Fluorescence image signal enhancement system and method based on three-dimensional DNA nanostructure

Abstract

The invention relates to the technical field of image processing, in particular to a fluorescence image signal enhancement system and method based on a three-dimensional DNA nanostructure. The method comprises the following steps: carrying out image identification on a fluorescence image to determine a plurality of diffraction spot areas; in the plurality of diffraction spot regions, distinguishing a first region corresponding to a single-molecule diffraction spot and a second region corresponding to an overlapped diffraction spot; fitting the gray level distribution of the first area to obtain first fitting information; distinguishing a mixed sub-region and a pure sub-region in the second region, and fitting the gray distribution of the pure sub-region by taking the gray distribution of the mixed sub-region as a constraint condition to obtain second fitting information; and according to the first fitting information and the second fitting information, executing a molecular localization task of the fluorescence image to obtain molecular localization information of the fluorescence image. According to the invention, the positioning precision of the fluorescence image in the case of diffraction spot overlapping can be improved.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and more specifically to a fluorescence image signal enhancement system and method based on three-dimensional DNA nanostructures. Background Technology

[0002] Fluorescence imaging technology using three-dimensional DNA nanostructures has become an important development direction in the field of bioimaging in recent years. DNA nanotechnology utilizes the self-assembly properties of DNA molecules to construct highly programmable three-dimensional structures and achieve precise molecular manipulation at the nanoscale. With the rapid development of biomedicine, molecular biology, and artificial intelligence, the demand for high-resolution, high-precision, and high-sensitivity fluorescence imaging technology is constantly increasing. Therefore, optimizing the construction of three-dimensional structures for fluorescence images by combining artificial intelligence technology to enhance fluorescence image signals has gradually become a key development issue.

[0003] In fluorescence imaging, molecular localization is generally performed using diffraction spots for two-dimensional Gaussian fitting, and molecular localization is obtained based on the final fitting result. However, in practice, there may be cases of overlapping diffraction spots. The gray-scale distribution in the overlapping diffraction spots destroys the Gaussian distribution characteristics of a single diffraction spot. In this case, the accuracy of directly locating the overlapping diffraction spots is very poor, and it will cause the localization of one diffraction spot to be lost, resulting in the final three-dimensional image accuracy not meeting the requirements. Summary of the Invention

[0004] The purpose of this invention is to provide a fluorescence image signal enhancement system and method based on three-dimensional DNA nanostructures, which solves the technical problem of low positioning accuracy of three-dimensional structures in fluorescence images in the prior art.

[0005] In a first aspect, one embodiment of the present invention provides a method for enhancing fluorescence image signals based on three-dimensional DNA nanostructures, the method comprising: Image recognition is performed on fluorescence images to identify multiple diffraction spot regions; In the plurality of diffraction spot regions, a first region corresponding to a single-molecule diffraction spot and a second region corresponding to an overlapping diffraction spot are distinguished, wherein the degree of matching between the region shape of the first region and the standard shape of the diffraction spot is greater than a matching threshold, and the degree of matching between the region shape of the second region and the standard shape is less than or equal to the matching threshold. The grayscale distribution of the first region is fitted to obtain first fitting information, wherein the first fitting information is used to determine the center position of the first region; In the second region, a mixed sub-region and a pure sub-region are distinguished, and the gray-level distribution of the mixed sub-region is used as a constraint condition to fit the gray-level distribution of the pure sub-region to obtain second fitting information, wherein the second fitting information is used to determine the center position of the second region; Based on the first fitting information and the second fitting information, a molecular localization task of the fluorescence image is performed to obtain the molecular localization information of the fluorescence image.

[0006] In one embodiment, the image recognition of the fluorescence image to determine multiple diffraction spot regions includes: Based on the grayscale segmentation threshold corresponding to the fluorescence image, the target pixel is identified among all the pixels included in the fluorescence image, wherein the grayscale segmentation threshold indicates the maximum grayscale of the image background of the fluorescence image; Multiple candidate regions are constructed based on the multiple target pixels described above; Among the multiple candidate regions, the candidate regions with a number of pixels greater than a threshold are determined as the diffraction spot regions, thus obtaining the multiple diffraction spot regions.

[0007] In one embodiment, the step of obtaining the grayscale segmentation threshold corresponding to the fluorescence image includes: The gray-level histogram of the fluorescence image is analyzed to determine multiple background gray levels, wherein the occurrence frequency of the background gray levels is less than the occurrence frequency of any gray level other than the multiple background gray levels in the gray-level histogram. Based on all pixels corresponding to the multiple background gray levels in the fluorescence image, multiple background regions are determined. Calculate the average and standard deviation of the gray values ​​of all pixels included in the multiple background regions to obtain the mean gray value and standard deviation of the background gray value; Based on the mean background gray level, the standard deviation of the background gray level, and a preset amplification factor, a gray level segmentation threshold corresponding to the fluorescence image is determined, wherein the amplification factor is used to amplify the numerical contribution of the standard deviation of the background gray level in the gray level segmentation threshold.

[0008] In one embodiment, the roundness of the region shape of the diffraction spot is used to indicate the degree of matching between it and the standard shape of the diffraction spot.

[0009] In one embodiment, distinguishing between the mixed sub-region and the pure sub-region in the second region includes: The second region is segmented to obtain multiple segmented sub-regions; Multiple region dividing lines are defined within each segmented sub-region, wherein the region dividing lines pass through the geometric center of the corresponding segmented sub-region, and the included angle between adjacent region dividing lines is a preset angle. Analyze the structural similarity between the two grandchild regions into which each segmented sub-region is divided by each region dividing line to obtain multiple equal division indices corresponding to each segmented sub-region. Among them, the multiple equal division indices corresponding to each segmented sub-region correspond one-to-one with its multiple region dividing lines. Among the plurality of segmented sub-regions, the segmented sub-region corresponding to the highest average score index is determined as the mixed sub-region, and the other segmented sub-regions besides the mixed sub-region are determined as the pure sub-regions.

[0010] In one embodiment, fitting the grayscale distribution of the pure sub-region to the grayscale distribution of the mixed sub-region as a constraint to obtain second fitting information includes: The grayscale distribution of the pure sub-region is fitted to obtain the fitting information to be determined; The regression distribution of gray values ​​in the mixed sub-region is determined based on the undetermined fitting information. Analyze the difference between the gray-level distribution of the mixed sub-region and the regression distribution to obtain the undetermined fitting deviation value; If the undetermined fitting deviation value is less than the deviation threshold, the undetermined fitting information is determined as the second fitting information; If the undetermined fitting deviation value is greater than or equal to the deviation threshold, the gray distribution of the mixed sub-region is used as a constraint condition to iteratively fit the gray distribution of the pure sub-region to obtain the second fitting information.

[0011] In one embodiment, analyzing the difference between the grayscale distribution of the mixed sub-region and the regression distribution to obtain an undetermined fitting deviation value includes: In all pixels included in the mixed sub-region, the absolute difference between the actual gray value of each pixel in the mixed sub-region and its predicted gray value in the regression distribution is calculated to obtain multiple gray deviation values. Calculate the average of the multiple grayscale deviation values ​​to obtain the undetermined fitting deviation value.

[0012] In one embodiment, the step of using the grayscale distribution of the mixed sub-region as a constraint to iteratively fit the grayscale distribution of the pure sub-region to obtain the second fitting information includes: A purity loss function is constructed based on the gray-level distribution of the pure sub-region, wherein the purity loss function is used to characterize the degree of difference between the gray-level distribution of the pure sub-region and the gray-level regression distribution obtained by fitting it; A hybrid loss function is constructed based on the grayscale distribution of the hybrid sub-region, wherein the hybrid loss function is used to characterize the degree of difference between the grayscale distribution of the hybrid sub-region and the grayscale regression distribution fitted by the pure sub-region; An iterative loss function is constructed based on the pure loss function and the mixed loss function, and the gray-level distribution of the pure sub-region is iteratively fitted based on the iterative loss function to obtain the second fitting information.

[0013] In one embodiment, constructing an iterative loss function based on the pure loss function and the mixed loss function includes: Calculate the sum of the area of ​​the pure sub-region and the area of ​​the mixed sub-region to obtain the sum of the areas; Calculate the ratio of the area of ​​the pure sub-region to the sum of the areas to obtain the purity coefficient; Calculate the ratio of the area of ​​the mixed sub-region to the sum of the areas to obtain the mixing coefficient; The iterative loss function is constructed based on the pure loss function, the mixed loss function, the pure coefficient, and the mixed coefficient.

[0014] Secondly, another embodiment of the present invention provides a fluorescence image signal enhancement system based on a three-dimensional DNA nanostructure, the system comprising: An image recognition module is used to perform image recognition on fluorescence images to determine multiple diffraction spot regions; The region subdivision module is used to distinguish between a first region corresponding to a single-molecule diffraction spot and a second region corresponding to an overlapping diffraction spot in the plurality of diffraction spot regions, wherein the degree of matching between the region shape of the first region and the standard shape of the diffraction spot is greater than a matching threshold, and the degree of matching between the region shape of the second region and the standard shape is less than or equal to the matching threshold. The first fitting module is used to fit the gray-level distribution of the first region to obtain first fitting information, wherein the first fitting information is used to determine the center position of the first region. The second fitting module is used to distinguish between mixed sub-regions and pure sub-regions in the second region, and to fit the gray distribution of the pure sub-regions to the gray distribution of the mixed sub-regions as a constraint condition to obtain second fitting information, wherein the second fitting information is used to determine the center position of the second region. The molecular localization module is used to perform a molecular localization task of the fluorescence image based on the first fitting information and the second fitting information, and obtain the molecular localization information of the fluorescence image.

[0015] Thirdly, in another embodiment of the present invention, an electronic device is provided, including a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the method described in the first aspect.

[0016] Fourthly, in another embodiment of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0017] The present invention has the following beneficial effects: This invention identifies diffraction spot regions in fluorescence images and fits the grayscale distribution of these regions only, effectively suppressing the mixing of grayscale information from the background region. This improves both the fitting effect and efficiency. Furthermore, it distinguishes between a first sub-region corresponding to a single-molecule diffraction spot and a second sub-region corresponding to an overlapping diffraction spot. Within the second sub-region, the grayscale distribution of the mixed sub-region corresponding to the overlapping part is used as a constraint to fit the grayscale distribution of the pure sub-region corresponding to the non-overlapping part. This ensures that the grayscale fitting result of the second sub-region matches its actual grayscale distribution, thereby improving the fitting accuracy of the fluorescence image when facing overlapping diffraction spots. Consequently, it improves the accuracy of molecular localization operations performed on the fluorescence image based on the fitting results, which in turn improves the localization accuracy of the three-dimensional structure of the fluorescence image. Attached Figure Description

[0018] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic flowchart of a fluorescence image signal enhancement method based on three-dimensional DNA nanostructures provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a fluorescence image signal enhancement system based on a three-dimensional DNA nanostructure provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0020] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the fluorescence image signal enhancement system and method based on three-dimensional DNA nanostructures proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0022] The specific details of the fluorescence image signal enhancement system and method based on three-dimensional DNA nanostructures provided by this invention are described below with reference to the accompanying drawings.

[0023] This invention proposes a fluorescence image signal enhancement method based on three-dimensional DNA nanostructures. Please refer to [link / reference]. Figure 1 The diagram illustrates a flowchart of a fluorescence image signal enhancement method based on three-dimensional DNA nanostructures according to an embodiment of the present invention. The method includes: Step S1: Perform image recognition on the fluorescence image to determine multiple diffraction spot regions.

[0024] The process of fluorescence imaging is as follows: by utilizing the photophysical properties of fluorescent molecules, specific optical signals are generated through the excitation-emission process (meaning that fluorescent molecules can absorb light of a specific wavelength and emit light of a longer wavelength), so as to convert invisible molecular information into visible optical signals. Then, the optical signals are accurately captured by the imaging system and converted into analyzable digital images.

[0025] In this invention, the fluorescence image specifically refers to a digital image acquired by processing fluorescent molecules using fluorescence imaging technology.

[0026] For example, the process of acquiring a fluorescence image can be as follows: Tetrahedral structures were constructed using the caDNAno tool. Four complementary strands self-assembled to form a tetrahedron with a side length of 5-20 nm. Single strand extensions were reserved at the vertices for fluorescent labeling. The GC content of each strand was 40-60% to avoid hairpin structures.

[0027] DNA strands were synthesized using a solid-phase phosphoramide method to ensure a length of 50-200 nt. Long strands >100 nt were separated by 20% polyacrylamide gel electrophoresis, and the target band was removed and eluted for recovery. Short strands <100 nt were synthesized by reversed-phase HPLC, and the main peak was collected and lyophilized for concentration.

[0028] Mix the various chains in a 1:1 molar ratio and add 1×TAE-Mg. 2+ The buffer solution consisted of 40 mM Tris, 20 mM acetic acid, 1 mM EDTA, and 12.5 mM MgCl2; then a gradient cooling process was used, reducing the temperature to 20 °C at a rate of 0.1 °C / second.

[0029] The chosen dye is Cy5, with excitation wavelengths of 550 / 650 nm and emission wavelengths of 570 / 670 nm, suitable for wide-field and confocal imaging. End labeling is employed, involving the introduction of amino groups at the 5' or 3' end of the DNA strand, coupled with dye via succinimide ester (NHS). The target concentration of the fluorescent label is diluted (e.g., from 100 nM to 10 nM) to reduce the number of molecules per unit area.

[0030] Immobilization was performed using glutaraldehyde. Specifically, 10 μL of the assembly solution was mixed with 10 μL of 4% glutaraldehyde and incubated at room temperature for 10 minutes. The reaction was terminated by adding 10 μL of 1 M Tris-HCl and centrifuged at 14000×g for 10 minutes to remove free glutaraldehyde. The solution was then resuspended in PBS containing 0.1% Triton X-100 and incubated on ice for 5 minutes to enhance dye penetration.

[0031] Immerse the glass coverslip in an ethanol solution containing 2% 3-aminopropyltriethoxysilane for 30 minutes, wash with water and dry; add 1 mg / mL PEG-SH solution and incubate at room temperature for 2 hours to block the remaining amino groups; block with PBS containing 1% BSA for 30 minutes to further reduce nonspecific binding.

[0032] Based on the above steps, fluorescent photon signals of the DNA structure are obtained, and the fluorescent photon signals received by the pre-set detector are then converted into pixelated digital images (i.e., the aforementioned fluorescent images).

[0033] The optical path configuration corresponding to the above detectors may include: Objective lens: 60×NA 1.42 oil immersion objective lens, lateral resolution ~150 nm, axial resolution ~500 nm; Excitation source: Cy5, corresponding to a 642nm laser, with a power of 5-20 mW / cm²; Filter group: Excitation filter is 642 / 10 nm (Cy5); Emission filter is 680 / 30 nm (Cy5). Sampling frequency: 15Hz.

[0034] In the actual image acquisition process, a 405nm laser can be used to briefly irradiate the sample to randomly activate a few dyes; then switch to a 561nm or 642nm laser for imaging until the dyes are bleached. During this process, an sCMOS camera is used to acquire fluorescence images of one-dimensional DNA nanostructures by back-illumination.

[0035] The sCMOS camera can be set to an exposure time of 100ms to ensure that each molecule generates sufficient charge, resulting in a grayscale value in the digital image that is greater than three standard deviations of the grayscale value of the background area, while avoiding dark current accumulation. The gain is set to 1-5× to avoid signal saturation, ensuring that the grayscale value does not exceed 80% of the ADC's maximum value. The pixel size is set to 6.5μm, and a 1.6× teleconverter is used to meet Nyquist sampling requirements.

[0036] It should be understood that there are multiple fluorescence images, and each fluorescence image consists of multiple pixels with gray values.

[0037] Specifically, the image recognition of the fluorescence image to determine multiple diffraction spot regions includes: Based on the grayscale segmentation threshold corresponding to the fluorescence image, the target pixel is identified among all the pixels included in the fluorescence image, wherein the grayscale segmentation threshold indicates the maximum grayscale of the image background of the fluorescence image; Multiple candidate regions are constructed based on the multiple target pixels described above; Among the multiple candidate regions, the candidate regions with a number of pixels greater than a threshold are determined as the diffraction spot regions, thus obtaining the multiple diffraction spot regions.

[0038] In a fluorescence image, there are fluorescent regions and background regions. The fluorescent region is the region corresponding to the DNA nanostructure, which is also the image region where the diffraction spot to be analyzed is located. Generally speaking, the gray value of the pixels in the fluorescent region is significantly higher than that in the background region. Based on this, the present invention uses a gray-scale segmentation threshold to conveniently distinguish the pixels corresponding to the fluorescent region (i.e., the aforementioned target pixels) from the pixels corresponding to the background region (i.e., other pixels besides the aforementioned target pixels), so as to quickly and accurately identify multiple diffraction spot regions included in the fluorescence image.

[0039] The reason for setting a quantity threshold is to filter out interference from discrete pixels introduced by extreme noise, so as to ensure the accuracy of the determined diffraction spot area.

[0040] The above quantity threshold can be set to 5 based on experience. In application, the above quantity threshold can also be adjusted adaptively according to actual needs.

[0041] The candidate region can be understood as a region consisting only of target pixels.

[0042] Furthermore, the step of obtaining the grayscale segmentation threshold corresponding to the fluorescence image includes: The gray-level histogram of the fluorescence image is analyzed to determine multiple background gray levels, wherein the occurrence frequency of the background gray levels is less than the occurrence frequency of any gray level other than the multiple background gray levels in the gray-level histogram. Based on all pixels corresponding to the multiple background gray levels in the fluorescence image, multiple background regions are determined. Calculate the average and standard deviation of the gray values ​​of all pixels included in the multiple background regions to obtain the mean gray value and standard deviation of the background gray value; Based on the mean background gray level, the standard deviation of the background gray level, and a preset amplification factor, a gray level segmentation threshold corresponding to the fluorescence image is determined, wherein the amplification factor is used to amplify the numerical contribution of the standard deviation of the background gray level in the gray level segmentation threshold.

[0043] For cases where the gray-level distribution of the background region in a fluorescence image matches a normal distribution, the gray-level histogram is analyzed to identify the multiple background gray-levels with the lowest frequency (meaning their frequency is less than the frequency of any gray-level in the histogram other than the multiple background gray-levels). Based on this, multiple corresponding background regions are identified in the fluorescence image. The gray-level segmentation threshold corresponding to the fluorescence image is calculated based on the average and standard deviation of the gray-level values ​​of all pixels in the multiple background regions. The matching gray-level segmentation threshold is adaptively determined according to the actual gray-level distribution of each fluorescence image to ensure the accuracy of the diffraction spot region identified accordingly.

[0044] After determining multiple background gray levels, multiple background undetermined regions can be constructed based on all pixels corresponding to these gray levels in the fluorescence image (each background undetermined region only includes pixels corresponding to the gray levels in the fluorescence image). Then, among these undetermined regions, the regions with the largest areas (i.e., the number of pixels they contain) are ranked from highest to lowest to be identified as the multiple background regions (the total number of background regions is less than the total number of undetermined background regions). This process avoids interference from extreme noise and ensures the accuracy of the determined background regions.

[0045] In one example, the total number of the aforementioned background grayscale levels can be 20, and the total number of the aforementioned background regions can be 3.

[0046] For example, the grayscale segmentation threshold corresponding to the fluorescence image It can be represented as: in, This represents the average grayscale value of the background. This represents the standard deviation of the background grayscale. This indicates the magnification factor (the magnification factor is at least greater than or equal to 3, and can be set to 4 based on experience).

[0047] Step S2: In the plurality of diffraction spot regions, distinguish the first region corresponding to the single-molecule diffraction spot from the second region corresponding to the overlapping diffraction spot.

[0048] Wherein, the degree of matching between the region shape of the first region and the standard shape of the diffraction spot is greater than the matching threshold, and the degree of matching between the region shape of the second region and the standard shape is less than or equal to the matching threshold.

[0049] It should be noted that in fluorescence imaging, only a few individual molecules are typically stained, resulting in sparse diffraction spots in the formed fluorescence image. However, in reality, adjacent molecules inevitably become stained, and their diffraction spots may be adjacent or even overlap in the image. Therefore, within the multiple diffraction spot regions of the fluorescence image, there will be a first region corresponding to a single-molecule diffraction spot and a second region corresponding to an overlapping diffraction spot. By distinguishing between these two, appropriate fitting processing can be performed on their respective grayscale distributions to ensure accurate positioning of the center of multiple diffraction spot regions in subsequent processes.

[0050] Specifically, since the standard shape of a diffraction spot is circular or elliptical, the degree of matching between the area shape of the diffraction spot and the standard shape is indicated by the roundness of the area shape.

[0051] In applications, the edges of each diffraction spot region can be extracted, and then the circularity can be calculated based on the length of the extracted region's edge (i.e., the perimeter of the diffraction spot region) and the area of ​​the corresponding diffraction spot region.

[0052] For example, the circularity of the diffraction spot region can be expressed as: in, This represents the area of ​​the diffraction spot region. This indicates the perimeter of the diffraction spot region.

[0053] In this example, the matching threshold can be set to 0.5 based on experience.

[0054] It should be noted that the distribution of staining molecules during the fluorescence shaping process is usually very sparse, and the number of diffraction spots formed at one time is basically less than 10. In addition, the molecular concentration is diluted, so the overlap of diffraction spots is relatively rare, and there is almost no complete overlap. Therefore, the complete overlap of diffraction spots is not considered in this invention.

[0055] Step S3: Fit the grayscale distribution of the first region to obtain the first fitting information.

[0056] The first fitting information is used to determine the center position of the first region.

[0057] Specifically, the process of fitting the grayscale distribution of the first region is as follows: In the single-molecule diffraction spot region (i.e., the first region), its geometric center is first determined, and then a two-dimensional Cartesian coordinate system is constructed with this geometric center as the origin (to reduce the number of iterations in the subsequent fitting process and improve computational efficiency), with the horizontal axis as the x-axis and the vertical axis as the y-axis.

[0058] The pre-constructed two-dimensional Gaussian function for fitting is as follows: in, These are the five parameters to be fitted. The coordinates of the molecular true center of a single-molecule diffraction spot region. These are the horizontal and vertical half-widths of the Gaussian distribution, respectively. The peak intensity is a Gaussian distribution; and in the formula... This represents the pixel coordinates within the traversed single-molecule diffraction spot region.

[0059] Two-dimensional Gaussian fitting is sensitive to the initial values ​​of the fitting parameters. If the initial values ​​deviate too far from the true values, it will cause the iteration to fail to converge or get stuck in a local optimum. Therefore, it is necessary to roughly estimate the initial parameters by using the intensity distribution of the diffraction spot to provide a reasonable starting point for subsequent iterations.

[0060] To avoid the above problems, the coordinates of the geometric center can be used as the initial values ​​of the true center coordinates of the molecule, and the maximum gray value in the diffraction spot region can be used as the initial value of the Gaussian distribution peak intensity. Light intensity profiles passing through the geometric center can be extracted along the x-axis (horizontal) and y-axis (vertical). For each profile, two pixels whose light intensity drops to half of the peak value can be found. The average of the distance between the two pixels found in each profile is the initial value of the horizontal half-width / vertical half-width.

[0061] The above process determines the initial values ​​of the five parameters for two-dimensional Gaussian fitting of the diffraction spot region. At this point, in order to accurately locate the true coordinates of the molecular center of the diffraction spot, iterative fitting is required. That is, starting from the initial values, the parameter values ​​of the five parameters are iterated, and all pixels in the diffraction spot region are traversed in each iteration. The fitting effect of the corresponding iteration is determined based on the fitting residuals of all pixels in the corresponding iteration, and the parameter values ​​used when the fitting effect is the best (that is, when the fitting residuals of all pixels in the corresponding iteration are the smallest) are determined as the target values.

[0062] First, within each diffraction spot region, the fitting residual corresponding to the pixel within the diffraction spot region is determined to be the absolute difference between the actual gray value of the pixel and the fitted regression value of the pixel. Then, the sum of squares of the fitting residuals of all pixels within a diffraction spot region is used as the objective function for the current iteration of fitting.

[0063] Then, by iteratively adjusting the values ​​of the five parameters (using the least squares method), the parameter values ​​corresponding to the convergence of the objective function are determined, and this information is used as the fitting information (when the diffraction spot region being processed is the first region, the resulting fitting information is the first fitting information). The fitting information contains... The corresponding parameter value is the two-dimensional coordinate of the center position of the corresponding diffraction spot region.

[0064] Step S4: In the second region, distinguish between the mixed sub-region and the pure sub-region, and use the gray-level distribution of the mixed sub-region as a constraint condition to fit the gray-level distribution of the pure sub-region to obtain the second fitting information.

[0065] The second fitting information is used to determine the center position of the second region.

[0066] Specifically, distinguishing between the mixed sub-region and the pure sub-region in the second region includes: The second region is segmented to obtain multiple segmented sub-regions; Multiple region dividing lines are defined within each segmented sub-region, wherein the region dividing lines pass through the geometric center of the corresponding segmented sub-region, and the included angle between adjacent region dividing lines is a preset angle. Analyze the structural similarity between the two grandchild regions into which each segmented sub-region is divided by each region dividing line to obtain multiple equal division indices corresponding to each segmented sub-region. Among them, the multiple equal division indices corresponding to each segmented sub-region correspond one-to-one with its multiple region dividing lines. Among the plurality of segmented sub-regions, the segmented sub-region corresponding to the highest average score index is determined as the mixed sub-region, and the other segmented sub-regions besides the mixed sub-region are determined as the pure sub-regions.

[0067] It should be noted that, as mentioned above, the number of diffraction spots in a fluorescence image is usually small, and the stacking of three or more diffraction spots is extremely rare. Therefore, the diffraction spot overlap situation addressed in this invention is mainly the case of two diffraction spots stacking.

[0068] For example, the above image segmentation operation can be performed based on the Simple Linear Iterative Clustering (SLIC) algorithm, and the number of segmented sub-regions obtained is three (that is, the number of superpixels corresponding to the SLIC algorithm is set to three).

[0069] The aforementioned mixed sub-regions and pure sub-regions can be understood as: in the second region, sub-regions with overlapping diffraction spots and sub-regions without overlapping diffraction spots. Specifically, in a second region, the number of mixed sub-regions is one, and the number of pure sub-regions is two; that is, the number of segmented sub-regions obtained from image segmentation in the second region is three.

[0070] In the above setup, after dividing the second region into three sub-regions, the region dividing lines are used to divide each sub-region equally. By analyzing the structural similarity between the two sub-regions after equal division, the mixed sub-regions and pure sub-regions can be accurately distinguished from the three sub-regions.

[0071] It should be understood that when two diffraction spots overlap, the shape of the overlapping region is relatively regular, while the shape of the non-overlapping region is usually more irregular. Therefore, when dividing each sub-region using a dividing line passing through the geometric center of the sub-region, the structural consistency of the two sub-regions formed after dividing the overlapping region is significantly higher than that of the two sub-regions formed after dividing the non-overlapping region.

[0072] For example, the preset angle can be set to 10 degrees or 20 degrees. For instance, when the preset angle is set to 10 degrees, 36 region dividing lines will be formed within a segmented sub-region.

[0073] In this invention, the structural similarity between two grandchild regions can be quantified based on the Structural Similarity Index (SSIM) between the two grandchild regions.

[0074] In the application, after distinguishing one mixed sub-region and two pure sub-regions from each second region, the area ratio of the mixed sub-region and the two pure sub-regions can be calculated separately. Then, the mean of the two area ratios is calculated. If the mean of the area ratio is greater than 0.7, it is judged as excessive overlap. At this time, it can be considered that the center of the two diffraction spots that make up the overlapping diffraction spots is likely located in the mixed sub-region. However, the gray-scale distribution of the mixed sub-region is difficult to directly fit with a two-dimensional Gaussian, which will lead to the center positioning accuracy of the corresponding two diffraction spots being unsatisfactory. Therefore, for such excessively overlapping diffraction spots, they are directly removed without further positioning to avoid affecting the overall positioning accuracy.

[0075] Conversely, if the mean of the above area ratios is less than or equal to 0.7, it is determined that there is no risk of excessive overlap. Subsequently, the gray-scale distribution of the mixed sub-region is used as a constraint condition, and the gray-scale distribution of the pure sub-region is fitted to obtain the corresponding second fitting information.

[0076] Further, the step of using the grayscale distribution of the mixed sub-region as a constraint condition to fit the grayscale distribution of the pure sub-region to obtain second fitting information includes: The grayscale distribution of the pure sub-region is fitted to obtain the fitting information to be determined; The regression distribution of gray values ​​in the mixed sub-region is determined based on the undetermined fitting information. Analyze the difference between the gray-level distribution of the mixed sub-region and the regression distribution to obtain the undetermined fitting deviation value; If the undetermined fitting deviation value is less than the deviation threshold, the undetermined fitting information is determined as the second fitting information; If the undetermined fitting deviation value is greater than or equal to the deviation threshold, the gray distribution of the mixed sub-region is used as a constraint condition to iteratively fit the gray distribution of the pure sub-region to obtain the second fitting information.

[0077] The process of fitting the gray-level distribution of the pure sub-region can be the same as the process of fitting the gray-level distribution of the first region mentioned above. To avoid repetition, it will not be described again.

[0078] The aforementioned undetermined fitting information includes the parameter values ​​of the five parameters of the aforementioned two-dimensional Gaussian function for the corresponding pure sub-region when the objective function converges. It should be understood that since there are two pure sub-regions, there are also two sets of undetermined fitting information. In this case, the regression distribution of the grayscale values ​​of the mixed sub-region includes the predicted grayscale values ​​of each pixel within the mixed sub-region. The acquisition process is as follows: based on the two sets of undetermined fitting information, the fitted regression value of each pixel within the mixed sub-region is calculated, and the sum of the two fitted regression values ​​for each pixel within the mixed sub-region is determined as the predicted grayscale value of that pixel.

[0079] The step of analyzing the difference between the gray-level distribution of the mixed sub-region and the regression distribution to obtain the undetermined fitting deviation value includes: In all pixels included in the mixed sub-region, the absolute difference between the actual gray value of each pixel in the mixed sub-region and its predicted gray value in the regression distribution is calculated to obtain multiple gray deviation values. Calculate the average of the multiple grayscale deviation values ​​to obtain the undetermined fitting deviation value.

[0080] For example, the undetermined fitting deviation value It can be represented as: Where m represents the total number of pixels included in the blended sub-region. This represents the actual grayscale value of the i-th pixel included in the blended sub-region. This represents the predicted grayscale value of the i-th pixel included in the mixed sub-region.

[0081] In the above setup, the computational complexity is reduced and the computational efficiency is improved by fitting the two pure sub-regions separately. Then, the undetermined fitting deviation value is calculated to quantify the fitting effect (the larger the undetermined fitting deviation value, the worse the fitting effect, and vice versa). By comparing the value between the undetermined fitting deviation value and the deviation threshold, it is determined whether the fitting accuracy of fitting the two pure sub-regions separately meets the requirements. If it does not meet the requirements (meaning the undetermined fitting deviation value is greater than or equal to the deviation threshold), a more accurate second fitting information is obtained by performing a complex fitting of the two pure sub-regions together, so as to balance the requirements of computational efficiency and computational accuracy.

[0082] For example, the above deviation threshold can be set to 0.5 or 1.

[0083] Furthermore, the step of using the grayscale distribution of the mixed sub-region as a constraint condition to iteratively fit the grayscale distribution of the pure sub-region to obtain the second fitting information includes: A purity loss function is constructed based on the gray-level distribution of the pure sub-region, wherein the purity loss function is used to characterize the degree of difference between the gray-level distribution of the pure sub-region and the gray-level regression distribution obtained by fitting it; A hybrid loss function is constructed based on the grayscale distribution of the hybrid sub-region, wherein the hybrid loss function is used to characterize the degree of difference between the grayscale distribution of the hybrid sub-region and the grayscale regression distribution fitted by the pure sub-region; An iterative loss function is constructed based on the pure loss function and the mixed loss function, and the gray-level distribution of the pure sub-region is iteratively fitted based on the iterative loss function to obtain the second fitting information.

[0084] The step of constructing an iterative loss function based on the pure loss function and the mixed loss function includes: Calculate the sum of the area of ​​the pure sub-region and the area of ​​the mixed sub-region to obtain the sum of the areas; Calculate the ratio of the area of ​​the pure sub-region to the sum of the areas to obtain the purity coefficient; Calculate the ratio of the area of ​​the mixed sub-region to the sum of the areas to obtain the mixing coefficient; The iterative loss function is constructed based on the pure loss function, the mixed loss function, the pure coefficient, and the mixed coefficient.

[0085] The process of iteratively fitting the gray distribution of the pure sub-region to the gray distribution of the mixed sub-region using the gray distribution of the mixed sub-region as a constraint can be understood as follows: first, constructing loss functions for the two pure sub-regions respectively, and using the fitting parameters of the two pure sub-regions to complete the construction of the loss function for the mixed sub-region; then, summing the above three loss functions to construct an iterative loss function, so that the fitting information obtained by the final iteration can not only match the gray distribution of the two pure sub-regions, but also match the gray distribution of the mixed sub-region, thereby realizing the association between the two pure sub-regions and the mixed sub-region during the fitting process.

[0086] The purity loss function can be understood as the objective function for the corresponding pure sub-region. The definition of the objective function can be found in the previous explanation, and will not be repeated here to avoid repetition. Similarly, the mixture loss function can be understood as the objective function for the mixture sub-region (the fitted regression value of each pixel in the mixture loss function is the aforementioned predicted gray value).

[0087] For example, if the two purity loss functions corresponding to the two purity sub-regions are respectively... , The mixing loss function for the mixed sub-region is .

[0088] A pure sub-region and a mixed sub-region form a temporary diffraction spot region. The overlapping area of ​​the two temporary diffraction spot regions is SC, and the non-overlapping areas are SA and SB, respectively. Therefore, the area proportion of the mixed sub-region in the two temporary diffraction spot regions (i.e., the aforementioned mixing coefficient) can be determined as follows: , The area proportions of the two pure sub-regions in the two temporary diffraction spot regions (i.e., the aforementioned purity coefficients) are respectively... , .

[0089] At this point, the loss function corresponding to a temporary diffraction spot region is: The corresponding loss function in another temporary diffraction spot region is: Combining the above two points yields the final loss function (i.e., the iterative loss function): Based on the aforementioned iterative loss function, the two sets of parameter values ​​for the two pure sub-regions are jointly iteratively updated. From all the fitting results, the two sets of parameter values ​​corresponding to the convergence of the final loss function are selected and concatenated to form the aforementioned second fitting information. The two center coordinates indicated by the two sets of parameter values ​​corresponding to the convergence of the final loss function are then considered. This indicates the center position of the two diffraction spots corresponding to the second region.

[0090] Step S5: Based on the first fitting information and the second fitting information, perform the molecular localization task of the fluorescence image to obtain the molecular localization information of the fluorescence image.

[0091] Based on the first and second fitting information mentioned above, the center position of each diffraction spot in each fluorescence image can be determined (two-dimensional form, lacking Z-axis height information).

[0092] After determining the two-dimensional coordinates of the center of each diffraction spot, the sample corresponding to the fluorescence image can be fixed on a glass slide. The slide can be moved along the Z-axis using a precision piezoelectric stage, with a movement range of (-500, 500) nm to cover the target imaging region, and a movement step size of 5 nm. At each Z-step, diffraction spot images of the fluorescent microspheres are acquired, and two-dimensional Gaussian fitting is performed on the diffraction spots at each step (the aforementioned process can be used to obtain accurate fitting results at each step). The half-width ratio r' and eccentricity e' of the diffraction spot (ellipse) are then extracted. The half-width ratio r' is the ratio of the horizontal half-width to the vertical half-width of the Gaussian fitting. The relative magnitude of r' to 1 reflects the stretching direction of the ellipse, i.e., when r'>1, the major axis of the ellipse is in the X direction, and when r'<1, the major axis of the ellipse is in the Y direction. The larger the eccentricity e', the flatter the ellipse, and it is positively correlated with the absolute value of Z.

[0093] Using Z as the abscissa and e' and r' as the ordinates, a calibration equation is obtained by polynomial fitting, where the fitting function is a quadratic function Z=a×e'^2+b×e'+c; finally, the calibration equation obtained by fitting is the relationship curve between the diffraction spot ellipse parameters and Z (that is, determining the parameter values ​​of parameters a, b, and c).

[0094] For each molecular diffraction spot obtained from the two-dimensional localization (the two-dimensional coordinates are obtained through the aforementioned first and second fitting information), a two-dimensional elliptical Gaussian fitting is performed to obtain the half-width ratio r and the eccentricity e.

[0095] When the half-width ratio r > 1, substitute the eccentricity e into the calibration equation Z = a × e^2 + b × e + c; when the half-width ratio r < 1, substitute the eccentricity e into the calibration equation Z = -(a × e^2 + b × e + c).

[0096] Repeat the above process to obtain the Z-axis information of the center of each fluorescent molecule (i.e., diffraction spot) in the fluorescence image.

[0097] By using the above steps, the center coordinates (x, y) and Z-axis coordinates (z) of each fluorescent molecule are obtained through two-dimensional localization. By integrating the two coordinates, the three-dimensional coordinates (x, y, z) of each fluorescent molecule can be formed. The set of information formed by the three-dimensional coordinates of all fluorescent molecules in the fluorescence image is the molecular localization information of the fluorescence image.

[0098] In summary, this invention effectively suppresses the mixing of grayscale information from the background region by identifying diffraction spot regions in fluorescence images and fitting only the grayscale distribution of these regions. This improves both the fitting effect and efficiency. Furthermore, it distinguishes between a first sub-region corresponding to a single-molecule diffraction spot and a second sub-region corresponding to an overlapping diffraction spot. Within the second sub-region, the grayscale distribution of the mixed sub-region corresponding to the overlapping part is used as a constraint condition to fit the grayscale distribution of the pure sub-region corresponding to the non-overlapping part. This ensures that the grayscale fitting result of the second sub-region matches its actual grayscale distribution, thereby improving the fitting accuracy of the fluorescence image when facing overlapping diffraction spots. Consequently, it improves the accuracy of molecular localization operations performed on the fluorescence image based on the fitting results, which in turn improves the localization accuracy of the three-dimensional structure of the fluorescence image.

[0099] This invention proposes a fluorescence image signal enhancement system based on three-dimensional DNA nanostructures. Please refer to [link / reference]. Figure 2 The diagram illustrates a schematic of a fluorescence image signal enhancement system 200 based on a three-dimensional DNA nanostructure according to an embodiment of the present invention. The system includes: Image recognition module 201 is used to perform image recognition on fluorescence images to determine multiple diffraction spot regions; The region subdivision module 202 is used to distinguish between a first region corresponding to a single-molecule diffraction spot and a second region corresponding to an overlapping diffraction spot in the plurality of diffraction spot regions, wherein the degree of matching between the region shape of the first region and the standard shape of the diffraction spot is greater than a matching threshold, and the degree of matching between the region shape of the second region and the standard shape is less than or equal to the matching threshold. The first fitting module 203 is used to fit the gray distribution of the first region to obtain first fitting information, wherein the first fitting information is used to determine the center position of the first region. The second fitting module 204 is used to distinguish between mixed sub-regions and pure sub-regions in the second region, and to fit the gray distribution of the pure sub-regions with the gray distribution of the mixed sub-regions as a constraint condition to obtain second fitting information, wherein the second fitting information is used to determine the center position of the second region. The molecular localization module 205 is used to perform a molecular localization task of the fluorescence image based on the first fitting information and the second fitting information to obtain the molecular localization information of the fluorescence image.

[0100] 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 of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the fluorescence image signal enhancement system based on three-dimensional DNA nanostructures and the fluorescence image signal enhancement method based on three-dimensional DNA nanostructures provided in the above embodiments belong to the same concept. The specific implementation process is detailed in the method embodiments and will not be repeated here.

[0101] This invention also provides an electronic device. Please refer to [link to relevant documentation]. Figure 3 The electronic device may include a processor 301, a memory 302, and a program 3021 stored in the memory 302 and capable of running on the processor 301.

[0102] When program 3021 is executed by processor 301, it can achieve the following: Figure 1 Any steps in the corresponding method embodiments and the achievement of the same beneficial effects will not be repeated here.

[0103] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by hardware related to program instructions, and the program can be stored in a readable medium.

[0104] This invention also provides a readable storage medium storing a computer program, which, when executed by a processor, can perform the above-described functions. Figure 1 Any step in the corresponding method embodiment can achieve the same technical effect, and will not be repeated here to avoid repetition.

[0105] The computer-readable storage medium of this invention can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0106] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0107] The program code contained on the storage medium can be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0108] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or terminal. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0109] This invention also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned steps to achieve the fluorescence image signal enhancement method based on three-dimensional DNA nanostructures provided in the above embodiments.

[0110] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0111] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for enhancing fluorescence image signals based on three-dimensional DNA nanostructures, characterized in that, The method includes: Image recognition is performed on fluorescence images to identify multiple diffraction spot regions; In the plurality of diffraction spot regions, a first region corresponding to a single-molecule diffraction spot and a second region corresponding to an overlapping diffraction spot are distinguished, wherein the degree of matching between the region shape of the first region and the standard shape of the diffraction spot is greater than a matching threshold, and the degree of matching between the region shape of the second region and the standard shape is less than or equal to the matching threshold. The grayscale distribution of the first region is fitted to obtain first fitting information, wherein the first fitting information is used to determine the center position of the first region; In the second region, a mixed sub-region and a pure sub-region are distinguished, and the gray-level distribution of the mixed sub-region is used as a constraint condition to fit the gray-level distribution of the pure sub-region to obtain second fitting information, wherein the second fitting information is used to determine the center position of the second region; Based on the first fitting information and the second fitting information, a molecular localization task of the fluorescence image is performed to obtain the molecular localization information of the fluorescence image.

2. The fluorescence image signal enhancement method based on three-dimensional DNA nanostructures according to claim 1, characterized in that, The image recognition of the fluorescence image to determine multiple diffraction spot regions includes: Based on the grayscale segmentation threshold corresponding to the fluorescence image, the target pixel is identified among all the pixels included in the fluorescence image, wherein the grayscale segmentation threshold indicates the maximum grayscale of the image background of the fluorescence image; Multiple candidate regions are constructed based on the multiple target pixels described above; Among the multiple candidate regions, the candidate regions with a number of pixels greater than a threshold are determined as the diffraction spot regions, thus obtaining the multiple diffraction spot regions.

3. The fluorescence image signal enhancement method based on three-dimensional DNA nanostructures according to claim 2, characterized in that, The steps for obtaining the grayscale segmentation threshold corresponding to the fluorescence image include: The gray-level histogram of the fluorescence image is analyzed to determine multiple background gray levels, wherein the occurrence frequency of the background gray levels is less than the occurrence frequency of any gray level other than the multiple background gray levels in the gray-level histogram. Based on all pixels corresponding to the multiple background gray levels in the fluorescence image, multiple background regions are determined. Calculate the average and standard deviation of the gray values ​​of all pixels included in the multiple background regions to obtain the mean gray value and standard deviation of the background gray value; Based on the mean background gray level, the standard deviation of the background gray level, and a preset amplification factor, a gray level segmentation threshold corresponding to the fluorescence image is determined, wherein the amplification factor is used to amplify the numerical contribution of the standard deviation of the background gray level in the gray level segmentation threshold.

4. The fluorescence image signal enhancement method based on three-dimensional DNA nanostructures according to claim 1, characterized in that, The roundness of the region shape of the diffraction spot is used to indicate the degree of matching between it and the standard shape of the diffraction spot.

5. The fluorescence image signal enhancement method based on three-dimensional DNA nanostructures according to claim 1, characterized in that, The step of distinguishing between mixed sub-regions and pure sub-regions in the second region includes: The second region is segmented to obtain multiple segmented sub-regions; Multiple region dividing lines are defined within each segmented sub-region, wherein the region dividing lines pass through the geometric center of the corresponding segmented sub-region, and the included angle between adjacent region dividing lines is a preset angle. Analyze the structural similarity between the two grandchild regions into which each segmented sub-region is divided by each region dividing line to obtain multiple equal division indices corresponding to each segmented sub-region. Among them, the multiple equal division indices corresponding to each segmented sub-region correspond one-to-one with its multiple region dividing lines. Among the plurality of segmented sub-regions, the segmented sub-region corresponding to the highest average score index is determined as the mixed sub-region, and the other segmented sub-regions besides the mixed sub-region are determined as the pure sub-regions.

6. The fluorescence image signal enhancement method based on three-dimensional DNA nanostructures according to claim 1, characterized in that, The step of fitting the grayscale distribution of the pure sub-region to the grayscale distribution of the mixed sub-region as a constraint to obtain second fitting information includes: The grayscale distribution of the pure sub-region is fitted to obtain the fitting information to be determined; The regression distribution of gray values ​​in the mixed sub-region is determined based on the undetermined fitting information. Analyze the difference between the gray-level distribution of the mixed sub-region and the regression distribution to obtain the undetermined fitting deviation value; If the undetermined fitting deviation value is less than the deviation threshold, the undetermined fitting information is determined as the second fitting information; If the undetermined fitting deviation value is greater than or equal to the deviation threshold, the gray distribution of the mixed sub-region is used as a constraint condition to iteratively fit the gray distribution of the pure sub-region to obtain the second fitting information.

7. The fluorescence image signal enhancement method based on three-dimensional DNA nanostructures according to claim 6, characterized in that, The analysis of the difference between the gray-level distribution of the mixed sub-region and the regression distribution to obtain the undetermined fitting deviation value includes: In all pixels included in the mixed sub-region, the absolute difference between the actual gray value of each pixel in the mixed sub-region and its predicted gray value in the regression distribution is calculated to obtain multiple gray deviation values. Calculate the average of the multiple grayscale deviation values ​​to obtain the undetermined fitting deviation value.

8. The fluorescence image signal enhancement method based on three-dimensional DNA nanostructures according to claim 6, characterized in that, The step of using the grayscale distribution of the mixed sub-region as a constraint condition to iteratively fit the grayscale distribution of the pure sub-region to obtain the second fitting information includes: A purity loss function is constructed based on the gray-level distribution of the pure sub-region, wherein the purity loss function is used to characterize the degree of difference between the gray-level distribution of the pure sub-region and the gray-level regression distribution obtained by fitting it; A hybrid loss function is constructed based on the grayscale distribution of the hybrid sub-region, wherein the hybrid loss function is used to characterize the degree of difference between the grayscale distribution of the hybrid sub-region and the grayscale regression distribution fitted by the pure sub-region; An iterative loss function is constructed based on the pure loss function and the mixed loss function, and the gray-level distribution of the pure sub-region is iteratively fitted based on the iterative loss function to obtain the second fitting information.

9. The fluorescence image signal enhancement method based on three-dimensional DNA nanostructures according to claim 8, characterized in that, The step of constructing an iterative loss function based on the pure loss function and the mixed loss function includes: Calculate the sum of the area of ​​the pure sub-region and the area of ​​the mixed sub-region to obtain the sum of the areas; Calculate the ratio of the area of ​​the pure sub-region to the sum of the areas to obtain the purity coefficient; Calculate the ratio of the area of ​​the mixed sub-region to the sum of the areas to obtain the mixing coefficient; The iterative loss function is constructed based on the pure loss function, the mixed loss function, the pure coefficient, and the mixed coefficient.

10. A fluorescence image signal enhancement system based on three-dimensional DNA nanostructures, characterized in that, The system includes: An image recognition module is used to perform image recognition on fluorescence images to determine multiple diffraction spot regions; The region subdivision module is used to distinguish between a first region corresponding to a single-molecule diffraction spot and a second region corresponding to an overlapping diffraction spot in the plurality of diffraction spot regions, wherein the degree of matching between the region shape of the first region and the standard shape of the diffraction spot is greater than a matching threshold, and the degree of matching between the region shape of the second region and the standard shape is less than or equal to the matching threshold. The first fitting module is used to fit the gray-level distribution of the first region to obtain first fitting information, wherein the first fitting information is used to determine the center position of the first region. The second fitting module is used to distinguish between mixed sub-regions and pure sub-regions in the second region, and to fit the gray distribution of the pure sub-regions to the gray distribution of the mixed sub-regions as a constraint condition to obtain second fitting information, wherein the second fitting information is used to determine the center position of the second region. The molecular localization module is used to perform a molecular localization task of the fluorescence image based on the first fitting information and the second fitting information, and obtain the molecular localization information of the fluorescence image.

Images