METHOD AND DEVICE FOR HIGH-RESOLUTION MICROSCOPIC IMAGING, COMPUTER DEVICE AND STORAGE MEDIUM

Description

Field of the Invention

[0001] The present disclosure relates to a super-resolution microscopic imaging method and apparatus, a computer device, and a storage medium in the field of computational imaging and super-resolution microscopic imaging.Background of the Invention

[0002] Super-resolution optical fluctuation imaging (SOFI), based on the fluctuation model of molecular intensity signal, uses the physical model of the random fluctuation of intensity of fluorescent molecules and does not rely on any hardware modulation, which is a flexible and highly cost-effective super-resolution means. Due to its hardware system-independent nature, it is flexibly coupled to different imaging modalities. The disadvantages of such apparatuses are the low temporal resolution of existing fluorescence super-resolution methods, requiring at least 500 to 1000 consecutive collections of images to achieve the desired high-quality super-resolution effect, which hinders their application in live cell super-resolution imaging. Therefore, there is a need to provide a method that maximizes the use of the fluorescence fluctuation behavior detectable in each measurement to achieve the desired high temporal resolution and high throughput.

[0003] ZHAO WEISONG ET AL: "Ultrafast super-resolution imaging via autocorrelation two-step deconvolution", SPIE PROCEEOINGS; [PROCEEOINGS OF SPIE ISSN 0277-786X], SPIE, US, vol. 11497, 20 August 2020 (2020-08-20), pages 114970V-114970V, discloses an imaging method achieving ultrafast super-resolution via autocorrelation two-step deconvolution.

[0004] CA 3124052 A1 teaches systems and methods for image processing.

[0005] Sami Koho ET AL: "Fourier ring correlation simplifies image restoration in fluorescence microscopy", Nature Communications, vol. 10, no. 1, 15 July 2019 (2019-07-15), proposes a blind image restoration method based on Fourier Ring Correlation (FRC).Summary of the Invention

[0006] The invention is set out in the appended set of claims.Brief Description of the Drawings

[0007] In order to explain the embodiments of the present disclosure or the technical solutions in the prior art more clearly, a brief introduction will be made to the accompanying drawings used in the embodiments or the description of the prior art. It is obvious that the drawings in the description below are only some embodiments of the present disclosure, and those ordinarily skilled in the art can obtain other drawings according to the structures shown in these drawings without creative work. Fig. 1 is a flowchart of a super-resolution microscopic imaging method according to Embodiment 1 of the present disclosure; Fig. 2 is an exemplary result diagram of three-dimensional spatial resolution enhancement according to Embodiment 1 of the present disclosure; Fig. 3 is an exemplary result diagram of temporal resolution enhancement according to Embodiment 1 of the present disclosure; Fig. 4 is an exemplary result diagram of non-parametric reconstruction capability according to Embodiment 1 of the present disclosure; Fig. 5 is an exemplary result diagram of an automated high-throughput reconstruction according to Embodiment 1 of the present disclosure; Fig. 6 is a structural block diagram of a super-resolution microscopic imaging apparatus according to Embodiment 2 of the present disclosure; and Fig. 7 is a structural block diagram of a computer device according to Embodiment 3 of the present disclosure. Detailed Description of the Embodiments

[0008] In order to make the objects, technical solutions, and advantages of the embodiments of the present disclosure clearer, the technical solutions of the embodiments of the present disclosure will be described clearly and completely with reference to the drawings of the embodiments of the present disclosure. It is obvious that the described embodiments are a part of the embodiments of the present disclosure rather than all the embodiments thereof, and all other embodiments obtained by the ordinarily skilled in the art without inventive effort are within the scope of protection of the present disclosure.Embodiment 1

[0009] As shown in Fig. 1, the embodiment provides a super-resolution microscopic imaging method based on a deconvolution enhancement implementation, including the following steps: At S101, a fluorescent signal sequence of a set of samples to be observed is collected.

[0010] In one embodiment, the number of frames of the fluorescent signal sequence is 20; the skilled in the art may understand that the number of frames of the fluorescent signal sequence may also be 10, 50, and the like. Each frame may represent a recorded image.

[0011] In one embodiment, for data with weak background or even no background, the background estimation operation may be removed, that is, setting the background parameter value b to zero, to avoid the removal of information; in particular, considering that an image under low-dose illumination generally shows only a low and stable background fluorescence noise distribution, a value exceeding the image mean value is directly set as zero, and the obtained residual image is used for subsequent background estimation.

[0012] In one embodiment, after step S101, other conditions than those described above, may further include: At S102, a background of the fluorescent signal is estimated using a wavelet transformation, and background noise is removed.

[0013] In one embodiment, step S102 specifically includes: At S1021, the background is estimated from a lowest frequency of an input image, such as the recorded image(s) using wavelet estimation.

[0014] In one embodiment, to extract the lowest band in the frequency domain, a two-dimensional Daubechies-6 wavelet filter is used to decompose the signal multilevel to seven levels.

[0015] At S1022, a wavelet inverse transformation is performed on an input image that may be the recorded image on a lowest frequency band to a spatial domain, a result is compared with half of a square root of the input image, and the two images are combined by keeping a minimum value of each pixel.

[0016] In one embodiment, to prevent accidental removal of small useful signals, the lowest frequency band is performed with a wavelet inverse transformation to the spatial domain; the result is compared to half of the square root of the input image (smoother information); and the two images are combined by keeping the minimum value for each pixel, removing high intensity pixels in the background due to inaccurate background estimation.

[0017] At S1023, estimated low-frequency band low-peak background data is taken as a new input image, and wavelet estimation is performed circularly, namely, repeating steps S1021 and S1022 until the preset number of cycles is reached.

[0018] In one embodiment, the preset number of cycles is set to 3 to estimate the true fluorescence background with the smallest distribution.

[0019] At S102, the number of iterations of deconvolution is determined.

[0020] In one embodiment, the determining the number of iterations of deconvolution specifically includes: using rolling FRC to determine, and taking the number of iterations at maximum rolling FRC resolution as the number of iterations of deconvolution.

[0021] In one embodiment, the using rolling FRC to determine, and taking the number of iterations at maximum rolling FRC resolution as the number of iterations of deconvolution specifically includes: At S1021, the rolling FRC is taken as a function of spatial frequency, and discretization of the spatial frequency of a rolling FRC curve is defined to calculate a discrete value of the corresponding spatial frequency.

[0022] In one embodiment, the rolling FRC measures the statistical correlation between two two-dimensional signals over a series of concentric rings in the Fourier domain, which may be viewed as a function of spatial frequency q i : FRC 12 q i = ∑ r ∈ q i F 1 r ⋅ F 2 * r ∑ r ∈ q i F 1 r 2 ⋅ ∑ r ∈ q i F 2 r 2 , where and represent the discrete Fourier transformation of the two signals; and the summation sign represents the sum of the pixels on the circumferential boundary of the corresponding spatial frequency q i .

[0023] The discretization of the spatial frequency of the rolling FRC curve is defined to calculate the discrete values of the corresponding spatial frequency; the maximum frequency f max corresponds to half of the inverse of the pixel size p s , that is, f max = 1 / (2 p s ); the rolling FRC curve is composed of N / 2 values, and the discretization step length Δf of the spatial frequency is: Δ f = f max N / 2 = 1 Np s

[0024] At S1022, the rolling FRC curve with noise is smoothed using an average filter with average window half-width.

[0025] In one embodiment, an average filter with average window half-width (typically equal to 3) is used to smooth the rolling FRC curve with noise.

[0026] At S1023, when the rolling FRC curve is lower than a given threshold, the frequency is defined as an effective cutoff frequency and the resolution as a reciprocal of the effective cutoff frequency, the given threshold representing a maximum spatial frequency of meaningful information outside the random noise.

[0027] In one embodiment, when the rolling FRC curve is lower than a given threshold, the frequency is defined as the effective cutoff frequency and the resolution as the reciprocal of the effective cutoff frequency, the given threshold representing the maximum spatial frequency of meaningful information outside of random noise; specifically, a common choice for the threshold is a fixed value threshold or a sigma factor curve. The fixed value is usually a 1 / 7 hard threshold, and the criterion for the sigma factor curve may be written as: σ i = σ factor N i / 2 , where N i represents the number of pixels in a ring of radius q i , and the most commonly used sigma factor is 3; if the two measured values only contain noise, then the rolling FRC curve may be expressed as FRC i = 1 / N i ; therefore, the 3-sigma factor curve is actually a frequency component with correlation degree three times greater than pure noise, which is determined as relatively valid information.

[0028] At S1024, the number of iterations at maximum resolution is taken as the number of iterations of deconvolution.

[0029] In one embodiment, the number of iterations at the maximum resolution is denoted as k; k is the number of iterations of deconvolution.

[0030] At S103, an iteration of pre-deconvolution is performed on each frame of the image(s) in the fluorescent signal sequence and outputting each frame to obtain output images, when the iteration reaches half of the number of iterations.

[0031] In one embodiment, when the iteration reaches half of the number of iterations determined in step S102, the iteration of pre-deconvolution is completed, and by performing pre-deconvolution, the effective on / off contrast and signal-to-noise ratio of the signal may be improved; and the iteration of deconvolution specifically includes: At S1031, an iterative formula is obtained by iteratively solving maximum likelihood in a spatial domain.

[0032] In one embodiment, the deconvolution uses a Richardson-Lucy (RL) algorithm; the deconvolution model is based on a Poisson noise model; and the way of solving a maximum likelihood iteratively in a spatial domain is used to obtain the following iterative formula: where x and y are spatial coordinates; h represents a PSF of a microscope; f represents a real fluorescent signal in the physical world; and g represents a signal acquired by a final microscope.

[0033] At S1032, according to the iterative formula, an iterative calculation of deconvolution is realized using an acceleration method based on vector extrapolation.

[0034] In one embodiment, in order to speed up the iterative convergence rate, the iterative calculation of the deconvolution is realized using an acceleration method based on vector extrapolation: y j + 1 = x j ⋅ h T ⋅ g h ⋅ x j v j = x j + 1 − y j α j + 1 = ∑ v j ⋅ v j − 1 ∑ v j − 1 ⋅ v j − 1 , x j + 1 = y j + 1 + α j + 1 ⋅ y j + 1 − y j where g is the reconstructed image constrained by previous prior knowledge; h represents the PSF of the microscope; x j+1< is the image after j+1 iterations; and α is an adaptive acceleration factor.

[0035] At S104, two reconstructions are performed on the output images.

[0036] In one embodiment, step S104 specifically includes: At S1041, a first reconstruction is performed on the output images using a fluctuation principle of the fluorescent signal.

[0037] At S10411, an expression of the fluorescent signal is acquired according to a PSF of a microscope, a luminance constant of a fluorescent molecule, and a function of fluctuation of the luminance of the fluorescent molecule with time.

[0038] In one embodiment, the sample is generally considered to be composed of N individual fluorescent molecules at rk; provided that the fluorescent molecules have independent molecular luminance over time, the fluorescent signal at r and time t is represented as: F r t = h r − r k ⋅ c k ⋅ ω k t , where h, c, and ω represent a PSF, a luminance constant of a molecular, and a function of fluctuation of luminance of the molecular with time of the corresponding microscope, respectively.

[0039] At S10412, according to the expression of the fluorescent signal, an expression of time cumulant with zero-time delay is acquired using individual fluctuation characteristics of each fluorescent molecule.

[0040] In one embodiment, the time cumulant is a second-order time cumulant.

[0041] In one embodiment, according to the fluorescence fluctuation super-resolution imaging technique, using this individual fluctuation characteristic of each fluorescent molecule, the relevant cumulant of each pixel along the t-axis is calculated to improve the resolution, and the second-order time cumulant G 2 of the zero delay is calculated to obtain the following formula: G 2 r = δF r t ⋅ δF r t t δF r t = F r t − F r t t , where 〈·〉 t is a time average function.

[0042] At S10413, the expression of the time cumulant is expanded, and a cross-correlation term of a time cumulant expansion is regarded as zero when a preset condition is met, to make the time cumulant being expressed as the sum of squares of a corresponding luminance constant weighted PSF.

[0043] In one embodiment, the expression of the second-order time cumulant G 2 is expanded to obtain the following formula: G 2 r = ∑ i , k h r − r i ⋅ h r − r k ⋅ c i ⋅ c k ⋅ δω i t δω k t t

[0044] Assuming that the luminescence intensity of each fluorescent molecule is an uncorrelated individual fluctuation, when i≠k (a preset condition), the cross-correlation term in the expansion formula is considered as zero, and the second-order time cumulant G 2 is expressed as the sum of the squares of the corresponding luminance constant γ-weighted PSF, as follows: G 2 r = ∑ i h r − r i 2 ⋅ c i 2 ⋅ δω i t 2 t = ∑ i h r − r i 2 ⋅ γ i

[0045] At S1042, a second reconstruction is performed on the image after the first reconstruction using sparsity-continuity joint constraints.

[0046] At S10421, a reconstruction model is constructed, the reconstruction model including a first term, a second term, and a third term, the first term being a fidelity term representing a distance between the image after the first reconstruction and a collected initial image, such as the recorded image, the second term representing a continuity constraint of the image after the first reconstruction, and the third term representing a sparsity constraint of the image after the first reconstruction.

[0047] In one embodiment, the reconstruction model is constructed as follows: arg min x , b λ 2 f − b − Ax 2 2 + R Hessian x + λ L 1 x 1 , where the first term on the left side is a fidelity term representing the distance between the restored image x and the collected initial image f; A is a PSF of the optical system; the second term and the third term are continuity and sparsity constraints, respectively; ∥·∥ 1 and ∥·∥ 2 are ℓ 1 and ℓ 2 norms respectively; and λ and λ L1 represent the weights of the fidelity and sparsity constraints, respectively; the Hessian matrix continuity prior for the xy-t(z) axis is defined as: R Hessian g = g xx g xy λ t g xt g yx g yy λ t g yt λ t g tx λ t g ty λ t g tt 1 = g xx 1 + g yy 1 + λ t g tt 1 + 2 g xy 1 + 2 λ t g xt 1 + 2 λ t g yt 1

[0048] At S10422, a second reconstruction is performed on the image after the first reconstruction using the reconstruction model.

[0049] At S106, a second set of iterations of deconvolution is performed on the image after the two reconstructions and outputting a super-resolution microscopic image, when the number of iterations reaches the determined number of iterations or half of the determined number of iterations of deconvolution.

[0050] In one embodiment, the second set of iterations of deconvolution is completed when the number of iteration reaches half of the number of iterations determined in step S102 (that is, k / 2 times), and both the number of iterations of pre-deconvolution and the second set of iteration of deconvolution are k / 2 times, just reaching the number of iterations determined in step S102; the specific content of the second setoff iterations of deconvolution may be seen in step S103, which will not be repeated herein; and by performing two sets of iterations of deconvolution (the first set of iteration of pre-deconvolution and the second set of iteration of deconvolution), the resolution is further improved while maintaining the image quality and minimizing artifacts.

[0051] Fig. 2 is an exemplary result diagram showing three-dimensional spatial resolution enhancement of the embodiments of the present disclosure, which can achieve a two-fold enhancement in three-dimensional spatial resolution compared to a conventional confocal mode, using the method of an embodiment of the present disclosure to measure the three-dimensional PSF of a quantum dot (QD525) sample, acquiring a sequence of images with a standard rotating disk confocal microscope, and separately extracting the transverse / axial half-height widths of such a molecular point source. The transverse / axial half-height widths of the conventional confocal mode are 325 / 655 nm, and the conventional fluorescence fluctuation mode is 295 / 510 nm, while the embodiment of the present disclosure is 135 / 334 nm.

[0052] Fig. 3 is an exemplary result diagram showing temporal resolution enhancement of the embodiments of the present disclosure (microtubules labeled with quantum dots in COS-7 cells). The embodiments of the present disclosure can achieve a high-quality super-resolution effect with only 20 frames while maintaining super-resolution, achieving a 50 to 100 times temporal resolution enhancement compared to the conventional fluorescence fluctuation technique SOFI, which requires 1000 frames of image reconstruction to maintain relatively stable performance with the decrease of the signal-to-noise ratio, whereas the embodiment of the present disclosure can reconstruct a two-point structure with high fidelity using only 20 frames of image under all conditions.

[0053] Fig. 4 and Fig. 5 are exemplary results showing the non-parametric reconstruction capability and automated high-throughput reconstruction, respectively. With the advantage of non-parameterization, an automatic parameter estimation method based on rolling FRC is proposed in the embodiments of the present disclosure to realize non-parameterization, to realize automatic super-resolution high-throughput imaging (microtubules labeled with quantum dots in COS-7 cells). The embodiment images a large field of view of 2.0 mm by 1.4 mm, containing more than 2,000 cells, composed of about 2, 400, 000, 000 (2.4 billion) pixels (32.5 nm by 32.5 nm each), and spanning a regional spatial dimension of almost five orders of magnitude. The embodiment requires only about 10 minutes for its reconstruction, compared to about 17 hours for conventional fluorescence fluctuation technique SOFI to achieve similar imaging performance.

[0054] It should be noted that while the method operations of the above embodiments are described in a particular order, this does not require or imply that the operations must be performed in the specific order, or that all illustrated operations must be performed to achieve the desired results. Rather, the depicted steps may change the order of execution. Additionally, or alternatively, certain steps may be omitted; a plurality of steps are combined into one step for execution; and / or one step is decomposed into a plurality of steps for execution.Embodiment 2

[0055] As shown in Fig. 6, the embodiment provides a super-resolution microscopic imaging apparatus including a collection module 601, a wavelet transformation module 602, a determining module 603, a pre-deconvolution module 604, a reconstruction module 605, and a deconvolution module 606, each of which is described in detail as follows: the collection module 601 is configured to record a fluorescent signal sequence of a set of samples to be observed; the wavelet transformation module 602 is configured to estimate a background of the fluorescent signal using a wavelet transformation and to remove background noise; the determining module 603 is configured to determine the number of iterations of deconvolution; the pre-deconvolution module 604 is configured to perform to perform a first set of iterations of pre-deconvolution on each frame of a recorded image in the fluorescent signal sequence, and outputting each frame to obtain output images when a number of iterations of pre-deconvolution reaches half of the number of the determined iterations of deconvolution; the reconstruction module 605 is configured to perform two reconstructions on the output images; and the deconvolution module 606 is configured to perform a second set of iterations of deconvolution on the twice-reconstructed image and outputting the reconstructed image to obtain a super-resolution microscopic image when a number of iterations of deconvolution reaches the number of determined iterations of deconvolution or half of the number of determined iterations of deconvolution.

[0056] It should be noted that the system provided in the embodiment is merely exemplified by the division of the above functional modules. In practical applications, the above functional allocation may be performed by different functional modules according to needs, that is, the internal structure is divided into different functional modules to perform all or part of the functions described above.Embodiment 3

[0057] The embodiment provides a computer device, which may be a computer, as shown in Fig. 7, including a processor 702 connected via a system bus 701, a memory, an input apparatus 703, a display 704, and a network interface 705. The processor is configured to provide computing and control capabilities; the memory includes a non-volatile storage medium 706 and an internal memory 707, the non-volatile storage medium 706 storing an operating system, computer programs, and a database, and the internal memory 707 providing an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The processor 702, when executing the computer programs stored in the memory, implements the super-resolution microscopic imaging method of the above Embodiment 1.Embodiment 4

[0058] The embodiment provides a storage medium, which is a computer-readable storage medium storing computer programs; the computer programs, when executed by a processor, implement the above super-resolution microscopic imaging method of Embodiment 1.

[0059] The present disclosure has the advantages of high flexibility and can be widely coupled to various imaging modalities, such as acoustic microscopy, namely photoacoustic and ultrasonic microscopy imaging technology; it has the advantage of high throughput, and the post-deconvolution of multiple iterations is used to further process the obtained single image, which can achieve a two-fold improvement in three-dimensional spatial resolution. High-quality super-resolution results can be achieved with only 20 frames while maintaining super-resolution, achieving a temporal resolution improvement of 50 to 100 times. With the advantage of non-parameterization, an automatic parameter estimation method based on rolling FRC is proposed to realize non-parameterization, to realize automatic super-resolution high-throughput imaging.

[0060] Note that the computer-readable storage medium of the embodiment may be either a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. The computer-readable storage medium may be, for example but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or element, or any combination thereof. More specific examples of the computer-readable storage medium may include, but are not limited to an electrical connection with one or more wires, a portable computer diskette, a hard disk, a random-access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage element, a magnetic storage element, or any suitable combination thereof.

[0061] In the embodiment, the computer-readable storage medium may be any tangible medium that contains or stores programs that may be used by or in connection with an instruction execution system, apparatus, or element. In the embodiment, the computer-readable signal medium may include data signals embodied in baseband or propagated as part of a carrier wave carrying the computer-readable programs. The propagated data signals may take many forms, including but not limited to, electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium may also be any computer-readable storage medium that can send, propagate, or transmit the programs for use by or in connection with the instruction execution system, apparatus, or element. The computer programs embodied on the computer-readable storage medium may be transmitted over any suitable medium including, but not limited to wire, fiber optic cable, RF (radio frequency), and the like, or any suitable combination thereof.

[0062] The computer-readable storage medium may be used in one or more programming languages or combinations thereof to write computer programs for executing the embodiment, including object-oriented programming languages such as Java, Python, C++, and conventional procedural programming languages, such as C language or similar programming languages. The programs may be executed entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (for example, through the Internet using an Internet service provider).

[0063] The computer program code required for the operation of various portions of the present application may be written in any one or more programming languages, including object-oriented programming languages such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, VB. NET, Python, and the like; conventional procedural programming languages such as C programming language, VisualBasic, Fortran2103, Perl, COBOL2102, PHP, ABAP; dynamic programming languages such as Python, Ruby, and Groovy, or other programming languages, and the like. The program codes may be executed entirely on the user's computer, partly on the user's computer as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server. In the latter case, the remote computer may be connected to the user's computer via any form of network, such as a LAN or WAN, or to an external computer (for example, via the Internet), or in a cloud computing environment, or as a service using for example, Software as a Service (SaaS).

[0064] The invention is defined by the appended claims.

Claims

1. A computer-implemented super-resolution microscopic imaging method, comprising: collecting a fluorescent signal sequence of a set of samples to be observed; determining a number of iterations of deconvolution; performing a first set of iterations of deconvolution on each frame of a recorded images in the fluorescent signal sequence until a number of the first set of iterations of deconvolution reaches half of the number of the determined iterations of deconvolution, and then outputting each frame to obtain output images; performing two reconstructions on the output images to obtain a twice-reconstructed image; and performing a second set of iterations of deconvolution on the twice-reconstructed image until a number of the second set of iterations of deconvolution reaches half of the number of determined iterations of deconvolution, and then outputting a super-resolution microscopic image, characterized in that the determining the number of iterations of deconvolution comprises: using rolling Fourier ring correlation (FRC) to determine, and to select the number of iterations at maximum rolling FRC resolution as the number of iterations of deconvolution, wherein the using rolling FRC to determine, and to select the number of iterations at maximum rolling FRC resolution as the number of iterations of deconvolution specifically comprises: taking the rolling FRC as a function of spatial frequency, and defining discretization of the spatial frequency of a rolling FRC curve to calculate a discrete value of the corresponding spatial frequency; smoothing the rolling FRC curve with noise using an average filter with average window half-width; defining, when the rolling FRC curve is lower than a given threshold, the frequency as an effective cutoff frequency and the resolution as a reciprocal of the effective cutoff frequency, the given threshold representing a maximum spatial frequency of meaningful information outside the random noise; and taking the number of iterations at maximum resolution as the number of iterations of deconvolution; wherein the performing two reconstructions on the output images comprises: performing a first reconstruction on the output images using a fluctuation principle of the fluorescent signal to obtain one or more images after the first reconstruction; and performing a second reconstruction on the one or more images after the first reconstruction using sparsity-continuity joint constraints; wherein the performing a first reconstruction on the output images using a fluctuation principle of the fluorescent signal comprises: acquiring an expression of the fluorescent signal according to a point spread function (PSF) of a microscope, a luminance constant of a fluorescent molecule, and a function of fluctuation of the luminance of the fluorescent molecule over time; wherein the samples are composed of N individual fluorescent molecules at rk, and the expression of the fluorescent signal at r and time t is represented as: F r t = h r − r k ⋅ c k ⋅ ω k t where h, c, and ω represent the PSF of the microscope, the luminance constant of the fluorescent molecular, and the function of fluctuation of luminance of the fluorescent molecular over time, respectively; acquiring, according to the expression of the fluorescent signal, an expression of second-order time cumulant G2 with zero-time delay for each pixel along a t-axis using individual fluctuation characteristics of each fluorescent molecule: G 2 r = δF r t ⋅ δF r t t δF r t = F r t − F r t t where 〈·〉t is a time average function; and expanding the expression of the second-order time cumulant G2 to obtain the following formula: G 2 r = ∑ i , k h r − r i ⋅ h r − r k ⋅ c i ⋅ c k ⋅ δω i t δω k t t and regarding a cross-correlation term of a time cumulant expansion formula as zero when i ≠ k, to make the second-order time cumulant G2 being expressed as the sum of squares of a corresponding luminance constant γ-weighted PSF, as follows: G 2 r = ∑ i h r − r i 2 ⋅ c i 2 ⋅ δω i t 2 t = ∑ i h r − r i 2 ⋅ γ i wherein the performing of the second reconstruction on the one or more images after the first reconstruction using sparsity-continuity joint constraints comprises: generating a reconstruction model, the reconstruction model comprising a first term, a second term, and a third term, the first term being a fidelity term representing a distance between the one or more images after the first reconstruction and the recorded image, the second term representing a continuity constraint of the one or more images after the first reconstruction, and the third term representing a sparsity constraint of the one or more images after the first reconstruction; and performing a second reconstruction on the one or more images after the first reconstruction applying the reconstruction model to the one or more images after the first reconstruction;2. The super-resolution microscopic imaging method according to claim 1, after collecting the fluorescent signal sequence of the set of samples to be observed, the method further comprises the steps of: estimating a background of the fluorescent signal using a wavelet transformation and removing the background, in particular background noise.

3. The super-resolution microscopic imaging method according to claim 2, wherein estimating the background of the fluorescent signal using a wavelet transformation and removing background specifically comprises: estimating the background from a lowest frequency of an input image, such as the recorded image, using wavelet estimation; performing a wavelet inverse transformation on a lowest frequency band to a spatial domain, comparing a result with half the square root of the input image, and combining the two images by keeping a minimum value of each pixel; and using the estimated low-frequency band low-peak background data as a new input image, and performing the wavelet estimation again until a preset number of cycles is reached.

4. The super-resolution microscopic imaging method according to any one of the preceding claims, wherein the first set of iterations of deconvolution comprises: obtaining an iterative formula by iteratively solving maximum likelihood in a spatial domain: where x and y are spatial coordinates; h represents a PSF of a microscope; f represents a real fluorescent signal in the physical world; and g represents a signal acquired by a final microscope; and executing, according to the iterative formula, a calculation of the first set of iterations of deconvolution using an acceleration method based on vector extrapolation: v j = x j + 1 − y j α j + 1 = ∑ v j ⋅ v j − 1 ∑ v j − 1 ⋅ v j − 1 x j + 1 = y j + 1 + α j + 1 ⋅ y j + 1 − y j where g is the reconstructed image constrained by previous prior knowledge; h represents the PSF of the microscope; xj+1 is the image after j+1 iterations; and α is an adaptive acceleration factor.

5. A super-resolution microscopic imaging apparatus, comprising: a collection module (601), configured to collect a fluorescent signal sequence of a set of samples to be observed; a determining module (603), configured to determine the number of iterations of deconvolution, wherein the determining the number of iterations of deconvolution comprises: using rolling Fourier ring correlation (FRC) to determine, and to select the number of iterations at maximum rolling FRC resolution as the number of iterations of deconvolution, wherein the using rolling FRC to determine, and to select the number of iterations at maximum rolling FRC resolution as the number of iterations of deconvolution specifically comprises: taking the rolling FRC as a function of spatial frequency, and defining discretization of the spatial frequency of a rolling FRC curve to calculate a discrete value of the corresponding spatial frequency; smoothing the rolling FRC curve with noise using an average filter with average window half-width; defining, when the rolling FRC curve is lower than a given threshold, the frequency as an effective cutoff frequency and the resolution as a reciprocal of the effective cutoff frequency, the given threshold representing a maximum spatial frequency of meaningful information outside the random noise; and taking the number of iterations at maximum resolution as the number of iterations of deconvolution; a first deconvolution module (604), configured to perform a first set of iterations of deconvolution on each frame of a recorded image in the fluorescent signal sequence until a number of the first set of iterations of deconvolution reaches half of the number of the determined iterations of deconvolution, and then output each frame to obtain output images; a reconstruction module (605), configured to perform two reconstructions on the output images; and a second deconvolution module (606), configured to perform a second set of iterations of deconvolution on the twice-reconstructed image until a number of the second set of iterations of deconvolution reaches half of the number of determined iterations of deconvolution, and then output the reconstructed image to obtain a super-resolution microscopic image; wherein the performing two reconstructions on the output images comprises: performing a first reconstruction on the output images using a fluctuation principle of the fluorescent signal to obtain one or more images after the first reconstruction; and performing a second reconstruction on the one or more images after the first reconstruction using sparsity-continuity joint constraints; wherein the performing a first reconstruction on the output images using a fluctuation principle of the fluorescent signal comprises: acquiring an expression of the fluorescent signal according to a point spread function (PSF) of a microscope, a luminance constant of a fluorescent molecule, and a function of fluctuation of the luminance of the fluorescent molecule over time; wherein the samples are composed of N individual fluorescent molecules at rk, and the expression of the fluorescent signal at r and time t is represented as: F r t = h r − r k ⋅ c k ⋅ ω k t where h, c, and ω represent the PSF of the microscope, the luminance constant of the fluorescent molecular, and the function of fluctuation of luminance of the fluorescent molecular over time, respectively; acquiring, according to the expression of the fluorescent signal, an expression of second-order time cumulant G2 with zero-time delay for each pixel along a t-axis using individual fluctuation characteristics of each fluorescent molecule: G 2 r = δF r t ⋅ δF r t t δF r t = F r t − F r t t where 〈·〉t is a time average function; and expanding the expression of the second-order time cumulant G2 to obtain the following formula: G 2 r = ∑ i , k h r − r i ⋅ h r − r k ⋅ c i ⋅ c k ⋅ δω i t δω k t t and regarding a cross-correlation term of a time cumulant expansion formula as zero when i ≠ k, to make the second-order time cumulant G2 being expressed as the sum of squares of a corresponding luminance constant γ-weighted PSF, as follows: G 2 r = ∑ i h r − r i 2 ⋅ c i 2 ⋅ δω i t 2 t = ∑ i h r − r i 2 ⋅ γ i wherein the performing of the second reconstruction on the one or more images after the first reconstruction using sparsity-continuity joint constraints comprises: generating a reconstruction model, the reconstruction model comprising a first term, a second term, and a third term, the first term being a fidelity term representing a distance between the one or more images after the first reconstruction and the recorded image, the second term representing a continuity constraint of the one or more images after the first reconstruction, and the third term representing a sparsity constraint of the one or more images after the first reconstruction; and performing a second reconstruction on the one or more images after the first reconstruction applying the reconstruction model to the one or more images after the first reconstruction;6. A computer device comprising a processor (702) and a memory for storing processor-executable programs, the processor, when executing the programs stored in the memory, implementing the super-resolution microscopic imaging method according to any one of claims 1 to 4.

7. A storage medium (706) storing programs thereon, the programs, when executed by a processor, implementing the super-resolution microscopic imaging method according to any one of claims 1 to 4.