A method and system for estimating parameters of FRET sensitized channel super-resolution SIM image reconstruction

Abstract

The application discloses a FRET sensitized channel super-resolution SIM image reconstruction parameter estimation method and system. The method comprises the following steps: obtaining a FRET three-channel original image data set and an optical transfer function OTF after structure light modulation; obtaining a FRET three-channel spread spectrum image by using Fourier transform and spectrum shift, and obtaining reconstruction parameters of DD, DA and AA channel image sets from the FRET three-channel spread spectrum image; replacing the structure light illumination frequency vector of the DA channel with the structure light illumination frequency vector in the reconstruction parameters of the DD channel image set; obtaining the initial phase and modulation depth of the DA channel image set by using a complex linear regression algorithm containing empirical mode decomposition; performing FRET three-channel structure light super-resolution linear Wiener reconstruction, and performing data processing based on a channel sensitization intensity measurement method to measure the FRET efficiency and the donor-acceptor concentration ratio of the FRET super-resolution image. The application enhances the robustness of the SIM-FRET method for low signal-to-noise ratio image reconstruction, and expands the usability of the SIM-FRET under extremely low FRET efficiency.

Description

Technical Field

[0001] This invention belongs to the technical field of fluorescence resonance energy transfer (FRET) detection technology and structured light excited super-resolution imaging (SIM), specifically relating to a method and system for estimating reconstruction parameters of FRET-sensitized channel super-resolution SIM images. Background Technology

[0002] Fluorescence resonance energy transfer (FRET) detection is currently the only technique capable of long-term, real-time detection of dynamic interactions and spatial distribution between biomolecules in a single living cell. Theoretically, it can quantitatively analyze protein-molecule interactions at the 10 nm scale in living cells using microscopic imaging. The development of super-resolution structured light microscopy (SR-SIM) has made it possible to observe subcellular spatial structures at the 100 nm scale with greater precision. SR-SIM offers low illumination intensity, fast imaging speed, and multicolor imaging capabilities. The fusion of SR-SIM and FRET technologies provides a new approach for achieving dynamic super-resolution quantitative SIM-FRET imaging in living cells. However, during FRET quantitative analysis, the receptor-sensitized channel (DA channel) image contains a large amount of Gaussian and Poisson noise due to the insufficient number of sensitized fluorescent molecules. This leads to low parameter estimation accuracy and numerous reconstruction artifacts during super-resolution image reconstruction, thus affecting quantitative FRET analysis.

[0003] To improve the accuracy of reconstruction parameters, the traditional SR-SIM method, when processing low-resolution images, often uses the same field of view to capture multiple consecutive time-series images, sums and averages them, and uses statistical methods to improve image resolution, thereby increasing the accuracy of parameter estimation and reducing reconstruction artifacts. However, this method has high requirements for exposure time and requires minimal phototoxicity and photobleaching during live-cell imaging. It also cannot guarantee that the averaged image will not produce motion artifacts or photobleaching in the reconstructed image due to cell movement. Therefore, combining SIM and FRET requires more images to accurately calculate the SIM reconstruction parameters. Summary of the Invention

[0004] The main objective of this invention is to overcome the shortcomings and deficiencies of the prior art and provide a method and system for estimating the reconstruction parameters of FRET-sensitized channel super-resolution SIM images.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] One aspect of the present invention provides a method for estimating reconstruction parameters of FRET-sensitized channel super-resolution SIM images, comprising the following steps:

[0007] Acquire a raw image data set of three FRET channels modulated by structured light; the three FRET channels include a donor-excited donor emission channel, i.e., the DD channel, a acceptor-excited acceptor emission channel, i.e., the DA channel, and a donor-excited acceptor emission channel, i.e., the AA channel.

[0008] Obtain the optical transfer function (OTF) of the three channels of FRET;

[0009] Fourier transform is used to obtain the spectral image of the FRET three-channel image, and the broadened spectral image is obtained by spectral shifting;

[0010] Reconstruction parameters for the DD channel image group, DA channel image group, and AA channel image group are obtained from the broadened spectral image.

[0011] Using the forward physical process model of SIM-FRET three-channel imaging, the structured light illumination frequency vector in the reconstruction parameters of the DA channel image group is replaced with the structured light illumination frequency vector in the reconstruction parameters of the DD channel image group in the previous step;

[0012] The initial phase and modulation depth of the DA channel image group are obtained using a complex linear regression algorithm that includes empirical mode decomposition;

[0013] FRET three-channel structured light super-resolution linear Wiener reconstruction was performed based on unified Wiener reconstruction parameters to obtain FRET three-channel super-resolution image data.

[0014] Data processing was performed using a channel sensitization intensity measurement method to measure FRET efficiency and donor-acceptor concentration ratio in FRET super-resolution images.

[0015] As a preferred technical solution, the acquisition of the FRET three-channel raw image data set modulated by structured light specifically involves:

[0016] The three-channel raw image data set based on the channel sensitization intensity measurement method was acquired using the SR-SIM system, including donor excitation and donor emission channel image set I. DD (r), Image group of donor-induced receptor emission channel I DA (r) and donor-induced receptor emission channel image group I AA (r), where r = (x, y) are the spatial coordinates of the fluorescence image pixels. Each set of original image data contains several different structured light direction angles, and each structured light azimuth angle contains several cosine structured light illumination wide-field fluorescence images with different phase differences.

[0017] As a preferred technical solution, the specific method for obtaining the optical transfer function (OTF) of the three FRET channels is as follows:

[0018] The point spread function (FPF) of the three channels of fluorescent microspheres was obtained by imaging with the SR-SIM system. After screening, segmentation, and averaging of multiple images, Fourier transform was performed to obtain the optical transfer function (OTF) of the three channels of FPF: H. DD (k), H DA (k), H AA (k).

[0019] As a preferred technical solution, the step of obtaining the spectral image of the FRET three-channel image using Fourier transform and obtaining the broadened spectral image through spectral shifting specifically involves:

[0020] First, the FRET three-channel image group I DD (r), I DA (r), I AA (r) performs a Fourier transform, I X (k)=FFT(I X (r)) to obtain the spectrum image corresponding to the FRET channel X, where k is the coordinate of the fluorescence image pixel transformed into the frequency domain;

[0021] The spectral image size of the three channels of FRET is expanded from m×n to 2m×2n.

[0022] The stretched FRET three-channel spectrum image was obtained.

[0023] As a preferred technical solution, the reconstruction parameters of the DD channel image group and the AA channel image group are obtained from the broadened spectral image using an iterative cross-correlation algorithm. The reconstruction parameters of the X channel include: the structured light illumination frequency vector. Modulation depth and initial phase

[0024] As a preferred technical solution, the step of using the forward physical process model of SIM-FRET three-channel imaging to replace the structured light illumination frequency vector in the reconstruction parameters of the DA channel image group with the structured light illumination frequency vector in the reconstruction parameters of the DD channel image group in the previous step specifically involves:

[0025] Calculate the structured light illumination frequency vector of the DA channel image Then, the vector distance d between it and the structured light illumination frequency vector in the reconstruction parameters of the DD channel image group is calculated. If d is greater than the set threshold of the magnitude of the DD channel structured light illumination frequency vector, the structured light illumination frequency vector of the DD channel image group is used. Replace the structured light illumination frequency vector of the DA channel image group Right now:

[0026]

[0027] As a preferred technical solution, the method of obtaining the initial phase and modulation depth of the DA channel image group using a complex linear regression algorithm including empirical mode decomposition specifically involves:

[0028] Based on the structured light illumination frequency vector of the obtained DA image group, complex linear regression is calculated on the overlapping region of the DA channel image group to obtain the modulation depth and initial phase of the DA channel image group:

[0029]

[0030]

[0031] s DA =∑(a i *b i ) / |a i | 2

[0032]

[0033]

[0034] Among them, s DA a i b i As an intermediate variable, This is a super-resolution low-frequency band image for the DA channel, where m is the magnitude of the spectral shift, typically m = +1 or -1. This is the m-band image of the DA channel. To The DA channel m-band image after frequency shifting, where the shift distance and direction are the structured light illumination frequency vector. The modulation depth of the structured light illumination frequency vector corresponding to the DA channel. The initial phase of the structured light illumination frequency vector corresponding to the DA channel;

[0035] Before determining the modulation depth and initial phase, empirical mode decomposition is used to separate s. DA The noise and image are extracted, and the intrinsic mode functions are obtained and summed to restore them;

[0036] The empirical mode decomposition method is used to separate s DA The noise and image in the image are as follows:

[0037]

[0038] Where I(n) represents the input signal, IMF m (n) represents M th The intrinsic modulus function, Res M(n) represents the residual.

[0039] As a preferred technical solution, the FRET three-channel structured light super-resolution linear Wiener reconstruction based on unified Wiener reconstruction parameters to obtain FRET three-channel super-resolution image data is specifically as follows:

[0040]

[0041] in, Here, k represents the Fourier transform of the corresponding image in the FRET channel X; k is the coordinate of the fluorescent image pixel transformed into the frequency domain. d This represents the spatial frequency of the FRET channel X at the corresponding structured light direction angle. The optical transfer function (OTF) of the SR-SIM system. This indicates that the OTF of the SR-SIM system corresponding to FRET channel X is in the frequency domain according to the spatial frequency vector k. d Spectrum shifting is performed; w represents the reconstructed Wiener parameters, and A(k) is the Gaussian apodization function;

[0042] In particular, the Wiener reconstruction parameters are kept consistent during the FRET three-channel super-resolution reconstruction process.

[0043] As a preferred technical solution, the data processing based on the channel sensitization intensity measurement method, specifically measuring the FRET efficiency and donor-acceptor concentration ratio of the FRET super-resolution image, involves:

[0044]

[0045]

[0046]

[0047] Among them, E SIM For the FRET efficiency under structured light super-resolution, Rc SIM Fc represents the donor-acceptor concentration ratio of FRET under structured light super-resolution. SIM denoted as α, where α is the receptor-sensitized emission fluorescence intensity, G is the sensitization quenching conversion factor, K is the donor-receptor concentration conversion factor, and a, b, c, and d are system crosstalk coefficients.

[0048] Another aspect of the present invention provides an estimation system for reconstruction parameters of FRET-sensitized channel super-resolution SIM images, applied to the above-mentioned estimation method for reconstruction parameters of FRET-sensitized channel super-resolution SIM images, including a module for obtaining original image data sets and optical transfer functions, a module for broadening spectral images, a module for solving and replacing reconstruction parameters, a module for obtaining initial phase and modulation depth, a module for super-resolution linear Wiener reconstruction of three-channel structured light, and a module for measuring FRET efficiency and donor-acceptor concentration ratio.

[0049] The original image data set and optical transfer function acquisition module are used to acquire the FRET three-channel original image data set modulated by structured light; the FRET three channels include a donor-excited donor emission channel, i.e., the DD channel, a acceptor-excited acceptor emission channel, i.e., the DA channel, and a donor-excited acceptor emission channel, i.e., the AA channel; and to acquire the optical transfer function (OTF) of the FRET three channels;

[0050] The original image data set and optical transfer function acquisition module are used to obtain the spectral image of the FRET three-channel image using Fourier transform, and to obtain the broadened spectral image by spectral shifting.

[0051] The reconstruction parameter solving and replacement module is used to obtain the reconstruction parameters of the DD channel image group, DA channel image group and AA channel image group from the broadened spectral image; using the forward physical process model of SIM-FRET three-channel imaging, the structured light illumination frequency vector in the reconstruction parameters of the DA channel image group is replaced with the structured light illumination frequency vector in the reconstruction parameters of the DD channel image group.

[0052] The initial phase and modulation depth acquisition module is used to acquire the initial phase and modulation depth of the DA channel image group using a complex linear regression algorithm that includes empirical mode decomposition.

[0053] The three-channel structured light super-resolution linear Wiener reconstruction module is used to perform FRET three-channel structured light super-resolution linear Wiener reconstruction based on unified Wiener reconstruction parameters to obtain FRET three-channel super-resolution image data.

[0054] The FRET efficiency and donor-acceptor concentration ratio measurement module is used for data processing based on the channel sensitization intensity measurement method to measure the FRET efficiency and donor-acceptor concentration ratio of FRET super-resolution images.

[0055] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0056] (1) By utilizing the physical relationship between E-FRET channels, this invention effectively solves the problem of inaccurate parameter estimation during image reconstruction caused by the low signal-to-noise ratio of the SIM-FRET sensitized channel image; at the same time, it avoids the problem of motion artifacts or image bleaching that may occur when taking multiple time-averaged images, and achieves accurate parameter estimation with high temporal resolution and clear super-resolution reconstruction; it enhances the robustness of the SIM-FRET method for low signal-to-noise ratio image reconstruction and expands the availability of SIM-FRET under extremely low FRET efficiency. Attached Figure Description

[0057] Figure 1 This is a flowchart of the FRET image reconstruction method based on spatial colocalization mask filtering and SR-SIM as described in the embodiments of the present invention;

[0058] Figure 2 This is the forward physical process model of SIM-FRET three-channel imaging as described in the embodiments of the present invention;

[0059] Figure 3 This is a flowchart illustrating the decision-making process of replacing the DA channel structured light illumination frequency vector with the DD channel image group structured light illumination frequency vector in an embodiment of the present invention.

[0060] Figure 4 This is a comparison chart of the initial phase trend calculated by the method of averaging multiple time-series images and the method of this invention as the average number of images increases;

[0061] Figure 5 This is a schematic diagram illustrating the SIM-FRET image reconstruction results and FRET efficiency of an embodiment of the present invention; wherein, Figure 5 (a) is the result of direct super-resolution reconstruction of the FRET three-channel image and local magnification. Figure 5 (b) is the result of super-resolution reconstruction of the FRET three-channel image after averaging multiple time-series images, followed by local magnification. Figure 5 (c) is the result of super-resolution reconstruction of the FRET three-channel image using the method of the present invention and local magnification.

[0062] Figure 6 This is a schematic diagram of the FRET efficiency according to an embodiment of the present invention; wherein Figure 6 (a) The FRET information results calculated from the directly super-resolution reconstructed FRET three-channel image. Figure 6 (b) The FRET information results of the three channels of FRET calculated from the super-resolution reconstructed image after averaging multiple temporal images. Figure 6 (c) is the result of super-resolution reconstruction of the FRET three-channel image using the method of the present invention to calculate the FRET information.

[0063] Figure 7This is a schematic diagram of the structure of an estimation system for super-resolution SIM image reconstruction parameters of a FRET-sensitized channel according to an embodiment of the present invention. Detailed Implementation

[0064] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative effort are within the scope of protection of the present application.

[0065] Example

[0066] like Figure 1 As shown, this embodiment provides a method for estimating reconstruction parameters of FRET-sensitized channel super-resolution SIM images. This method can be applied to super-resolution observation of the activities of biomacromolecules in living cells, enabling more refined spatiotemporal dynamic imaging of molecular structures and their functions within cells. It provides a powerful technical means to reveal the inter-regulatory relationships between subcellular structures, molecular structures, and their functions. The method includes the following steps:

[0067] S1. Acquire the raw image data set of the three-channel FRET after structured light modulation; the three-channel FRET includes the donor-excited donor emission channel DD, the acceptor-excited acceptor emission channel DA, and the donor-excited acceptor emission channel AA. For different FRET donors and acceptors, the corresponding excitation and emission channels of FRET are also different. In this embodiment, the fluorescent protein pair GFP-mCherry is selected as the FRET donor and acceptor pair. The donor excitation channel uses a 488nm laser as the excitation light of the donor, and the donor emission channel uses a 525±15nm channel as the emission fluorescence detection channel of the donor. The acceptor excitation channel uses a 561nm laser as the excitation light of the acceptor, and the acceptor emission channel uses a 600±15nm channel as the emission fluorescence detection channel of the acceptor.

[0068] The SR-SIM system was used to acquire a set of three-channel raw image data, including the donor channel image set, based on the channel sensitization intensity measurement method (E-FRET method). (Image set of fluorescence detected by DD in donor-excited donor emission channel), image set of acceptor channel (Image set of fluorescence detected by receptor-induced receptor emission channels AA) and FRET channel image set (A set of fluorescence images detected by the donor-excited acceptor emission channel DA); where r = (x, y) are the spatial coordinates of the fluorescence image pixels. Each set of raw image data contains several different structured light direction angles, and each structured light azimuth angle contains several wide-field fluorescence images modulated by cosine structured light with different phase differences.

[0069] Specifically, in this embodiment, each set of original image data includes three different structured light direction angles θ1, θ2 and θ3, and each structured light azimuth angle includes three structured light modulated wide-field fluorescence images with different phase differences n1, n2 and n3. A total of 27 structured light modulated wide-field fluorescence images are acquired in the DD, DA and AA channels, which is a preferred technical solution.

[0070] S2. Obtain the optical transfer function (OTF) of the three channels of FRET;

[0071] The corresponding optical transfer function (OTF) H of the FRET channel X X (k) The point spread function (PSF) obtained from photographs of fluorescent microspheres is obtained by screening, segmentation, averaging multiple images, and then performing a Fourier transform. In this embodiment, the photographs of fluorescent microspheres are first screened, selecting those with regular shapes and relatively far distance from other fluorescent microspheres. Then, the selected fluorescent microsphere images are segmented, and after averaging multiple images, a Fourier transform is performed to obtain the optical transfer function (OTF). The H of the FRET DD channel... DD (k) Obtained by measuring 80 nm fluorescent microspheres with 488 nm excitation / 525 nm emission; FRET DA, AA channel OTFH DA (k) and H AA (k) was obtained by measuring 80 nm fluorescent microspheres with 561 nm excitation / 600 nm emission.

[0072] S3. Obtain the broadened spectral image of the FRET three-channel image;

[0073] For S1 in I DD (r), I DA (r), I AA (r) performs a Fourier transform, I X (k)=FFT(I X (r)) yields the spectral image corresponding to the FRET channel X, where k is the coordinate of the fluorescence image pixel transformed into the frequency domain.

[0074] Furthermore, during the super-resolution spectrum shifting process, the size of the shifted image may exceed the original spectrum image size. Therefore, the size of the FRET three-channel spectrum image is widened from m×n to 2m×2n. Obtaining the stretched FRET three-channel spectral image can avoid the loss of image information caused by the spectral shift exceeding the original image spectral size, thus ensuring the accuracy of super-resolution reconstruction.

[0075] S4. Using an iterative cross-correlation algorithm, reconstruction parameters for the DD channel image group and the AA channel image group are obtained from the broadened spectral image. Specifically, the reconstruction parameters for the X channel are: the structured light illumination frequency vector. Modulation depth and initial phase

[0076] Using the iterative cross-correlation algorithm described in the existing technique "Gustafsson MGL, et al. Three-dimensional resolution doubling in wide-field fluorescence microscopy by structured illumination. Biophys J, 2008", the structured light illumination frequency vectors of two channels are obtained from the DD channel image group and AA channel image group of structured light modulation, respectively. Furthermore, the modulation depth is obtained based on the aforementioned complex linear regression method. and initial phase

[0077] In particular, due to the modulation depth Compared to the initial phase It is more susceptible to interference from inaccurate overlapping regions. When calculating the modulation depth, the overlapping region selected is smaller than the region selected when calculating the initial phase, so as to ensure that the distortion of relative fluorescence intensity between the reconstructed image and the original image is minimized, thereby meeting the requirement of fluorescence intensity fidelity during FRET calculation. In this embodiment, the size of the modulation depth overlapping region is approximately 0.8 times the size of the initial phase region.

[0078] S5. Using the forward physical process model of SIM-FRET three-channel imaging, replace the structured light illumination frequency vector in the reconstruction parameters of the DA channel image group with the structured light illumination frequency vector in the reconstruction parameters of the DD channel image group in S4.

[0079] (1) According to Figure 2 The model shown is a forward physical process model for SIM-FRET three-channel imaging. The FRET sample is first subjected to... Two types of excitation light excitation, among which, These are the average intensity of the excitation light, It is a two-dimensional structured light superimposed from two excitation lights. The modulation depth of the structured light. The structured light illumination frequency vector, This is the initial phase. Under these two excitation lights, the FRET sample produced different fluorescent molecule distributions: S D (r) and S A (r), and superimposed with two different excitation lights to produce donor excitation image groups Em D (r)=Ex D (r)·S D (r) and receptor excitation image group Em A (r)=Ex A (r)·S A (r), based on existing E-FRET technology [ Chen Tongsheng ; Sun Han ; Zhuang Zhengfei "Method and Application of Correction Factor for Linear Separation and Quantification FRET System Based on Measurement of Excitation-Emission Spectra of Cell Samples in the Same System" National Invention Patent 2021.3, Authorized Patent No.: CN202110270760.4, Donor Excitation Image Group Em D (r) By switching between different wavelength donor or acceptor filters, DD channel image groups I are obtained respectively. DD (r) and DA channel image group I DA (r):

[0080]

[0081]

[0082] in, The average intensity of the image, The modulation depth is the result of the light passing through the filter. The structured light illumination frequency vector of the image. This is the initial phase.

[0083] Under ideal conditions, based on optical diffraction relations, the DD channel image group and the DA channel image group... All images were generated by donor-excited image group Em D (r) provide, i.e. Furthermore, this value does not change when passing through the filter, so and The following relationships should be present:

[0084]

[0085] Where ε represents a small error, typically present in the algorithm's solution. During the process, it is generally believed that it will not affect the subsequent calculation of the initial phase and modulation depth.

[0086] (2) Table 1 shows a statistical comparison of the structured illumination frequency vectors calculated after averaging multiple time-series images of the SIM-FRET DD channel image group and DA channel image group for three different samples. According to existing technology [Fan Junchao, Huang Xiaoshuai, Tan Shan, "A Low Signal-to-Noise Ratio Image Reconstruction Method and System" National Invention Patent 2019.7, Authorized Patent No.: CN201710200970.X], averaging multiple time-series images can improve the image signal-to-noise ratio from a statistical perspective. The reconstruction parameter results obtained by this method have been proven to be usable as a control result. The structured illumination frequency vectors of the corresponding DD channel image group and DA channel image group were calculated using the iterative cross-correlation algorithm, and the distance between their vectors was calculated. The results in Table 1 are all less than 0.1 pixel units, which is less than the allowable error ε, proving that the relationship is valid.

[0087]

[0088] Table 1. Spatial frequency comparison of structured light between the DD channel image group and the DA channel image group of the three sample groups.

[0089] Furthermore, such as Figure 3 The flowchart illustrates the decision-making process for replacing the structured light illumination frequency vector of the DA channel with the structured light illumination frequency vector of the DD channel image group. Since the DD and DA channel images share the same field of view, the DD channel image group is first used as the reference image to determine if the peak signal-to-noise ratio (PSNR) of the DA channel image is sufficient for subsequent parameter estimation. A PSNR greater than 20 indicates high image information reliability; conversely, a PSNR less than 20 indicates low signal-to-noise ratio in the DA channel image group, requiring correction using the reconstruction parameters from the DD channel image group. When the PSNR is greater than 20, for the DA channel, if the vector distance between the structured light illumination frequency vector calculated using the traditional iterative cross-correlation algorithm and the structured light illumination frequency vector in the reconstructed parameters of the DD channel image group exceeds a set threshold for the magnitude of the DD channel structured light illumination frequency vector (i.e., exceeding the allowable error ε; in this embodiment, the set threshold is 0.5%), it also indicates that... It is no longer reliable and requires the use of structured light illumination frequency vectors from DD channel image groups. Replace the structured light illumination frequency vector of the DA channel image group Right now: This completes the calculation of the initial phase and modulation depth of the subsequent DA channel.

[0090] S6. Obtain the initial phase of the DA channel image group using a complex linear regression algorithm that includes empirical mode decomposition. and modulation depth

[0091] Based on the structured light illumination frequency vector of the DA image group obtained in step S5, complex linear regression is calculated on the overlapping region of the DA channel image group to obtain the modulation depth and initial phase of the DA channel image group:

[0092]

[0093]

[0094]

[0095]

[0096]

[0097] Among them, s DA a i DA b i DA As an intermediate variable, This is a super-resolution low-frequency band image for the DA channel, where m is the magnitude of the spectral shift, typically m = +1 or -1. This is the m-band image of the DA channel. To The DA channel m-band image after frequency shifting, where the shift distance and direction are the structured light illumination frequency vector. The modulation depth of the structured light illumination frequency vector corresponding to the DA channel. This represents the initial phase of the structured light illumination frequency vector corresponding to the DA channel.

[0098] In particular, due to the low signal-to-noise ratio of the DA channel image group, the intermediate variables in linear regression... and Contains a lot of noise, directly affecting s DA The summation in step S6 can lead to inaccurate calculations of the modulation depth and initial phase. Therefore, in this embodiment, empirical mode decomposition (EMD) is used to separate the modulation depth and initial phase before calculating them. DA Noise and images in the image:

[0099]

[0100] Where I(n) represents the input signal, IMF m (n) represents M th The intrinsic modulus function, Res M (n) represents the residual. The EMD method decomposes the input signal into several intrinsic mode functions and residuals, thereby achieving the purpose of removing noise and obtaining the main components of the input signal. In this embodiment, for s DAAfter EMD processing, the intrinsic mode functions are obtained and summed to restore them. Then, the summation and modulation depth and initial phase are calculated in the above steps. This is a preferred method in this embodiment.

[0101] like Figure 4 The comparison chart shows the initial phase trends calculated using the method of averaging multiple time-series images and the method of this invention. It can be seen that the result obtained by this invention using only one image is consistent with the initial phase result obtained by averaging 5 images. Moreover, as the number of average time-series images increases, the initial phase result does not change significantly. The error between the result obtained by averaging 5 images and the result obtained by averaging 5 images is less than 0.05, which is within the allowable error range.

[0102] S7. Based on the unified Wiener reconstruction parameters, perform FRET three-channel structured light super-resolution linear Wiener reconstruction to obtain FRET three-channel super-resolution image data.

[0103]

[0104] in, Here, k represents the Fourier transform of the corresponding image in the FRET channel X; k is the coordinate of the fluorescent image pixel transformed into the frequency domain. d This represents the spatial frequency of the FRET channel X at the corresponding structured light direction angle. The optical transfer function (OTF) of the SR-SIM system. This indicates that the OTF of the SR-SIM system corresponding to FRET channel X is in the frequency domain according to the spatial frequency vector k. d Spectrum shifting is performed; w represents the reconstructed Wiener parameters, and A(k) is the Gaussian apodization function.

[0105] In particular, in the above formula for FRET three-channel structured light super-resolution linear Wiener reconstruction, each set of original image data contains three different structured light direction angles, and each structured light azimuth angle contains three different phase differences. This selection is a preferred method in this embodiment.

[0106] Furthermore, during the super-resolution linear Wiener reconstruction process, changes in the Wiener parameters can affect the grayscale values ​​of the reconstructed super-resolution image. By maintaining the consistency of the Wiener reconstruction parameters during FRET three-channel super-resolution reconstruction, the distortion of the relative fluorescence intensity between the reconstructed image and the original image can be minimized, thereby meeting the requirement of fluorescence intensity fidelity during FRET calculation. In this embodiment, the Wiener parameters are selected as a constant between 0.1 and 0.5 based on the signal-to-noise ratio of the three-channel image.

[0107] S8. Based on the channel sensitization intensity measurement method, E-FRET data processing is performed to measure the FRET efficiency and donor-acceptor concentration ratio of the FRET super-resolution image.

[0108]

[0109]

[0110]

[0111] Among them, E SIM For the FRET efficiency under structured light super-resolution, Rc SIM Fc represents the donor-acceptor concentration ratio of FRET under structured light super-resolution. SIM denoted as α, where α is the receptor-sensitized emission fluorescence intensity, G is the sensitization quenching conversion factor, K is the donor-receptor concentration conversion factor, and a, b, c, and d are system crosstalk coefficients.

[0112] Furthermore, the sensitization quenching conversion factor G, the donor-acceptor concentration conversion factor k, and the system crosstalk coefficients a, b, c, and d can be determined by preparing two standard plasmid samples with different fixed FRET efficiencies and a donor-acceptor concentration ratio of 1:1, as well as by transfecting donor and acceptor plasmid samples separately. For different FRET donor-acceptor pairs and FRET measurement systems, the system parameters such as a, b, c, d, G, and K will also be different. In this embodiment, a = 0.010135, b = 0.001037, c = 0.001641, d = 0.111086, G = 0.621361, and k = 0.292009 were obtained through pre-measurement.

[0113] To further demonstrate the effect of using DD channel super-resolution reconstruction parameters to calculate DA channel super-resolution image parameters on super-resolution SIM-FRET, and to compare the results with those of super-resolution SIM-FRET using direct DA channel calculation and super-resolution SIM-FRET using a multi-image averaging parameter estimation method, we used GFP-mCherry as the donor-acceptor pair and, combined with the actual parameters from the above examples, captured intracellular mitochondrial images using an existing SIM-FRET system. To obtain the reconstruction effect of multiple average images, multiple time-series images were continuously captured for each of the nine SIM super-resolution images from the three FRET channels, such as... Figure 4 The figure shows the trend of the initial phase after averaging multiple images. It can be seen that the initial phase result has stabilized after averaging 5 images. Therefore, in this embodiment, 5 time-series images are taken consecutively, for a total of 3*9*5=135 images. The first group of experiments uses a single time-series image group to directly perform super-resolution reconstruction and calculate its FRET efficiency. The second group of experiments uses the super-resolution images after averaging 5 images to reconstruct and calculate the FRET efficiency. The above two groups serve as the control group. The third group reconstructs and calculates the FRET efficiency according to the steps of this embodiment and serves as the experimental group for comparison with the control group.

[0114] Figure 5These are schematic diagrams illustrating image reconstruction from the three sets of experiments described above; where... Figure 5 (a) presents the direct super-resolution reconstructed DA channel image and the local magnification results. Figure 5 (b) presents the reconstructed image and local magnification results after averaging five images. Figure 5 (c) shows the reconstructed image and local magnification results obtained using the method of the present invention. From Figure 5 (a) Figure 5 (b) and Figure 5 (c) The comparison shows that both the multi-image averaging method and the present invention effectively reduce noise in the reconstructed images, preserve more fine structure, and the reconstructed images have more obvious biological information structure; the peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) of the three reconstructed images are calculated, where Figure 5 (a) has PSNR and SSIM of 19.87 and 0.045, respectively. Figure 5 (b) The PSNR and SSIM are 28.17 and 0.1209, respectively. Figure 5 (c) shows that the PSNR and SSIM are 24.72 and 0.0920, respectively. It can be seen that the present invention has a significant improvement in image quality compared with the direct super-resolution reconstruction result, and is close to the algorithm of multiple average temporal images that have been denoised by statistical methods.

[0115] Figure 7 These are schematic diagrams of the FRET results of the three sets of experiments mentioned above; where... Figure 6 (a) The FRET information results calculated from the DA channel image obtained by direct super-resolution reconstruction are presented. Figure 6 (b) The FRET information results calculated using 5 averaged reconstructed images are presented. Figure 6 (c) The FRET information results calculated from the reconstructed image using the method of the present invention are given; it can be seen that... Figure 6 (a) The results differed significantly from the other two groups in the standard deviation of FRET efficiency (Ed) and donor-acceptor concentration ratio (Rc), due to Figure 5 (a) The reconstruction results contain a large amount of noise and artifacts, resulting in a low signal-to-noise ratio, which in turn leads to a small amount of effective FRET data, resulting in a large standard deviation of FRET efficiency and donor-receptor concentration ratio; the average multiple time-series image method ( Figure 5 (b) shows a high signal-to-noise ratio and contains more effective FRET information. Figure 6 (b) The FRET efficiency Ed and donor-acceptor concentration ratio Rc are significantly better than those of the previous methods. Figure 6 (a) However, the average multiple time-series image method requires a long shooting time, and the samples themselves cannot undergo significant movement, otherwise motion artifacts are easily generated; the reconstructed image of this invention Figure 5 (c) FRET results obtained Figure 6 (c) FRET signal comparison Figure 6 (b) They are consistent and close to the true value of FRET theory, which shows that the method of the present invention meets the fidelity requirements for calculating fluorescence intensity in FRET. At the same time, compared with the multi-image averaging method, the present invention only needs to take one image to obtain effective FRET information, which takes 1 / 5 of the time of averaging multiple time-series images, greatly improving the temporal resolution and avoiding motion artifacts, further demonstrating the superiority of the present invention.

[0116] like Figure 7 As shown, in another embodiment of this application, an estimation system for reconstruction parameters of FRET-sensitized channel super-resolution SIM image is provided. The system includes a raw image data set and optical transfer function acquisition module, a spectral image broadening module, a reconstruction parameter solving and replacement module, an initial phase and modulation depth acquisition module, a three-channel structured light super-resolution linear Wiener reconstruction module, and a FRET efficiency and donor-acceptor concentration ratio measurement module.

[0117] The original image data set and optical transfer function acquisition module are used to acquire the FRET three-channel original image data set modulated by structured light; the FRET three channels include a donor-excited donor emission channel, i.e., the DD channel, a acceptor-excited acceptor emission channel, i.e., the DA channel, and a donor-excited acceptor emission channel, i.e., the AA channel; and to acquire the optical transfer function (OTF) of the FRET three channels;

[0118] The original image data set and optical transfer function acquisition module are used to obtain the spectral image of the FRET three-channel image using Fourier transform, and to obtain the broadened spectral image by spectral shifting.

[0119] The reconstruction parameter solving and replacement module is used to obtain the reconstruction parameters of the DD channel image group, DA channel image group and AA channel image group from the broadened spectral image; using the forward physical process model of SIM-FRET three-channel imaging, the structured light illumination frequency vector in the reconstruction parameters of the DA channel image group is replaced with the structured light illumination frequency vector in the reconstruction parameters of the DD channel image group.

[0120] The initial phase and modulation depth acquisition module is used to acquire the initial phase and modulation depth of the DA channel image group using a complex linear regression algorithm that includes empirical mode decomposition.

[0121] The three-channel structured light super-resolution linear Wiener reconstruction module is used to perform FRET three-channel structured light super-resolution linear Wiener reconstruction based on unified Wiener reconstruction parameters to obtain FRET three-channel super-resolution image data.

[0122] The FRET efficiency and donor-acceptor concentration ratio measurement module is used for data processing based on the channel sensitization intensity measurement method to measure the FRET efficiency and donor-acceptor concentration ratio of FRET super-resolution images.

[0123] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure can be divided into different functional modules to complete all or part of the functions described above. This system is an estimation method for FRET-sensitized channel super-resolution SIM image reconstruction parameters applied to the above embodiments.

[0124] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0125] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A method for estimating reconstruction parameters of FRET-sensitized channel super-resolution SIM images, characterized in that, Includes the following steps: Acquire a raw image data set of three FRET channels modulated by structured light; the three FRET channels include a donor-excited donor emission channel, i.e., the DD channel, a acceptor-excited acceptor emission channel, i.e., the DA channel, and a donor-excited acceptor emission channel, i.e., the AA channel. Obtain the optical transfer function (OTF) of the three channels of FRET; Fourier transform is used to obtain the spectral image of the FRET three-channel image, and the broadened spectral image is obtained by spectral shifting; Reconstruction parameters for the DD channel image group, DA channel image group, and AA channel image group are obtained from the broadened spectral image. Using the forward physical process model of SIM-FRET three-channel imaging, the structured light illumination frequency vector in the reconstruction parameters of the DA channel image group is replaced with the structured light illumination frequency vector in the reconstruction parameters of the DD channel image group in the previous step; The initial phase and modulation depth of the DA channel image group are obtained using a complex linear regression algorithm that includes empirical mode decomposition; FRET three-channel structured light super-resolution linear Wiener reconstruction was performed based on unified Wiener reconstruction parameters to obtain FRET three-channel super-resolution image data. Data processing was performed using a channel sensitization intensity measurement method to measure FRET efficiency and donor-acceptor concentration ratio in FRET super-resolution images.

2. The method for estimating FRET-sensitized channel super-resolution SIM image reconstruction parameters according to claim 1, characterized in that, The specific steps for obtaining the FRET three-channel raw image data set modulated by structured light are as follows: A set of three-channel raw image data based on the channel sensitization intensity measurement method was acquired using the SR-SIM system, including a set of donor-excited donor-emission channel images. Image group of donor-excited receptor emission channels and donor-induced receptor emission pathway imaging group ,in The spatial coordinates of the fluorescence image pixels are represented by the original image data set. Each set of original image data contains several different structured light direction angles, and each structured light azimuth angle contains several cosine structured light illumination wide-field fluorescence images with different phase differences.

3. The method for estimating reconstruction parameters of FRET-sensitized channel super-resolution SIM images according to claim 1, characterized in that, The specific steps for obtaining the optical transfer function (OTF) of the three FRET channels are as follows: The point spread function (FPF) of the three channels of fluorescent microspheres was obtained by imaging with the SR-SIM system. After screening, segmentation, and averaging of multiple images, Fourier transform was performed to obtain the optical transfer function (OTF) of the three channels of FPF. .

4. The method for estimating FRET-sensitized channel super-resolution SIM image reconstruction parameters according to claim 1, characterized in that, The specific steps involve obtaining the spectral image of the FRET three-channel image using Fourier transform and then obtaining the broadened spectral image through spectral shifting: First, the FRET three-channel image group , Perform Fourier transform, The spectral image corresponding to the FRET channel X is obtained, where k is the coordinate of the fluorescence image pixel transformed into the frequency domain; The size of the FRET three-channel spectral image was reduced from... Expand to , The stretched FRET three-channel spectrum image was obtained.

5. The method for estimating FRET-sensitized channel super-resolution SIM image reconstruction parameters according to claim 1, characterized in that, The reconstruction parameters of the DD channel image group and the AA channel image group are obtained from the broadened spectral image using an iterative cross-correlation algorithm. The reconstruction parameters of channel X include: structured light illumination frequency vector. Modulation depth and initial phase .

6. The method for estimating reconstruction parameters of FRET-sensitized channel super-resolution SIM images according to claim 1, characterized in that, The forward physical process model utilizing SIM-FRET three-channel imaging replaces the structured light illumination frequency vector in the DA channel image group reconstruction parameters with the structured light illumination frequency vector in the DD channel image group reconstruction parameters from the previous step, specifically as follows: Calculate the structured light illumination frequency vector of the DA channel image Then, the vector distance between it and the structured light illumination frequency vector in the reconstruction parameters of the DD channel image group is calculated. if When the structured light illumination frequency vector magnitude of the DD channel is greater than the set threshold, the structured light illumination frequency vector of the DD channel image group is used. Replace the structured light illumination frequency vector of the DA channel image group ,Right now: 。 7. The method for estimating reconstruction parameters of FRET-sensitized channel super-resolution SIM images according to claim 1, characterized in that, The specific steps for obtaining the initial phase and modulation depth of the DA channel image group using a complex linear regression algorithm incorporating empirical mode decomposition are as follows: Based on the structured light illumination frequency vector of the obtained DA image group, complex linear regression is calculated on the overlapping region of the DA channel image group to obtain the modulation depth and initial phase of the DA channel image group: ； ； ； ； ； in, As an intermediate variable, For DA channel super-resolution low-frequency band images, For the magnitude of the spectrum shift, , This is the m-band image of the DA channel. To The DA channel m-band image after frequency shifting, where the shift distance and direction are the structured light illumination frequency vector. , The modulation depth of the structured light illumination frequency vector corresponding to the DA channel. This represents the initial phase of the structured light illumination frequency vector corresponding to the DA channel; Before determining the modulation depth and initial phase, empirical mode decomposition is used to separate the modulation depth and initial phase. The noise and image are extracted, and the intrinsic mode functions are obtained and summed to restore them; The use of empirical mode decomposition method for separation The noise and image in the image are as follows: ； in Indicates the input signal. express The intrinsic modulus function, Represents the residual.

8. The method for estimating FRET-sensitized channel super-resolution SIM image reconstruction parameters according to claim 1, characterized in that, The FRET three-channel structured light super-resolution linear Wiener reconstruction based on unified Wiener reconstruction parameters yields FRET three-channel super-resolution image data, specifically: ； in, , where is the Fourier transform of the corresponding image in the FRET channel X; k is the coordinate of the fluorescent image pixel transformed into the frequency domain. This represents the spatial frequency of the FRET channel X at the corresponding structured light direction angle. The optical transfer function (OTF) of the SR-SIM system. This indicates that the OTF of the SR-SIM system corresponding to FRET channel X is in the frequency domain space according to the spatial frequency vector. Perform spectrum shifting; To reconstruct Wiener parameters, It is a Gaussian apodization function; Maintain consistency of Wiener reconstruction parameters during FRET three-channel super-resolution reconstruction.

9. The method for estimating FRET-sensitized channel super-resolution SIM image reconstruction parameters according to claim 1, characterized in that, The data processing based on the channel sensitization intensity measurement method involves measuring the FRET efficiency and donor-acceptor concentration ratio in the FRET super-resolution image, specifically as follows: ； ； ； in, The FRET efficiency under structured light super-resolution. The FRET donor-acceptor concentration ratio under structured light super-resolution. denoted as α, where α is the receptor-sensitized emission fluorescence intensity, G is the sensitization quenching conversion factor, K is the donor-receptor concentration conversion factor, and a, b, c, and d are system crosstalk coefficients.

10. A system for estimating reconstruction parameters of FRET-sensitized channel super-resolution SIM images, characterized in that, An estimation method for reconstruction parameters of a super-resolution SIM image using a FRET-sensitized channel, applicable to any one of claims 1-9, includes a module for obtaining the original image data set and optical transfer function, a module for broadening the spectral image, a module for solving and replacing reconstruction parameters, a module for obtaining the initial phase and modulation depth, a module for super-resolution linear Wiener reconstruction using three-channel structured light, and a module for measuring FRET efficiency and donor-acceptor concentration ratio. The original image data set and optical transfer function acquisition module are used to acquire the FRET three-channel original image data set modulated by structured light; the FRET three channels include a donor-excited donor emission channel, i.e., the DD channel, a acceptor-excited acceptor emission channel, i.e., the DA channel, and a donor-excited acceptor emission channel, i.e., the AA channel; and to acquire the optical transfer function (OTF) of the FRET three channels; The original image data set and optical transfer function acquisition module are used to obtain the spectral image of the FRET three-channel image using Fourier transform, and to obtain the broadened spectral image by spectral shifting. The reconstruction parameter solving and replacement module is used to obtain the reconstruction parameters of the DD channel image group, DA channel image group and AA channel image group from the broadened spectral image; using the forward physical process model of SIM-FRET three-channel imaging, the structured light illumination frequency vector in the reconstruction parameters of the DA channel image group is replaced with the structured light illumination frequency vector in the reconstruction parameters of the DD channel image group. The initial phase and modulation depth acquisition module is used to acquire the initial phase and modulation depth of the DA channel image group using a complex linear regression algorithm that includes empirical mode decomposition. The three-channel structured light super-resolution linear Wiener reconstruction module is used to perform FRET three-channel structured light super-resolution linear Wiener reconstruction based on unified Wiener reconstruction parameters to obtain FRET three-channel super-resolution image data. The FRET efficiency and donor-acceptor concentration ratio measurement module is used for data processing based on the channel sensitization intensity measurement method to measure the FRET efficiency and donor-acceptor concentration ratio of FRET super-resolution images.

Images