulses of different wavelengths that are offset in time can also be evaluated as a spectral channel. This type of channel formation is also called Pulsed Interleaved Excitation (PIE).

In 3 a three-dimensional decay behavior 30 is shown by way of example, which extends over a multiplicity (specifically 3 in this figure) of optical channels ch. As a possible embodiment, the invention provides that a decay behavior 30 can be detected in this multidimensional form. The models and adjustment methods are then also designed accordingly, so that all channels are adjusted at the same time, possibly with different adjustment parameters.

The ch axis can, for example, refer to a detection wavelength of the luminescence and thus divide the decay behavior into optical channels on which the decay behavior can be represented as a two-dimensional decay behavior 31, 32, 33, and 34. It can be seen here that the decay behavior 30 in the central channel also has a temporal distribution within a spectral channel (32, 33). This can be useful, for example, due to a further temporally offset excitation pulse (Pulsed-Interleaved-Excitation: PIE) with a different wavelength. The entirety of the two-dimensional decay behavior 31, 32, 33, 34 in a multi-dimensional detection space is generally referred to as the decay behavior 30 (analogously also 10, 11, 12, 20) in the context of the invention.

4 . Shows a schematic representation of a pattern family MF. The pattern family MF includes a large number of patterns MF1, MF2, MF3, each of which is associated with at least one characteristic variable, such as an index, a luminescence lifetime and/or another characteristic.

In the example shown, a pattern family MF is shown whose patterns MF1, MF2, MF3 have successively longer average luminescence lifetimes.

As already described, these patterns can be used as a second model. However, use as the first pattern is also conceivable.

Claims (19)

Method for determining luminescence lifetimes of a luminescent sample, the method comprising at least a first and a second method section (V1, V2), the first method section (V1) being carried out at least once or several times, with each run-through of the first method section (V1) at least the following steps are carried out: - detecting a first time-correlated luminescence measurement on a first sample (P1, P11), - determining a first decay behavior (10) of the luminescence from the first measurement of the first sample (P1, P11); - Determining a first model (M1, M11) which is adapted to the first decay behavior (10,11) using a first adaptation method or which corresponds to the first decay behavior (10,11) in the form of a first pattern; where if the first method section (V1) is carried out several times, the first method section is carried out in each case with a further first sample (P12), so that for each further first sample (P12) a further first decay behavior (12) is determined and a further first Model (M12) is determined and is adapted to the respective further first decay behavior (12) with the first adaptation method and/or corresponds to the further first decay behavior (12) in the form of a further first pattern, with at least one first sample (P1, P11, P12) has a variety of luminescence lifetimes; wherein the second method section (V2) comprises at least the following steps: - acquiring a second time-correlated luminescence measurement on a second sample (P2), which can comprise the first sample (P1) or the first samples (P11, P12); - Determining a second decay behavior (20) of the luminescence from the second measurement of the second sample (P2); - Adapting an overall model (MG) to the second decay behavior (20), the overall model comprising a linear combination of each first model (M1, M11, M12) and a second model (M2), the linear combination for each first model (M1, M11, M12) comprises a first coefficient (g1) of the linear combination associated with each of these first models (M1, M11, M12) and a second coefficient (g2) of the linear combination associated with the second model (M2), wherein each first coefficient (g1) and the second coefficient (g2) are varied with a second adaptation method in such a way that the overall model (MG) is adapted to the second decay behavior (20); wherein, during the adjustment with the second adjustment method, at least one adjustment parameter of the second model (M2) is additionally adjusted, so that one or more luminescence lifetime(s) assigned to the second sample and one or more amplitude(s) associated thereto from the adapted second model (M2) are determined. procedure after claim 1 , characterized in that the adaptation of the first model (M1, M11, M12) to the first decay behavior (10, 11, 12) is determined with the first adaptation method such that a deviation between the first model (M1, M11, M12) and the first decay behavior (10, 11, 12) is minimized, in particular wherein the first adaptation method comprises an iterative optimization method, in particular wherein the first adaptation method comprises a maximum likelihood (MLE) method or a least squares (least squares) method. procedure after claim 1 or 2 , characterized in that the adaptation of the overall model (MG) is determined using the second adaptation method in such a way that a deviation between the overall model (MG) and the second decay behavior (20) is minimized, in particular the second adaptation method comprising an iterative optimization method, in particular wherein the second fitting method comprises a maximum likelihood (MLE) method or a least squares (least squares) method. Method according to one of the preceding claims, characterized in that for each first sample (P1, P11, P12) a number of luminescence lifetimes and associated amplitudes assigned to the respective first sample is determined from the respective first model (M1, M11, M12). Method according to one of the preceding claims, characterized in that the first model (M1, M12, M12) comprises a first parametric model (M1, M11) which has a large number of luminescence lifetimes and amplitudes associated therewith as adaptation parameters, wherein in the first method section with the luminescence lifetimes and the associated amplitudes of the first model (M1, M11) are varied in the first adaptation method in such a way that the first model (M1, M11) is adapted to the first decay behavior, in particular with the determined luminescence lifetimes and the associated amplitudes in the second method section of the first model (M1, M11) are kept constant and the associated amplitudes are multiplied by the second adaptation method in the second method section by a common scaling factor, in particular by the first coefficient (g1) of the linear combination. Method according to one of the preceding claims, characterized in that the second model (M2) comprises or consists of a second parametric model which has one or more luminescence lifetimes and amplitudes associated therewith as adaptation parameters, with the second adaptation method additionally to each first coefficient ( g1) and the second coefficient (g2), the luminescence lifetimes and the amplitudes associated therewith of the second model are varied in such a way that the overall model is adapted to the second decay behavior. Method according to one of the preceding claims, characterized in that the first and / or the second parametric model (M1, M11, M2) comprises a linear combination of exponential functions, the luminescence lifetimes as adjustment parameters in the exponents of the exponential functions and the associated amplitudes as adjustment parameters form the respective prefactors of the exponential functions. Method according to one of the preceding claims, characterized in that at least one first model (M12) comprises a first pattern which is determined in particular from the first decay behavior (12), in particular wherein the first pattern (12) corresponds to the first decay behavior, in particular wherein the first pattern corresponds to the first decay behavior minus a measurement background, so that when the first pattern is multiplied by the first coefficient, the measurement background does not is scaled with, in particular wherein the first pattern includes all spectral and / or optical channels of the first decay behavior. Method according to one of the preceding claims, characterized in that the second model (M2) comprises a second pattern which is determined in particular on the basis of the second decay behavior, in particular the second pattern corresponds to the second decay behavior minus the first model, in particular the second pattern includes all spectral and/or optical channels of the second decay behavior. Method according to one of the preceding claims, characterized in that the first and the second decay behavior (10, 11, 12, 20) each comprises at least two spectral and/or optical channels, in particular the channels having different wavelength ranges of a detected light and/or a Include excitation light, or include different polarizations of the detected light and / or the excitation light, in particular the first and the second model are adapted to all channels of the respective decay behavior. Method according to one of the preceding claims, characterized in that in the second method section, when adapting the overall model (MG) to the second decay behavior (20), the second model (M2) is assigned a pattern (MF1, MF2, MF3) which is part of a pattern family (MF) consisting of a plurality of different predetermined patterns (MF1, MF2, MF3), and wherein the adaptation to the overall model (MG) is performed for each of the patterns (MF1, MF2, MF3) of the pattern family (MF) and that Pattern (MF1, MF2, MF3) is selected as the second model (M2), which together with the first model (M1, M11, M12) achieves the best adaptation to the second decay behavior (20), in particular with each pattern (MF1, MF2 , MF3) from the pattern family (MF) is assigned at least one luminescence lifetime. Method according to one of the preceding claims, characterized in that the first and the second measurement comprises an imaging fluorescence lifetime microscopy measurement (FLIM), in particular wherein the (respective) first and the second decay behavior (10, 20) each consist of at least one sample area of Interest (region of interest (ROI)) is determined, in particular wherein the sample region of interest can also consist of a single pixel, in particular wherein the first and/or the second decay behavior is determined from a plurality of sample regions of interest. Method according to one of the preceding claims, characterized in that the first and the second measurement are a point measurement, in particular the first and the second measurement being recorded in a fluorescence lifetime spectrometer or by a point measurement with a microscope. Method according to one of the preceding claims, characterized in that in the second method section the first and the second model (M1, M11, M12, M2) are temporally shifted relative to one another during adaptation with the second adaptation method. Method according to one of the preceding claims, characterized in that the first and the second measurement are time-correlated single-photon measurements or time-correlated photon number-resolved measurements. Method according to one of the preceding claims, characterized in that each first and second decay behavior is a histogram (31, 32, 34, 33) or can be represented as a histogram, in particular each histogram showing one or more of the spectral and/or optical channels (ch) includes. Method according to one of the preceding claims, characterized in that in the method both individually and in combination - the second sample comprises a FRET sample, having at least one FRET pair with a donor and an acceptor, which participates in the FRET process and the the first sample corresponds to the FRET pair, but the acceptor is absent or does not participate in the FRET process, - the second sample comprises at least one luminescent component, and the first sample comprises a luminescent or an autofluorescent background, and/or - the second sample comprises at least one luminescent component, and the first sample comprises a luminescent background of a measuring device component. System for the time-resolved and time-correlated measurement of a decay behavior of luminescence, having at least the following components: - At least one light source for pulsed or modulated optical excitation of a luminescent sample; - at least one detector which is set up to detect the luminescence in a time-correlated manner for the excitation and in a time-resolved manner and to output a detector signal; - a recording device that is set up to generate a digital electronic data structure from the detector signal and a synchronization signal, which comprises at least one repetition rate or modulation frequency of the light source, from which a decay behavior of the detected luminescence can be determined, - an evaluation device which is set up to generate the decay behavior from the data structure and to carry out an adaptation of a model with an adaptation method to the decay behavior. Computer program or a computer program product comprising instructions that cause the system of claim 18 the method steps according to one of Claims 1 until 17 executes

Images