CN-122024232-A - Fluorescent image signal enhancement system and method based on three-dimensional DNA nano structure

CN122024232ACN 122024232 ACN122024232 ACN 122024232ACN-122024232-A

Abstract

The invention relates to the technical field of image processing, in particular to a fluorescent image signal enhancement system and method based on a three-dimensional DNA nano structure. The method comprises the steps of carrying out image recognition on a fluorescent image to determine a plurality of diffraction spot areas, distinguishing a first area corresponding to single-molecule diffraction spots from a second area corresponding to overlapping diffraction spots in the plurality of diffraction spot areas, carrying out fitting on gray distribution of the first area to obtain first fitting information, distinguishing a mixed subarea and a pure subarea in the second area, carrying out fitting on gray distribution of the pure subarea by taking gray distribution of the mixed subarea as a constraint condition to obtain second fitting information, and carrying out molecular positioning tasks of the fluorescent image according to the first fitting information and the second fitting information to obtain molecular positioning information of the fluorescent image. The invention can improve the positioning precision of the fluorescent image when facing the overlapping condition of diffraction spots.

Inventors

CAI QILIANG
LI GANG
KONG DEMING
WANG CHENYU
ZHAO YAN
ZHU LINA
KANG SHAOSAN
LIU ZHIFEI
WANG ZHUN
ZHANG BAOSHUAI

Assignees

天津医科大学第二医院

Dates

Publication Date: 20260512
Application Date: 20260203

Claims (10)

1. A method for enhancing fluorescent image signals based on three-dimensional DNA nanostructures, the method comprising: performing image recognition on the fluorescent image to determine a plurality of diffraction spot areas; Distinguishing a first area corresponding to a single-molecule diffraction spot from a second area corresponding to an overlapping diffraction spot in the plurality of diffraction spot areas, wherein the degree of matching between the area shape of the first area and the standard shape of the diffraction spot is greater than a matching threshold value, and the degree of matching between the area shape of the second area and the standard shape is less than or equal to the matching threshold value; Fitting the gray distribution of the first region to obtain first fitting information, wherein the first fitting information is used for determining the central position of the first region; Distinguishing a mixed subarea from a pure subarea in the second area, and fitting the gray distribution of the pure subarea by taking the gray distribution of the mixed subarea as a constraint condition to obtain second fitting information, wherein the second fitting information is used for determining the center position of the second area; and executing a molecular positioning task of the fluorescent image according to the first fitting information and the second fitting information to obtain molecular positioning information of the fluorescent image.
2. The method of claim 1, wherein the image recognition of the fluorescent image to determine the plurality of diffraction spot areas comprises: Identifying target pixel points in all pixel points included in the fluorescent image according to a gray segmentation threshold corresponding to the fluorescent image, wherein the gray segmentation threshold indicates the maximum gray of the image background of the fluorescent image; Constructing a plurality of candidate areas based on a plurality of target pixel points; And determining the candidate areas with the number of the included pixel points larger than a number threshold value as the diffraction spot areas in the plurality of candidate areas so as to obtain the plurality of diffraction spot areas.
3. The method for enhancing a fluorescence image signal based on a three-dimensional DNA nanostructure according to claim 2, wherein the step of obtaining the gray segmentation threshold corresponding to the fluorescence image comprises: Analyzing a gray level histogram of a fluorescent image to determine a plurality of background gray levels, wherein the occurrence frequency corresponding to the background gray levels is smaller than the occurrence frequency corresponding to any gray level except the plurality of background gray levels in the gray level histogram; Determining a plurality of background areas based on all pixel points corresponding to the plurality of background gray levels in the fluorescent image; Respectively calculating the average value and standard deviation of gray values of all pixel points included in the background areas to obtain a background gray average value and a background gray standard deviation; And determining a gray scale segmentation threshold corresponding to the fluorescent image according to the background gray scale average value, the background gray scale standard deviation and a preset amplification factor, wherein the amplification factor is used for amplifying the numerical contribution of the background gray scale standard deviation in the gray scale segmentation threshold.
4. The method of claim 1, wherein the roundness-like shape of the area of the diffraction spot is used to indicate the degree of matching with the standard shape of the diffraction spot.
5. The method of claim 1, wherein the distinguishing the mixed subregion from the pure subregion in the second region comprises: image segmentation is carried out on the second region so as to obtain a plurality of segmentation subareas; Determining a plurality of region dividing lines in each dividing sub-region, wherein the region dividing lines pass through the geometric center of the corresponding dividing sub-region, and the included angle between the adjacent region dividing lines is a preset angle; Analyzing the structural similarity degree between two grandchild areas, which are cut by each area dividing line, of each dividing sub-area to obtain a plurality of equipartition indexes corresponding to each dividing sub-area, wherein the equipartition indexes corresponding to each dividing sub-area are in one-to-one correspondence with the area dividing lines corresponding to each dividing sub-area; among the plurality of divided subregions, a divided subregion corresponding to the highest average division index is determined as the mixed subregion, and other divided subregions except for the mixed subregion are determined as the pure subregion.
6. The method for enhancing a fluorescence image signal based on a three-dimensional DNA nanostructure according to claim 1, wherein fitting the gray scale distribution of the pure subregion using the gray scale distribution of the mixed subregion as a constraint condition to obtain second fitting information comprises: fitting the gray distribution of the pure subregion to obtain undetermined fitting information; Determining regression distribution of gray values of the mixed subregion according to the undetermined fitting information; analyzing the difference between the gray distribution of the mixed subregion and the regression distribution to obtain a undetermined fitting deviation value; determining the pending fit information as the second fit information if the pending fit deviation value is less than a deviation threshold; and under the condition that the deviation value of the to-be-determined fitting is larger than or equal to the deviation threshold value, taking the gray distribution of the mixed subarea as a constraint condition, and carrying out iterative fitting on the gray distribution of the pure subarea so as to obtain the second fitting information.
7. The method of claim 6, wherein analyzing the difference between the gray scale distribution of the mixed subregion and the regression distribution to obtain the to-be-fitted deviation value comprises: Calculating absolute differences between actual gray values of each pixel point in the mixed subarea and predicted gray values of the pixel points in the regression distribution in all pixel points included in the mixed subarea to obtain a plurality of gray deviation values; And calculating the average value of the gray level deviation values to obtain the undetermined fitting deviation value.
8. The method for enhancing a fluorescence image signal based on a three-dimensional DNA nanostructure according to claim 6, wherein iteratively fitting the gray scale distribution of the pure subregion using the gray scale distribution of the mixed subregion as a constraint condition to obtain the second fitting information comprises: Constructing a pure loss function based on the gray distribution of the pure subarea, wherein the pure loss function is used for representing the degree of difference between the gray distribution of the pure subarea and gray regression distribution obtained by fitting the gray distribution of the pure subarea; Constructing a mixing loss function based on the gray distribution of the mixing subregion, wherein the mixing loss function is used for representing the degree of difference between the gray distribution of the mixing subregion and the gray regression distribution obtained by fitting the pure subregion; And constructing an iterative loss function according to the pure loss function and the mixed loss function, and performing iterative fitting on the gray distribution of the pure subarea based on the iterative loss function to obtain the second fitting information.
9. The method of claim 8, wherein constructing an iterative loss function from the pure loss function and the mixed loss function comprises: calculating the sum of the area of the pure subarea and the area of the mixed subarea to obtain an area sum value; Calculating the ratio of the area of the pure subarea to the sum of the areas to obtain a pure coefficient; calculating the ratio of the area of the mixing subarea to the area sum value to obtain a mixing coefficient; and constructing the iterative loss function according to the pure loss function, the mixed loss function, the pure coefficient and the mixed coefficient.
10. A fluorescent image signal enhancement system based on three-dimensional DNA nanostructures, the system comprising: The image recognition module is used for carrying out image recognition on the fluorescent image so as to determine a plurality of diffraction spot areas; the area subdivision module is used for distinguishing a first area corresponding to a single-molecule diffraction spot from a second area corresponding to an overlapped diffraction spot in the plurality of diffraction spot areas, wherein the matching degree between the area shape of the first area and the standard shape of the diffraction spot is larger than a matching threshold value, and the matching degree between the area shape of the second area and the standard shape is smaller than or equal to the matching threshold value; The first fitting module is used for fitting the gray distribution of the first area to obtain first fitting information, wherein the first fitting information is used for determining the central position of the first area; the second fitting module is used for distinguishing a mixed subarea from a pure subarea in the second area, taking the gray distribution of the mixed subarea as a constraint condition, and fitting the gray distribution of the pure subarea to obtain second fitting information, wherein the second fitting information is used for determining the center position of the second area; and the molecular positioning module is used for executing the molecular positioning task of the fluorescent image according to the first fitting information and the second fitting information to obtain the molecular positioning information of the fluorescent image.

Description

Fluorescent image signal enhancement system and method based on three-dimensional DNA nano structure Technical Field The invention relates to the technical field of image processing, in particular to a fluorescent image signal enhancement system and method based on a three-dimensional DNA nano structure. Background The fluorescence imaging technology of three-dimensional DNA nano-structure is an important development direction in the field of biological imaging in recent years, and the DNA nano-technology can construct a highly programmable three-dimensional structure by utilizing the self-assembly characteristic of DNA molecules and realize accurate molecular control on nano scale. 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 increasing, so that the optimization of the three-dimensional structure construction of fluorescence images by combining with the artificial intelligence technology to enhance the fluorescence image signals has gradually become an important development problem. When molecular positioning is carried out in fluorescence imaging, two-dimensional Gaussian fitting is generally carried out by utilizing diffraction spots, molecular positioning is obtained based on a final fitting result, but in practice, the situation that diffraction spots overlap possibly exists, gray distribution in the overlapped diffraction spots damages Gaussian distribution characteristics of single diffraction spots, at the moment, the accuracy of positioning the overlapped diffraction spots is poor, one diffraction spot is lost in positioning, and the accuracy of a final three-dimensional image is not satisfied. Disclosure of Invention The invention aims to provide a fluorescent image signal enhancement system and method based on a three-dimensional DNA nano structure, which are used for solving the technical problem of low positioning precision of a fluorescent image three-dimensional structure in the prior art. In a first aspect, an embodiment of the present invention provides a fluorescence image signal enhancement method based on three-dimensional DNA nanostructures, the method comprising: performing image recognition on the fluorescent image to determine a plurality of diffraction spot areas; Distinguishing a first area corresponding to a single-molecule diffraction spot from a second area corresponding to an overlapping diffraction spot in the plurality of diffraction spot areas, wherein the degree of matching between the area shape of the first area and the standard shape of the diffraction spot is greater than a matching threshold value, and the degree of matching between the area shape of the second area and the standard shape is less than or equal to the matching threshold value; Fitting the gray distribution of the first region to obtain first fitting information, wherein the first fitting information is used for determining the central position of the first region; Distinguishing a mixed subarea from a pure subarea in the second area, and fitting the gray distribution of the pure subarea by taking the gray distribution of the mixed subarea as a constraint condition to obtain second fitting information, wherein the second fitting information is used for determining the center position of the second area; and executing a molecular positioning task of the fluorescent image according to the first fitting information and the second fitting information to obtain molecular positioning information of the fluorescent image. In one embodiment, the image recognition of the fluorescence image to determine a plurality of diffraction spot areas includes: Identifying target pixel points in all pixel points included in the fluorescent image according to a gray segmentation threshold corresponding to the fluorescent image, wherein the gray segmentation threshold indicates the maximum gray of the image background of the fluorescent image; Constructing a plurality of candidate areas based on a plurality of target pixel points; And determining the candidate areas with the number of the included pixel points larger than a number threshold value as the diffraction spot areas in the plurality of candidate areas so as to obtain the plurality of diffraction spot areas. In one embodiment, the step of acquiring the gray segmentation threshold corresponding to the fluorescence image includes: Analyzing a gray level histogram of a fluorescent image to determine a plurality of background gray levels, wherein the occurrence frequency corresponding to the background gray levels is smaller than the occurrence frequency corresponding to any gray level except the plurality of background gray levels in the gray level histogram; Determining a plurality of background areas based on all pixel points corresponding to the plurality of background gray levels in the fluorescent image; Respectively cal