An immunoblot membrane strip interpretation method and system based on image processing

Abstract

This invention discloses an image processing-based method for interpreting immunoblot membrane strips. The steps are as follows: S1, acquiring a membrane strip image, which includes the areas where quality control points and detection points are located on the membrane strip, and performing image correction processing directly or after image correction before proceeding to step S2; S2, preprocessing the membrane strip image to obtain a grayscale image; S3, identifying the position coordinates and radius of the quality control points on the membrane strip based on the grayscale image, and performing horizontal correction to obtain a corrected grayscale image; S4, identifying the position coordinates of the detection points on the membrane strip based on the corrected grayscale image and the position coordinates and radius of the quality control points obtained in step S3; S5, calculating the relative grayscale value of the detection points and comparing it with the detection results of standards of different concentrations, converting it into a concentration value for interpretation. The beneficial effects of this invention are: it can accurately judge dot-like membrane strips, and the test results are highly reliable; combined with dot-like membrane strip detection, it provides a high-throughput, rapid interpretation method and system.

Description

Technical Field

[0001] This invention belongs to the field of biomedical detection technology, and relates to biomedical image processing, specifically to an immunoblot strip interpretation method and system based on image processing. Background Technology

[0002] Western blotting is a hybridization technique that combines high-resolution gel electrophoresis and immunochemical analysis. It offers advantages such as large analytical capacity, high sensitivity, and strong specificity, making it one of the most commonly used methods for detecting protein characteristics, expression, and distribution. The Western blotting detection process includes sample loading, (incubation) reaction, washing, drying, and interpretation. Currently, most interpretation methods involve scanning the membrane strip with a general-purpose scanner to generate images, followed by outputting results using proprietary interpretation software. However, general-purpose scanners are consumer products, characterized by rapid technological advancements and inconsistent grayscale levels across different areas, making them unsuitable for biomedical applications and difficult to integrate with automated Western blotting interpretation equipment. Interpretation methods based on camera image acquisition offer superior production and interpretation stability compared to scanner-based methods and can be integrated into fully automated Western blotting interpreters.

[0003] Immunoblot strips can be classified into strip-shaped and dot-shaped detection strips based on the shape of the detection area. For example... Figure 1 For membrane strips of the same size, dot-shaped detection membrane strips have the advantages of high detection point density and a wider range of detection items. However, when actually using a camera to acquire images, due to the membrane strip manufacturing process and camera lens distortion, the detection points captured in the image may deviate from the standard position, or the size of the membrane strip images acquired by different acquisition devices may differ due to different focal lengths. These factors bring certain difficulties to image processing and judgment. Patent document CN108562750A discloses a method for interpreting images of imprinted membrane strips acquired by photography, but this method is applicable to the interpretation of membrane strip images with bar detection lines, but cannot interpret dot-shaped membrane strips. Patent document CN113920423A discloses an immunoblot image recognition method, which also cannot interpret dot-shaped membrane strips. Summary of the Invention

[0004] In view of this, the present invention provides an image processing-based method for interpreting immunoblot strips. Through reasonable image processing and analysis techniques, it is possible to more accurately interpret membrane strip images of small-sized blot regions, especially dot-shaped membrane strip images.

[0005] The technical solution is as follows:

[0006] A method for interpreting immunoblot strips based on image processing, the key steps of which are as follows:

[0007] S1. Obtain the membrane strip image, which includes the areas where the quality control points and detection points are located on the membrane. Proceed directly to step S2 or perform image correction processing as needed before proceeding to step S2.

[0008] S2, preprocess the membrane strip image to obtain a grayscale image;

[0009] S3, based on the grayscale image, identify the position coordinates and radius of the quality control points on the membrane strip, and perform horizontal correction to obtain a corrected grayscale image. In the corrected grayscale image, the quality control points in the same row are arranged along the horizontal direction.

[0010] S4. Based on the corrected grayscale image and the position coordinates and radius of the quality control point obtained in step S3, identify the position coordinates of the detection point on the membrane strip.

[0011] S5. Calculate the relative gray value of the detection point and compare it with the detection results of standards of different concentrations, and convert it into a concentration value for interpretation.

[0012] In one implementation, step S3 includes:

[0013] S31, Based on the characteristics of the grayscale image, select parameters to perform a binarization operation to obtain a binarized image, in preparation for identifying quality control points of the membrane strip image;

[0014] A Cartesian coordinate system is established with the two adjacent sides of the membrane strip image as the x-axis and y-axis, respectively.

[0015] S32, Perform feature recognition on the binarized image to obtain a candidate contour set including quality control points;

[0016] S33, according to the set rules for the quality control point features of the membrane strip, the quality control point contours are selected from the candidate contour set, thereby calculating the coordinates and radius of the quality control points.

[0017] In one embodiment, in step S32 above, the Suzuki contour recognition algorithm or the Hough circle detection algorithm is used to perform contour search operation on the binarized image.

[0018] In step S33, the lower limit value of the contour area is used as the feature of the membrane strip quality control point. Imprint point contours with contour areas above the lower limit value of the contour area are selected to obtain the candidate contour set.

[0019] The method for calculating the coordinates and radius of the quality control point is as follows: calculate the largest inscribed circle for each contour in the candidate contour set, obtain the center coordinates and radius of the inscribed circle, and confirm whether it is a quality control point contour based on the center coordinates and radius.

[0020] In one embodiment, step S4 includes:

[0021] S41, Based on the positional relationship between the detection points and quality control points in the membrane strip design and the coordinates of the quality control points, the ideal coordinates of the detection points are initially calculated;

[0022] S42, search again for the actual coordinates of the detection point near the ideal coordinates of the detection point;

[0023] S43, If a detection point can be found in step S42, the actual coordinates in step S42 are used as the position coordinates of the detection point; otherwise, the ideal coordinates in step S41 are used as the position coordinates of the detection point.

[0024] In one implementation, step S42 uses a pattern matching algorithm or a contour recognition algorithm for searching.

[0025] In one embodiment, the imprints on the membrane strip are arrays of 2 rows and multiple columns, wherein the first and last columns of the arrays are quality control points;

[0026] In step S3, after identifying the quality control points and obtaining their coordinates, the grayscale image in step S2 is horizontally corrected so that the y-values ​​of the quality control points in the same row are the same, thus obtaining the corrected grayscale image.

[0027] In the corrected grayscale image, the coordinates of the two quality control points in the first row, first column and last column are (xl1, yl1) and (xr1, yr1) respectively, and the coordinates of the two quality control points in the second column, first column and last column are (xl2, yl2) and (xr2, yr2) respectively. The radius of the quality control points is rz; where yl1 = yr1.

[0028] The specific process of step S41 is as follows:

[0029] Based on the coordinates, radius, and number of points per row of the membrane strip obtained after horizontal correction in step S3, calculate the ideal coordinates of each detection point. The coordinates of the detection point in the 1st row and nth column are:

[0030] (x 1n y 1n )=(xl1+n*spacedist,yl1),

[0031] The coordinates of the detection point in the 2nd row and nth column are: (x 2n y 2n )=(xl2+n*spacedist,yl2),

[0032] in N is the number of detection points per row.

[0033] In one implementation, step S42 specifically involves performing a cross-correlation operation within a rectangular area with a side length of 2*search_dist, centered on the ideal coordinates of each detection point. The pixel coordinates corresponding to the maximum grayscale value of the pixel in the cross-correlation operation result are then used as candidate coordinates for the detection point.

[0034]

[0035] The method for determining the kernel of the correlation matrix is ​​as follows:

[0036]

[0037] Where kernel_size = rz * 2;

[0038] The specific process of step S43 is as follows: divide the maximum value of the relevant operation result in step S42 by the area of ​​the kernel of the relevant matrix to obtain the mean value of the relevant operation result, and then compare the mean value with a set threshold. If it is greater than the threshold, the candidate coordinates are taken as the actual coordinates and the actual coordinates are confirmed as the position coordinates of the detection point. If it is less than the threshold, the candidate coordinates are discarded and the ideal coordinates are taken as the position coordinates of the detection point.

[0039] In one implementation, step S5 includes:

[0040] S51, calculate the absolute grayscale mean of the detection point, point_mean_abs, using the following formula:

[0041]

[0042] S52, calculate the mean grayscale value of the detection point, pointback_mean_abs, using the following formula:

[0043]

[0044] S53, calculate the relative gray value of the detection point, point_mean_rela, using the following formula:

[0045] poin_mean_rela=point_mean_abs-pointback_mean_abs;

[0046] S54. Calculate the concentration of the detection point based on the pre-established gray-concentration curve. The gray-concentration curve is established by using different concentration standards and testing them using the same method to obtain the gray-concentration detection results of the standards. Based on the gray-concentration detection results of the standards and the concentration data, the gray-concentration curve is obtained by curve fitting.

[0047] In one implementation, in step S1 above, if the image comes from a scanner, it is used directly; if it is acquired from a camera, the following steps are taken for image correction:

[0048] S11, coarsely adjust the camera position or membrane strip position so that the membrane strip is roughly in the center of the image;

[0049] S12, identify the specific position of the membrane strip, and finely adjust the camera a second time based on the difference between the specific position of the membrane strip and the position of the image center so that the membrane strip is located in the center of the image, and acquire the image and correct the distortion.

[0050] In one embodiment, step S2 includes:

[0051] S21, Identify the outline of the membrane strip and cut it;

[0052] S22, perform noise filtering on the cropped membrane strip image;

[0053] S23, the filtered image is converted to grayscale to obtain the grayscale image.

[0054] In one implementation, the filtering method in step S22 may be mean filtering or Gaussian filtering.

[0055] The second objective of this invention is to provide a judgment system. Its technical solution is as follows:

[0056] A judgment system, the key of which is that it includes an image acquisition device and a computer, the computer including a memory and a calculator;

[0057] The image signal output terminal of the image acquisition device is connected to the image signal input terminal of the computer, and the input image information is stored in the memory.

[0058] The memory also stores an image interpretation program that can run on the calculator. When the image interpretation program runs, it performs the image processing and interpretation process of steps S2 to S5 as described above, and outputs the interpretation result.

[0059] Compared with existing technologies, the beneficial effects of this invention are: it can accurately judge dot-like membrane strips, and the test results are highly reliable; for image data acquired by a camera, the combination of distortion correction and algorithms reduces the difficulty of image processing and improves accuracy; combined with dot-like membrane strip detection, it provides a high-throughput, fast interpretation method and system. Attached Figure Description

[0060] Figure 1 The diagram shows three types of dot-shaped membrane strips, where (a), (b), and (c) have one, two, and multiple rows of imprinted dots, respectively.

[0061] Figure 2 This is a schematic diagram of the interpretation method of the present invention;

[0062] Figure 3 Intermediate image during the interpretation process of the multi-allergen-specific antibody IgE detection membrane strip. Detailed Implementation

[0063] The present invention will be further described below with reference to the embodiments and accompanying drawings.

[0064] Example 1

[0065] like Figure 2 An image processing-based method for interpreting immunoblot strips is provided for interpreting strips with circular immunoblot dots, including control points and detection points. The steps are as follows:

[0066] S1. Obtain the membrane strip image, which includes the areas where the quality control points and detection points are located on the membrane. Proceed directly to step S2 or perform image correction processing as needed before proceeding to step S2.

[0067] S2, preprocess the membrane strip image to obtain a grayscale image;

[0068] S3, based on the grayscale image, identify the position coordinates and radius of the quality control point on the membrane strip, and perform horizontal correction to obtain the corrected grayscale image;

[0069] S4. Based on the corrected grayscale image and the position coordinates and radius of the quality control point obtained in step S3, identify the position coordinates of the detection point on the membrane strip.

[0070] S5. Calculate the relative gray value of the detection point and compare it with the detection results of standards of different concentrations, and convert it into a concentration value for interpretation.

[0071] Specifically:

[0072] In step S1, if the image comes from a scanner, it is used directly; if it comes from a camera, the following steps are taken for image correction:

[0073] S11, coarsely adjust the camera position or membrane strip position so that the membrane strip is roughly in the center of the image;

[0074] S12, identify the specific position of the membrane strip, and finely adjust the camera a second time based on the difference between the specific position of the membrane strip and the position of the image center so that the membrane strip is located in the center of the image, and acquire the image and correct the distortion.

[0075] Step S2 includes:

[0076] S21, Identify the outline of the membrane strip and cut it;

[0077] S22, perform noise filtering on the cropped membrane strip image. The filtering method can be mean filtering, Gaussian filtering, etc.

[0078] S23, the filtered image is converted to grayscale to obtain the grayscale image.

[0079] Step S3 includes:

[0080] S31, Based on the characteristics of the grayscale image, select appropriate parameters to perform binarization operation on it to obtain a binarized image, in preparation for identifying quality control points of the membrane strip image;

[0081] A Cartesian coordinate system is established with the two adjacent sides of the membrane strip image as the x-axis and y-axis, respectively.

[0082] S32, perform feature recognition on the binarized image, and use the Suzuki contour recognition algorithm or the Hough circle detection algorithm to perform contour search operation on the binarized image to obtain a candidate contour set including quality control points; the features of the membrane strip quality control points include size and position.

[0083] S33, according to the membrane strip quality control point feature setting rules, quality control point contours are selected from the candidate contour set, thereby calculating the coordinates and radius of the quality control points. Specifically, using the set lower limit value of the contour area as the membrane strip quality control point feature, imprint point contours with contour areas above the lower limit value are selected to obtain the candidate contour set; the method for calculating the coordinates and radius of the quality control points is as follows: for each contour in the candidate contour set, the largest inscribed circle is calculated to obtain the center coordinates and radius of the inscribed circle, and the center coordinates and radius are used to confirm whether it is a quality control point contour.

[0084] Step S4 includes:

[0085] S41, Based on the positional relationship between the detection points and quality control points in the membrane strip design and the coordinates of the quality control points, the ideal coordinates of the detection points are initially calculated;

[0086] S42, search again for the actual coordinates of the detection point near the ideal coordinates of the detection point; the search algorithm can be a pattern matching algorithm, or the Suzuki contour recognition algorithm as in step S32;

[0087] S43, If a detection point can be found in step S42, the actual coordinates in step S42 are used as the position coordinates of the detection point; otherwise, the ideal coordinates in step S41 are used as the position coordinates of the detection point.

[0088] Step S5 includes:

[0089] S51, calculate the absolute grayscale mean of the detection point, point_mean_abs, using the following formula:

[0090]

[0091] S52, calculate the mean grayscale value of the detection point, pointback_mean_abs, using the following formula:

[0092]

[0093] S53, calculate the relative gray value of the detection point, point_mean_rela, using the following formula:

[0094] poin_mean_rela=point_mean_abs-pointback_mean_abs;

[0095] S54. Calculate the concentration of the detection point based on the pre-established gray-concentration curve. The gray-concentration curve is established by using different concentration standards and testing them using the same method to obtain the gray-concentration detection results of the standards. Based on the gray-concentration detection results of the standards and the concentration data, the gray-concentration curve is obtained by curve fitting.

[0096] The method of the present invention is specifically illustrated using a multi-allergen-specific antibody IgE detection assay.

[0097] The immunoblot strips were from the Allergen-Specific Antibody IgE Detection Kit (Enzyme Immunoblotting) from Hangzhou Zhejiang University Disun Biotechnology Co., Ltd., kit number I-sIgE-HH-2801. After removing the quality control points, the strips have 2 rows and 14 columns, totaling 28 detection items. The test samples were from clinical serum. Following the kit instructions, after sample addition, incubation, washing, and air drying, the following steps were performed:

[0098] 1) Use a self-made camera-based image acquisition device to acquire images of the membrane strip, obtaining image A of the membrane strip being tested, such as... Figure 3 The image acquisition device is equipped with a position adjustment mechanism, which can adjust the camera position bidirectionally in the horizontal plane.

[0099] a) The coarse adjustment mechanism is used to adjust the camera position so that the membrane strip reaches the center of the image, the image is acquired, and distortion correction is performed; an image coordinate system is established with the upper left corner of the image as the zero point, the horizontal axis as the x-axis, and the vertical axis as the y-axis. The distortion correction formula is:

[0100]

[0101]

[0102] in, (x0, y0) is the distortion center. Based on the known lens parameter λ, substituting (x0, y0) into the distortion correction formula yields image A0.

[0103] b) Identify the specific position of the membrane strip, fine-tune the position adjustment mechanism to center the membrane strip in the image, acquire image A1 and perform distortion correction, the distortion correction is the same as in step a; specifically:

[0104] i. The Sobel operator is used to perform edge detection on image A0 in the X and Y directions to obtain image A2. In this embodiment, the Sobel operator has a 3×3 matrix kernel.

[0105] ii. Perform image grayscale inversion, erosion, and binarization operations on image A2 to obtain image A3. In this embodiment, the erosion kernel size is 3×3, and the operation is repeated 3 times.

[0106] iii. The Suzuki algorithm is used to perform contour search on image A3. In this embodiment, all contours in image A3 are extracted to obtain a contour set. Then, according to the set rules, the contour rectangle of the membrane strip is selected from the contour set to obtain the center coordinates of the contour rectangle.

[0107] iv. Based on the difference between the center coordinates of the outline rectangle and the center coordinates of the image, fine-tune the position adjustment mechanism to bring the membrane strip to the center position of the image, acquire the image, and correct the distortion to obtain image A.

[0108] 2) Identify the membrane strip border in image A, crop it, and perform filtering and grayscale preprocessing on the cropped image to obtain image B;

[0109] a) Identify the membrane strip border in image A and crop it to obtain image B1, wherein the method for identifying the membrane strip border is the same as in step 1).

[0110] b) Apply Gaussian filtering to image B1 to obtain image B2, where the Gaussian convolution kernel formula is:

[0111]

[0112] In this embodiment, the convolution kernel size is set to 5 and σ = 1.5 based on the size of image B1.

[0113] c) Convert the filtered image B2 to grayscale image B, where the grayscale conversion formula is:

[0114] Gray=0.39*Red+0.5*Green+0.11*Blue.

[0115] 3) Identify the location and size of the quality control points in image B.

[0116] a) Adaptive binarization is performed on image B to obtain image C1. In this embodiment, the pixel comparison area size of adaptive binarization is 131. Within the comparison range, pixels with a gray value 15 smaller than the surrounding pixels are set to 255, and the rest are set to 0.

[0117] b) The Suzuki contour recognition algorithm is used to perform contour search on the binarized image C1. The obtained contour set is then selected and sorted according to a set of rules to obtain a candidate contour set for quality control points. In this embodiment, the contour set is sorted according to the x-axis coordinate, and the area of ​​the contour should be greater than 100.

[0118] c) Calculate the largest inscribed circle for each profile in the candidate profile set of quality control points, obtain the center coordinates and radius of the inscribed circle, and confirm whether it is a quality control point profile based on the center coordinates and radius. In this embodiment, the x-coordinates of the two quality control points on both sides of the membrane strip should be similar. Let the coordinates of the upper left quality control point be (cxl1, cyl1), the coordinates of the upper right quality control point be (cxr1, cyr1), the coordinates of the lower left quality control point be (cxl2, cyl2), and the coordinates of the lower right quality control point be (cxr2, cyr2), with a radius of rz.

[0119] d) Horizontal correction of membrane strip imprint points. The purpose of correction is to make the y-coordinates of the left and right quality control points on the same row of the membrane strip equal, so as to facilitate the calculation of the position coordinates of the detection points. Specifically, let the upper left quality control point be the center, the horizontal x-direction of the upper left quality control point be the baseline, and the angle between the line connecting the upper left and upper right quality control points and the baseline be θ.

[0120] but

[0121] Then, centering on the top left quality control point, rotate counterclockwise by θ to horizontally correct the membrane strip image B, obtaining image C. The rotation formula is:

[0122] x′=cxl1+(x-cxl1)cosθ+(y-cyl1)sinθ

[0123] y′=cyl1+(x-cxl1)sinθ+(y-cyl1)cosθ

[0124] Where (x, y) are points in image B, and (x′, y′) are points in the corrected image C. Similarly, substituting the coordinates of the quality control points in image B into the correction formula yields the coordinates of the top-left quality control point (xl1, yl1), the top-right quality control point (xr1, yr1), the bottom-left quality control point (xl2, yl2), and the bottom-right quality control point (xr2, yr2) in the corrected image C.

[0125] 4) As shown in Figure C2, calculate and identify the position of the detection point in the corrected image C.

[0126] a) Calculate the position coordinates of each detection point based on the quality control point coordinates, radius, and number of points per row of the membrane strip obtained in step 3). The coordinates of the detection point in the 1st row and nth column are:

[0127] (x 1n y 1n )=(xl1+n*spacedist,yl1),

[0128] The coordinates of the detection point in the 2nd row and nth column are:

[0129] (x 2n y 2n )=(xl2+n*spacedist,yl2),

[0130] in N is the number of detection points per row; in this embodiment, N = 14.

[0131] b) Using the coordinates calculated for each detection point in the above steps as the center, perform cross-correlation within a rectangle with a side length of 2*search_dist. The maximum pixel grayscale value of the cross-correlation result is used as the candidate coordinates for the detection point.

[0132]

[0133]

[0134] Where kernel_size = rz * 2.

[0135] c) Divide the maximum value of the relevant calculation results in the above steps by the area of ​​the kernel of the relevant matrix to obtain the mean value of the relevant calculation results. Then compare the mean value with the set threshold. If it is greater than the threshold, take the candidate coordinate as the actual coordinate and confirm the actual coordinate as the detection point coordinate. If it is less than the threshold, discard the candidate coordinate and select the calculated coordinate in step a) as the detection point coordinate.

[0136] 5) Calculate the relative gray value of each detection point based on the location coordinates of the detection points obtained in step 4), and obtain the judgment result based on the pre-established gray-concentration curve.

[0137] a) Calculate the absolute grayscale mean of the detection points, point_mean_abs, using the following formula:

[0138]

[0139] b) Calculate the mean grayscale value of the detection point, pointback_mean_abs, using the following formula:

[0140]

[0141] c) Calculate the relative gray value of the detection point, point_mean_rela, using the following formula:

[0142] poin_mean_rela=point_mean_abs-pointback_mean_abs;

[0143] d) Calculate the concentration at the detection point based on the pre-established gray-concentration curve. A series of standards with different concentrations are tested. This method is used to obtain the relative gray values ​​of the standard detection points, thus obtaining the relative gray values ​​and concentration data pairs of the standards. Then, a gray-concentration curve is generated through curve fitting. A cubic polynomial curve or a Logistic curve can be used for fitting. In this embodiment, the gray-concentration curve formula established based on pre-calibrated experiments is an inverse function of the four parameters of the Logistic curve.

[0144]

[0145] Where X is the concentration, Y is the relative gray value point_mean_rela, and the fitting parameters are A1 = -0.76163, A2 = 16378.6, X0 = 2727.2, and p = 0.641.

[0146] Substituting the relative gray values ​​of each detection point into the above formula, the concentration results of the detection items are finally obtained, as shown in Table 1. Each detection item in Table 1 corresponds one-to-one with the detection point on the membrane strip.

[0147] Table 1. Concentration interpretation results at each detection point (unit: IU / mL)

[0148]

[0149] Example 2

[0150] An interpretation system includes an image acquisition device and a computer, the computer including a memory and a calculator;

[0151] The image signal output terminal of the image acquisition device is connected to the image signal input terminal of the computer, and the input image information is stored in the memory.

[0152] The memory also stores an image interpretation program that can run on the calculator. When the image interpretation program runs, it performs the image processing and interpretation process as described in steps S2 to S5 of Embodiment 1 and outputs the interpretation result.

[0153] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention. Those skilled in the art, under the guidance of the present invention, can make various similar representations without departing from the spirit and claims of the present invention, and such modifications all fall within the protection scope of the present invention.

Claims

1. A method for interpreting immunoblot strips based on image processing, characterized in that... The steps are as follows: S1. Obtain the membrane strip image, which includes the areas where the quality control points and detection points are located on the membrane. Proceed directly to step S2 or perform image correction processing as needed before proceeding to step S2. S2, preprocess the membrane strip image to obtain a grayscale image; S3, based on the grayscale image, identify the position coordinates and radius of the quality control points on the membrane strip, and perform horizontal correction to obtain a corrected grayscale image. In the corrected grayscale image, the quality control points in the same row are arranged along the horizontal direction. S4. Based on the corrected grayscale image and the position coordinates and radius of the quality control point obtained in step S3, identify the position coordinates of the detection point on the membrane strip. S5, calculate the relative gray value of the detection point and compare it with the detection results of standards of different concentrations, and convert it into a concentration value for interpretation; In step S3, the horizontal correction process is as follows: Let the top-left quality control point be the center, the horizontal x-direction of the top-left quality control point be the baseline, and the angle θ between the line connecting the top-left and top-right quality control points and the baseline be θ. then ; Then, centering on the top-left quality control point, rotate counterclockwise by θ to horizontally correct the grayscale image of the membrane strip, obtaining the corrected grayscale image. The rotation formula is: ； ； Where (x, y) are points in the grayscale image of the membrane strip, () represents a point in the corrected grayscale image.

2. The image processing-based immunoblot membrane strip interpretation method according to claim 1, characterized in that, Step S3 includes: S31, Based on the characteristics of the grayscale image, select parameters to perform a binarization operation to obtain a binarized image, in preparation for identifying quality control points of the membrane strip image; A Cartesian coordinate system is established with the two adjacent sides of the membrane strip image as the x-axis and y-axis, respectively. S32, Perform feature recognition on the binarized image to obtain a candidate contour set including quality control points; S33, according to the set rules for the quality control point features of the membrane strip, the quality control point contours are selected from the candidate contour set, thereby calculating the coordinates and radius of the quality control points.

3. The image processing-based immunoblot membrane strip interpretation method according to claim 2, characterized in that: In step S32, the Suzuki contour recognition algorithm or the Hough circle detection algorithm is used to perform contour search operation on the binarized image. In step S33, the lower limit value of the contour area is used as the feature of the membrane strip quality control point. Imprint point contours with contour areas above the lower limit value of the contour area are selected to obtain the candidate contour set. The method for calculating the coordinates and radius of the quality control point is as follows: calculate the largest inscribed circle for each contour in the candidate contour set, obtain the center coordinates and radius of the inscribed circle, and confirm whether it is a quality control point contour based on the center coordinates and radius.

4. The image processing-based immunoblot membrane strip interpretation method according to claim 3, characterized in that, Step S4 includes: S41, Based on the positional relationship between the detection points and quality control points in the membrane strip design and the position coordinates of the quality control points, the ideal coordinates of the detection points are initially calculated; S42, search again for the actual coordinates of the detection point near the ideal coordinates of the detection point; S43, If a detection point can be found in step S42, the actual coordinates in step S42 are used as the position coordinates of the detection point; otherwise, the ideal coordinates in step S41 are used as the position coordinates of the detection point.

5. The image processing-based immunoblot membrane strip interpretation method according to claim 4, characterized in that: The imprints on the membrane strip are arrays of 2 rows and multiple columns, wherein the first and last columns of the array are quality control points; In step S3, after identifying the quality control points and obtaining their coordinates, the grayscale image in step S2 is horizontally corrected so that the y-values ​​of the quality control points in the same row are the same, thus obtaining the corrected grayscale image. In the corrected grayscale image, the coordinates of the two quality control points in the first row and first column and the last column are (xl1, yl1) and (xr1, yr1) respectively, and the coordinates of the two quality control points in the second row and first column and the last column are (xl2, yl2) and (xr2, yr2) respectively. The radius of the quality control points is rz; where yl1 = yr1. The specific process of step S41 is as follows: Based on the coordinates, radius, and number of points per row of the membrane strip obtained after horizontal correction in step S3, calculate the ideal coordinates of each detection point. The coordinates of the detection point in the 1st row and nth column are: ， The coordinates of the detection point in the 2nd row and the n th column are: , wherein N is the number of detection points per row.

6. The method for interpreting immunoblot strips based on image processing according to claim 5, characterized in that: The specific process of step S42 is as follows: taking the ideal coordinates of each detection point as the center, a cross-correlation operation is performed within a rectangular area with a side length of 2*search_dist. The pixel coordinates corresponding to the maximum pixel grayscale value of the cross-correlation operation result are the candidate coordinates of the detection point. ； The method for determining the kernel of the correlation matrix is ​​as follows: ， in ; The specific process of step S43 is to divide the maximum value of the relevant calculation result in step S42 by... The mean of the relevant calculation results is obtained by calculating the kernel area. Then, the mean is compared with a set threshold. If it is greater than the threshold, the candidate coordinates are taken as the actual coordinates and the actual coordinates are confirmed as the position coordinates of the detection point. If it is less than the threshold, the candidate coordinates are discarded and the ideal coordinates are taken as the position coordinates of the detection point.

7. The image processing-based immunoblot membrane strip interpretation method according to claim 1, characterized in that, Step S5 includes: S51, calculate the absolute grayscale mean of the detection point, point_mean_abs, using the following formula: ； S52, calculate the mean grayscale value of the detection point, pointback_mean_abs, using the following formula: ； S53, calculate the relative gray value of the detection point, point_mean_rela, using the following formula: ； S54. Calculate the concentration of the detection point based on the pre-established gray-concentration curve. The gray-concentration curve is established by using different concentration standards and testing them using the same method to obtain the gray-concentration detection results of the standards. Based on the gray-concentration detection results of the standards and the concentration data, the gray-concentration curve is obtained by curve fitting.

8. The image processing-based immunoblot membrane strip interpretation method according to claim 1, characterized in that: In step S1, if the image comes from a scanner, it is used directly; if it comes from a camera, the following steps are taken for image correction: S11, coarsely adjust the camera position or membrane strip position so that the membrane strip is roughly in the center of the image; S12, identify the specific position of the membrane strip, and finely adjust the camera a second time based on the difference between the specific position of the membrane strip and the position of the image center so that the membrane strip is located in the center of the image, and acquire the image and correct the distortion.

9. The image processing-based immunoblot membrane strip interpretation method according to claim 1, characterized in that, Step S2 includes: S21, Identify the outline of the membrane strip and cut it; S22, perform noise filtering on the cropped membrane strip image; S23, the filtered image is converted to grayscale to obtain the grayscale image.

10. An interpretation system characterized by: Includes an image acquisition device and a computer, the computer including a memory and a calculator; The image signal output terminal of the image acquisition device is connected to the image signal input terminal of the computer, and the input image information is stored in the memory. The memory also stores an image interpretation program that can run on the calculator. When the image interpretation program runs, it performs the image processing and interpretation process of steps S2 to S5 as described in any one of claims 1 to 9, and outputs the interpretation result.

Images