Image processing method and device for lipoprotein subtyping

Abstract

This invention discloses an image processing method and apparatus for lipoprotein subtyping, belonging to the field of biodetection technology. The method includes: acquiring an original image; preprocessing the original image to obtain a target image; acquiring multiple single-tube images based on the target image; extracting effective regions from each of the multiple single-tube images to obtain multiple Regions of Interest (ROIs); analyzing each ROI to obtain multiple analysis results; and for each analysis result, determining the low-density lipoprotein (LDL) particle type based on the LDL average particle size. This application solves the technical problem in existing lipoprotein subtyping detection methods that cannot calculate the LDL average particle size and determine the LDL particle type when total cholesterol values ​​are missing, achieving the technical effect of accurately calculating the LDL average particle size and determining the LDL particle type even when total cholesterol values ​​are missing.

Description

Technical Field

[0001] The present invention relates to the field of biological detection technology, and in particular to an image processing method and device for lipoprotein subtyping. Background Art

[0002] Lipoproteins are spherical particles that carry cholesterol and triglycerides in plasma. According to their composition and density, they can be divided into very low-density lipoproteins, low-density lipoproteins and high-density lipoproteins. Accurate detection of lipoprotein subtypes is of great significance. The existing lipoprotein detection system still has some shortcomings during use: first, the existing technical solutions fail to perform necessary correction processing on the original image after image acquisition, and directly conduct subsequent analysis, which affects the accuracy of the analysis results; second, in the image preprocessing link, most of the existing solutions do not split the single test tube image from the original image, and do not locate the test tube area, which affects the detection efficiency and applicability; then, the existing solutions mainly rely on the total cholesterol value to calculate the cholesterol content of each lipoprotein subtype. However, in actual applications, total cholesterol information is not available at any time, which makes it impossible to carry out analysis when cholesterol data is missing; at the same time, in the process of drawing the peak type diagram of lipoprotein subtypes, the existing schemes do not normalize the horizontal axis, resulting in morphological differences in the spectrum obtained under different detection conditions, affecting the repeatability and comparability of the results. Summary of the invention

[0003] The present application provides an image processing method and device for lipoprotein subtyping, aiming to solve the technical problem in the existing lipoprotein subtyping detection method that the average particle size of low-density lipoprotein cannot be calculated and the type of low-density lipoprotein particles cannot be determined when the total cholesterol value is missing.

[0004] In view of the above problems, the present application provides an image processing method and device for lipoprotein subtyping.

[0005] The first aspect disclosed in the present application provides an image processing method for lipoprotein subtyping, the method comprising: obtaining an original image containing lipoprotein subtype bands; preprocessing the original image to obtain a target image, the preprocessing comprising grayscale correction, distortion correction, rotation correction, and test tube area centering processing; extracting a single test tube image according to the target image to obtain a plurality of single test tube images; extracting effective areas from the plurality of single test tube images respectively to obtain a plurality of ROI areas, wherein each ROI area corresponds to a very low density lipoprotein VLDL band to a high density lipoprotein HDL band in the test tube image; analyzing the plurality of ROI areas respectively to obtain a plurality of analysis results, wherein each analysis result comprises a lipoprotein subtyping peak pattern diagram and a low density lipoprotein average particle size value; for each analysis result, obtaining a low density lipoprotein particle type according to the low density lipoprotein average particle size value therein, wherein the low density lipoprotein particle types comprise type A, type B, and intermediate.

[0006] Another aspect disclosed in the present application provides an image processing device for lipoprotein subtyping, the device comprising: an original image acquisition unit, used to acquire an original image containing lipoprotein subtype bands; an image preprocessing unit, used to preprocess the original image to obtain a target image, wherein the preprocessing includes grayscale correction, distortion correction, rotation correction, and test tube area centering processing; a test tube image extraction unit, used to extract a single test tube image according to the target image to obtain multiple single test tube images; and an effective area extraction unit, used to extract effective areas from the multiple single test tube images respectively. , obtaining multiple ROI regions, wherein each ROI region corresponds to the very low density lipoprotein VLDL band to the high density lipoprotein HDL band in the test tube image; an ROI region analysis unit, used to analyze the multiple ROI regions respectively, to obtain multiple analysis results, wherein each analysis result includes a lipoprotein subtyping peak type diagram and an average particle size value of low density lipoprotein; a type acquisition unit, used to obtain the type of low density lipoprotein particles according to the average particle size value of low density lipoprotein for each analysis result, wherein the low density lipoprotein particle types include type A, type B and intermediate type.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] The invention adopts the method of obtaining an original image containing lipoprotein subtype bands; preprocessing the original image to obtain a target image, wherein the preprocessing process includes grayscale correction, distortion correction, rotation correction and test tube area centering processing. Compared with the prior art, the invention realizes the correction processing of the original captured image, which is helpful to improve the accuracy of subsequent analysis; extracting a single test tube image according to the target image, obtaining multiple single test tube images, and realizing automatic extraction of single test tube images; extracting effective areas from multiple single test tube images respectively, obtaining multiple ROI areas, and realizing automatic extraction of effective areas of pixel grayscale without manual intervention; analyzing multiple ROI areas respectively, obtaining multiple analysis results, and realizing analysis when the total cholesterol value is unknown; and obtaining the corresponding low-density lipoprotein particle type according to the low-density lipoprotein average particle size value. The technical solution solves the technical problem that the low-density lipoprotein average particle size value and the low-density lipoprotein particle type cannot be calculated in the absence of the total cholesterol value in the existing lipoprotein subtyping detection method, and achieves the technical effect of accurately calculating the low-density lipoprotein average particle size value and judging the low-density lipoprotein particle type in the absence of the total cholesterol value.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic diagram of a flow chart of an image processing method for lipoprotein subtyping provided in an embodiment of the present application;

[0011] Figure 2 It is a target image in an image processing method for lipoprotein subtyping provided in an embodiment of the present application;

[0012] Figure 3 A plurality of single test tube images in an image processing method for lipoprotein subtyping provided in an embodiment of the present application;

[0013] Figure 4 A plurality of ROI regions extracted in an image processing method for lipoprotein subtyping in an embodiment of the present application;

[0014] Figure 5 An analysis result without a total cholesterol value in an image processing method for lipoprotein subtyping extracted in an embodiment of the present application;

[0015] Figure 6 The analysis result of the total cholesterol value in an image processing method for lipoprotein subtyping extracted in the embodiment of the present application;

[0016] Figure 7 A schematic structural diagram of an image processing device for lipoprotein subtyping provided in an embodiment of the present application.

[0017] Explanation of the reference numerals: original image acquisition unit 11 , image preprocessing unit 12 , test tube image extraction unit 13 , effective area extraction unit 14 , ROI area analysis unit 15 , type acquisition unit 16 . DETAILED DESCRIPTION

[0018] The overall idea of ​​the technical solution provided by this application is as follows:

[0019] The embodiment of the present application provides an image processing method and device for lipoprotein subtyping to solve the problems existing in the prior art such as insufficient image preprocessing, inaccurate extraction of effective areas, and analysis methods limited by total cholesterol values.

[0020] First, the original image containing lipoprotein subtype bands is obtained and preprocessed, including grayscale correction, distortion correction, rotation correction and test tube area centering processing, to improve the image quality. Then, multiple single test tube images are automatically extracted from the preprocessed target image; then, the very low density lipoprotein VLDL band to the high density lipoprotein HDL band is used as the target area, and the region of interest is automatically extracted from the single test tube image to obtain multiple ROI areas. Subsequently, each ROI area is analyzed separately to obtain multiple analysis results including lipoprotein subtype peak type diagram and low density lipoprotein average particle size value. Afterwards, the type of low density lipoprotein particles is determined according to the low density lipoprotein average particle size value. Compared with the prior art, the present application has been systematically optimized in image processing, single test tube image extraction and image analysis, and has achieved significant improvements in automation, detection accuracy, result reliability, etc., and can obtain particle size distribution information and determine the type of low density lipoprotein particles in the absence of total cholesterol data.

[0021] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically described below in conjunction with the drawings in the specification.

[0022] Embodiment 1, as Figure 1 As shown, the embodiment of the present application provides an image processing method for lipoprotein subtyping, the method comprising:

[0023] S1: Get the original image containing the lipoprotein subtype bands.

[0024] Specifically, first, obtain an original image containing lipoprotein subtype bands. For example, use an imaging device such as a camera or a scanner to image the gel after the electrophoresis analysis to obtain an original image, which clearly reflects the distribution of lipoprotein subtype bands on the gel. Among them, the original image can be a color image, such as an RGB three-channel image, or a grayscale image. The resolution of the original image is set according to actual needs, for example, an image with a resolution of 300dpi, 600dpi or higher is used. The format of the original image is a digital image format such as BMP, JPG, TIFF, PNG, etc.

[0025] S2: Preprocessing the original image to obtain a target image, wherein the preprocessing includes grayscale correction, distortion correction, rotation correction, and test tube area centering processing.

[0026] Specifically, after obtaining the original image, a series of preprocessing operations are performed on the original image to improve the accuracy and reliability of subsequent analysis, so as to correct the deviations or interference factors in the original image, make the image more standardized, and facilitate the extraction of key information.

[0027] The preprocessing of the original image includes grayscale correction, distortion correction, rotation correction, and test tube area centering. Among them, grayscale correction is used to reduce the deviation of the image in terms of brightness, contrast, etc., so that the grayscale distribution of the image is more uniform and consistent; distortion correction is used to correct the perspective distortion or lens distortion that may exist in the original image, so that the image presents the effect of an orthographic perspective; rotation correction is used to correct the rotation angle of the original image, so that the lipoprotein bands appear vertically distributed in the image, such as by detecting the inclination angle of the test tube edge or the strip edge, and making corresponding rotation transformations to achieve this; test tube area centering is used to display the target test tube area in the center of the image, which is convenient for the subsequent segmentation of a single test tube image, such as by detecting the overall position of the test tube area and translating it to the center of the image. Figure 2 As shown, after the above preprocessing, the original image is converted into a standardized target image, which lays the foundation for subsequent test tube segmentation, image analysis and other processing.

[0028] S3: Extracting a single test tube image according to the target image to obtain a plurality of single test tube images.

[0029] Specifically, after obtaining the preprocessed target image, the image of each individual test tube is extracted from it to prepare for the subsequent ROI area segmentation and analysis. Since the actual shooting contains multiple parallel test tubes, it is necessary to effectively segment and extract the sub-images of each test tube.

[0030] First, perform a vertical projection analysis on the target image to obtain the upper and lower boundaries of the test tube area. Since the pixel value in the test tube area is higher than that in the background area, the image is summed in the vertical direction to obtain a clear peak interval, which corresponds to the vertical position of the test tube. Then, combined with the preset width information of the test tube, the test tube area is equally divided in the horizontal direction to obtain multiple candidate test tube sub-areas, where the preset width is determined according to the actual specifications of the test tube. Next, by detecting the sudden change of the pixel value at the edge of each candidate test tube sub-area, the horizontal position and width of the test tube are corrected to eliminate possible deviations and extract its accurate left and right boundaries. Afterwards, as Figure 3 As shown in the figure, according to the determined test tube boundary position, the sub-image of each test tube is segmented from the target image to obtain multiple single test tube images. When extracting single test tube images, some test tube positions may be empty, that is, there is no test tube placed at the position or no sample is added to the test tube. In this case, the pixel mean or variance and other features in the candidate test tube area are analyzed to make judgments and eliminations, ensuring that the extracted single test tube images are valid.

[0031] S4: extracting effective areas from the multiple single test tube images respectively to obtain multiple ROI areas, wherein each ROI area corresponds to a very low density lipoprotein VLDL band to a high density lipoprotein HDL band in the test tube image.

[0032] Specifically, after the single test tube image is successfully extracted, the target region of interest, i.e., the ROI region, is further segmented from each single test tube image. The ROI region refers to the effective interval containing the lipoprotein bands, specifically ranging from the starting position of the very low density lipoprotein (VLDL) band to the ending position of the high density lipoprotein (HDL) band.

[0033] First, the single test tube image is scanned line by line, and the grayscale average value of each row of pixels is calculated. Since the pixel value corresponding to the lipoprotein band is higher than that of the background area, it presents an obvious peak feature. Then, a grayscale threshold is set, and the grayscale average value of each row obtained by scanning is compared with the threshold. If the grayscale average value of a row is higher than the threshold, it is considered that the row corresponds to the lipoprotein band area; otherwise, it is considered to be the background area. In this way, a binary mask image representing the band area is obtained. Next, the binary mask image is subjected to morphological processing, such as opening and closing operations, to eliminate possible noise or breaks. Then, through connected domain analysis, the largest connected area is extracted, which is the complete band area. After that, as Figure 4As shown, according to the upper and lower boundaries of the strip area, the ROI area is cut out from the single test tube image to obtain multiple ROI areas. Considering that the positions of the VLDL and HDL strips are relatively fixed, the specific upper and lower boundary positions of the ROI area can be determined by an empirical proportional coefficient. For example, the part of the area above the upper boundary of the VLDL strip and the part of the area below the lower boundary of the HDL strip are removed, and only the main strip area in the middle is retained as the ROI.

[0034] Through the above steps, the ROI region containing VLDL to HDL is accurately extracted from each single test tube image. In practical applications, due to differences in samples and experimental conditions, the specific position and spacing of lipoprotein bands may be slightly different. Therefore, when determining the boundaries of the ROI region, a certain tolerance range can be appropriately set to ensure the robustness and reliability of the extraction.

[0035] In addition, in some embodiments, other image segmentation algorithms, such as edge detection, region growing, etc., can also be used to automatically extract the ROI region. According to the characteristics of the lipoprotein bands, such as edge gradient, regional grayscale difference, etc., the segmentation parameters can be dynamically adjusted to adapt to different image conditions.

[0036] S5: Analyze the multiple ROI regions respectively to obtain multiple analysis results, wherein each analysis result includes a lipoprotein subtyping peak pattern diagram and a low-density lipoprotein average particle size value.

[0037] Specifically, after obtaining the ROI area image of each test tube sample, it is quantitatively analyzed to extract indicators reflecting the distribution characteristics of lipoprotein subtypes. The lipoprotein status of the sample is evaluated by mapping the x-axis coordinate to a fixed length, generating a standardized lipoprotein subtype peak graph, and calculating the average particle size of low-density lipoprotein.

[0038] First, according to the lipoprotein subtype classification rules, the ROI area is divided into different areas such as VLDL, IDL, LDL, HDL, etc., each area corresponds to a specific x-axis coordinate range. Then, the x-axis coordinate range of each lipoprotein subtype area is mapped to a standardized coordinate system with a fixed length. Through coordinate mapping, the lipoprotein subtype distribution curves of different samples will have the same x-axis length. Then, as Figure 5 As shown, in the standardized coordinate system, the grayscale value of the ROI image is interpolated or resampled to obtain a standardized lipoprotein subtype peak diagram. Subsequently, based on the standardized lipoprotein subtype peak diagram, the area proportion of each subtype peak (such as LDL1 to LDL7) in the LDL region is calculated. After that, the average particle size value of LDL particles, that is, the average particle size value of low-density lipoprotein, is obtained by weighted averaging combined with the preset coefficients.

[0039] By analyzing multiple ROI regions separately, standardized lipoprotein subtype peak patterns and low-density lipoprotein average particle size values ​​are extracted from each ROI region image. Compared with the existing technology, the x-axis coordinate is mapped to a fixed length, eliminating the difference in the length of the spectrum between different samples, improving the repeatability and comparability of the experimental results, ensuring the consistency of the lipoprotein subtype peak patterns, and providing a reliable basis for subsequent sample comparison and data analysis.

[0040] S6: For each analysis result, according to the average particle size value of the low-density lipoprotein, obtain the type of low-density lipoprotein particles, and the types of low-density lipoprotein particles include type A, type B and intermediate type.

[0041] Specifically, after obtaining the analysis results of each ROI area, the corresponding low-density lipoprotein particle type is determined based on the average particle size value of the low-density lipoprotein.

[0042] First, the average particle size of low-density lipoprotein is compared with a preset particle size threshold. The preset particle size threshold has two particle size thresholds, corresponding to the typical particle size ranges of type A and type B low-density lipoprotein particles, respectively. The particle size is defined as the characteristic of type A particles and will be less than or equal to The particle size is defined as the characteristic of type B particles. Then, based on the comparison results, the type of LDL particles corresponding to different average LDL particle size values ​​is determined, so as to obtain the average LDL particle size value in the absence of total cholesterol data and determine the type of LDL particles. If the average LDL particle size value is greater than or equal to the characteristic threshold of type A particles, the type of LDL particles is determined to be type A; if the average LDL particle size value is less than or equal to the characteristic threshold of type B particles, the type of LDL particles is determined to be type B; if the average particle size value is between type A and type B, the type of LDL particles is classified as intermediate.

[0043] Furthermore, the embodiment of the present application also includes:

[0044] Step S210: performing grayscale correction and distortion correction on the original image to obtain a first processed image;

[0045] Step S220: extracting a central test tube region from the first processed image using a pixel statistics method, obtaining a regional slope according to left and right boundaries of the central test tube region, and performing rotation correction on the first processed image based on the regional slope to obtain a second processed image;

[0046] Step S230: According to the upper and lower boundaries and the left and right boundaries of the central test tube area, the offset between the central test tube area and the center of the original image is obtained, and according to the offset, the central test tube area is moved to the center position of the second processed image to obtain the target image.

[0047] In a feasible implementation, after obtaining the original image containing a plurality of test tubes, a series of preprocessing operations are performed on the original image to improve the accuracy and reliability of subsequent analysis. First, grayscale correction and distortion correction are performed on the original image to obtain a first processed image. For example, the original image is processed by using methods such as histogram equalization and gamma correction to reduce the deviation of the original image in terms of brightness, contrast, etc., so that the grayscale distribution of the image is more uniform and consistent; and the image is corrected by using the distortion coefficient obtained by camera calibration to correct the perspective distortion or lens distortion in the image, so that the image presents an orthographic projection characteristic, thereby obtaining the first processed image.

[0048] Then, by analyzing the distribution characteristics of the pixel values ​​of each row or column of the first processed image, the upper and lower and left and right boundaries of the central test tube area are determined. Since the pixel values ​​in the test tube area are usually higher than those in the background area, the boundary position is determined by detecting the mutation point of the pixel value. According to the extracted left and right boundary coordinates, the inclination angle of the central test tube area is calculated, and the entire image is subjected to a corresponding rotation transformation so that the test tube area remains vertical in the image, thereby obtaining the second processed image. Subsequently, according to the upper and lower boundaries and left and right boundaries of the central test tube area, the position offset of the area in the second processed image is obtained, and the central test tube area is moved to the center position of the second processed image through a translation transformation to obtain the target image, which lays the foundation for subsequent single test tube extraction, ROI area segmentation and other operations.

[0049] Furthermore, the embodiment of the present application also includes:

[0050] Step S310: extracting a central test tube region in the target image, wherein the central test tube region includes a plurality of single test tube images;

[0051] Step S320: dividing the central test tube area into a plurality of candidate test tube areas according to a preset width of the single test tube image;

[0052] Step S330: for each of the candidate test tube regions, respectively determine the left and right boundaries thereof, and correct the boundary of each of the candidate test tube regions according to the determined left and right boundaries, to obtain a plurality of test tube region images;

[0053] Step S340: judging the grayscale distribution of the multiple test tube area images, excluding images without test tubes, and obtaining multiple single test tube images.

[0054] In a preferred embodiment, after obtaining the preprocessed target image, the image of each individual test tube is extracted therefrom to prepare for the subsequent ROI area segmentation and analysis. Since one image contains multiple parallel test tubes during actual shooting, it is necessary to effectively segment and extract the sub-images of each test tube from the central test tube area to obtain multiple single test tube images.

[0055] First, the central test tube area in the target image is extracted. This area contains all the individual test tube images to be analyzed, thereby excluding irrelevant background areas in the target image, reducing the size of the data for subsequent processing, and improving computational efficiency. Then, the extracted central test tube area is divided equally according to the preset width of the individual test tube image to obtain multiple candidate test tube areas. The preset width is set according to parameters such as the actual physical size of the test tube and the magnification of the imaging system to ensure that each candidate area can completely contain a test tube image, thereby preliminarily dividing the central test tube area into multiple sub-areas, providing a basis for further boundary correction.

[0056] Next, the boundaries of each candidate test tube region are corrected to obtain an accurate single test tube image region. Specifically, the left and right boundaries of each candidate region are determined separately. For example, by analyzing the horizontal projection histogram of the pixel values ​​in the candidate test tube region, the true left and right boundaries of the test tube are inferred based on the position and width of the projection peak and other features. According to the corrected boundary position, the candidate region is cropped or expanded to obtain a complete and accurate single test tube region image, effectively eliminating the segmentation error caused by factors such as uneven test tube spacing and changes in tube wall width. Subsequently, the corrected test tube region image is screened to exclude images without test tubes, and the final single test tube image set is obtained. Specifically, the grayscale distribution characteristics of each test tube region image, such as grayscale mean, variance, etc., are analyzed. For areas without test tubes, their grayscale distribution is significantly different from that of areas with test tubes. By setting threshold conditions, invalid areas are automatically identified and eliminated, thereby obtaining truly effective multiple single test tube images.

[0057] Furthermore, the embodiment of the present application also includes:

[0058] Step S410: for each single test tube image, the grayscale average value of pixels in each row of the image is calculated along the length direction of the test tube to obtain a one-dimensional grayscale array;

[0059] Step S420: setting a preset threshold, clearing the grayscale values ​​in the one-dimensional grayscale array that are higher than the preset threshold to obtain an updated one-dimensional grayscale array;

[0060] Step S430: searching for continuous zero-value intervals in the updated one-dimensional grayscale array to determine the upper boundary of the very low-density lipoprotein VLDL band and the lower boundary of the high-density lipoprotein HDL band;

[0061] Step S440: extracting a region located in the single test tube image as a ROI region according to the upper boundary of the very low density lipoprotein VLDL band and the lower boundary of the high density lipoprotein HDL band.

[0062] In a feasible implementation, after the single test tube images are successfully extracted, the target region of interest, i.e., the ROI region, is further segmented from each single test tube image, wherein the ROI region specifically refers to the effective interval containing the lipoprotein band to be analyzed, and specifically ranges from the upper boundary of the very low density lipoprotein band to the lower boundary of the high density lipoprotein band.

[0063] First, each single test tube image is scanned line by line, and the grayscale average of each row of pixels in the image is calculated to obtain a one-dimensional grayscale array that represents the grayscale distribution of the test tube image. Specifically, each row of the image is traversed, the grayscale values ​​of all pixels in the row are extracted, and their arithmetic average is calculated, so that the two-dimensional grayscale distribution of the single test tube image is converted into a one-dimensional grayscale array, reflecting the grayscale change trend along the length of the test tube. Secondly, the one-dimensional grayscale array is thresholded, and the grayscale values ​​above the preset threshold are cleared to obtain an updated one-dimensional grayscale array. Among them, the preset threshold is set according to prior knowledge such as the average grayscale level of the test tube image and the typical grayscale range of the lipoprotein band. Through thresholding, the grayscale values ​​of the background area and the non-target strip area are suppressed to zero, while the grayscale value of the target strip area is retained, which effectively highlights the position of the target area and creates conditions for subsequent boundary positioning.

[0064] Subsequently, continuous zero value intervals are searched in the updated one-dimensional grayscale array to determine the upper boundary of the VLDL band and the lower boundary of the HDL band. Specifically, starting from the starting position of the updated one-dimensional grayscale array, each element of the array is traversed to record the starting position and the ending position of the continuous zero value. Among them, there will be three continuous zero value intervals in the updated one-dimensional grayscale array, corresponding to the background area above the VLDL band, between the VLDL band and the HDL band, and below the HDL band. By locating the ending position of the first continuous zero value interval and the starting position of the third continuous zero value interval, the upper boundary of the VLDL band and the lower boundary of the HDL band can be determined. Then, according to the located upper boundary of the VLDL band and the lower boundary of the HDL band, the corresponding area is extracted from the original single test tube image to obtain the ROI area. Specifically, using these two boundary positions as the upper and lower boundaries and the left and right boundaries of the test tube image as the left and right boundaries, the target area of ​​interest is cropped out from the single test tube image, thereby further focusing the single test tube image to the effective area containing the lipoprotein bands to be analyzed, laying the foundation for subsequent detailed analysis.

[0065] Furthermore, the embodiment of the present application also includes:

[0066] Step S510: establishing a standardized coordinate system;

[0067] Step S530: in the standardized coordinate system, the regional ranges of VLDL, IDL, LDL and HDL are divided, and the curve area ratio of each region is calculated to form a lipoprotein subtyping peak diagram;

[0068] Step S540: Calculate the average particle size of low-density lipoprotein according to the area ratio of the LDL region curve in the lipoprotein subtyping peak diagram according to the following formula:

[0069] Mean LDL Particle Size=∑α i L i ;

[0070] Among them, Mean LDL Particle Size is the average particle size of low-density lipoprotein, L i Represents the area proportion of each low-density lipoprotein subtype, and the low-density lipoprotein subtype includes LDL-1, LDL-2, LDL-3, LDL-4, LDL-5, LDL-6, LDL-7, α i represents the preset empirical parameters for each LDL subtype;

[0071] Step S550: respectively calculating the average particle size values ​​of low-density lipoprotein in multiple ROI regions to obtain multiple analysis results.

[0072] In a preferred embodiment, after obtaining the ROI area image of each test tube, it is subjected to quantitative analysis, and by generating a lipoprotein subtyping peak diagram, the average particle size value of low-density lipoprotein is calculated without the need for total cholesterol data.

[0073] In order to achieve comparability and consistency of the analysis results of different test tubes, the horizontal coordinates of the ROI area were linearly transformed and mapped to a standardized coordinate system of fixed length. Through coordinate transformation, the ROI areas of different test tubes were aligned and scaled uniformly. Then, in the standardized coordinate system, the regional ranges of VLDL, IDL, LDL, and HDL were divided, and the curve area ratio of each area was calculated to generate a lipoprotein subtype peak diagram. According to the relative position and mobility difference of lipoprotein subtypes in the electrophoresis spectrum, the starting and ending positions of each area in the standardized coordinate system were predefined. Then, the grayscale curve of the ROI area image was integrated and calculated to obtain the curve area under each area. The ratio of the area of ​​each area to the total area was taken as the area ratio of the area, and a peak diagram reflecting the relative content of lipoprotein subtypes was generated, so that the ROI area image was converted into quantitative lipoprotein subtype distribution information.

[0074] Then, the average particle size of low-density lipoprotein is calculated based on the area ratio of the LDL region curve in the lipoprotein subtype peak diagram. In order to quantitatively characterize the size of LDL particles, the weighted average method is used to combine the area ratio of each LDL subtype and the preset empirical parameters to calculate the average particle size of low-density lipoprotein. The calculation formula is: Mean LDL ParticleSize = ∑α i L i ; Mean LDL Particle Size is the average particle size of low-density lipoprotein, L i Represents the area proportion of each low-density lipoprotein subtype, which includes LDL-1, LDL-2, LDL-3, LDL-4, LDL-5, LDL-6, LDL-7, α i Represents the preset empirical parameters of each low-density lipoprotein subtype. Subsequently, the calculation of the low-density lipoprotein average particle size value is repeated for the ROI area of ​​each test tube to obtain a set of low-density lipoprotein average particle size values ​​as multiple analysis results to achieve horizontal comparison and analysis of multiple test tubes.

[0075] Furthermore, the embodiment of the present application also includes:

[0076] Step S610: If the average particle size of low-density lipoprotein is greater than or equal to Then the type of low-density lipoprotein particles is type A;

[0077] Step S620: If the average particle size of low-density lipoprotein is less than or equal to Then the type of low-density lipoprotein particles is type B;

[0078] Step S630: If the average particle size of low-density lipoprotein is greater than and less than The type of low-density lipoprotein particles is intermediate.

[0079] In a preferred embodiment, after obtaining the analysis results of each ROI area, the type of low-density lipoprotein particles is further determined. First, it is determined whether the average particle size of low-density lipoprotein is greater than or equal to If the mean LDL particle size is greater than or equal to Then, it is classified as type A. Then, it is determined whether the average particle size of low-density lipoprotein is less than or equal to If the mean LDL particle size is less than or equal to It is classified as type B. and The type of low-density lipoprotein particles is automatically determined and classified by dividing the low-density lipoprotein particles into three types: type A, type B and intermediate.

[0080] Furthermore, the embodiment of the present application also includes:

[0081] Step S710: obtaining the total cholesterol value and obtaining the area ratio of each intermediate density lipoprotein subtype in the intermediate density lipoprotein;

[0082] Step S720: Based on the total cholesterol value and the area ratio of each intermediate density lipoprotein subtype, the low density lipoprotein cholesterol content is calculated according to the following formula:

[0083] Total LDL-C=∑L i Chol+∑M j Chol;

[0084] Among them, Total LDL-C is the low-density lipoprotein cholesterol content, Chol is the total cholesterol value, L i is the area proportion of each low-density lipoprotein subtype, M j The area proportion of each intermediate density lipoprotein subtype, wherein the intermediate density lipoprotein subtype includes MDL-C, MDL-B, and MDL-A.

[0085] In a feasible embodiment, after obtaining the lipoprotein subtype peak pattern and the low-density lipoprotein average particle size value, if the total cholesterol measurement value is provided at the same time, the absolute content of low-density lipoprotein cholesterol in the sample can be further calculated.

[0086] First, the total cholesterol value of the sample was obtained, and the area percentage data of each subtype of intermediate density lipoprotein (MDL-C, MDL-B, MDL-A) were extracted from the lipoprotein subtype peak diagram. Then, the total cholesterol value, the area percentage of each subtype of low density lipoprotein, and the area percentage of each subtype of intermediate density lipoprotein were used to calculate the low density lipoprotein cholesterol content using the following formula:

[0087] Total LDL-C=∑L i Chol+∑M j Chol;

[0088] Among them, Total LDL-C is the low-density lipoprotein cholesterol content, Chol is the total cholesterol value, L i is the area proportion of each low-density lipoprotein subtype, M j The area proportion of each intermediate density lipoprotein subtype, wherein the intermediate density lipoprotein subtype includes MDL-C, MDL-B, and MDL-A.

[0089] For example, Figure 6 As shown in the figure, for a test tube with a total cholesterol of 120 mg / dL, the following results were obtained through lipoprotein subtyping analysis: LDL1 accounts for 20.6% of the area, LDL2 accounts for 9.7%, LDL3 accounts for 2.0%, and LDL4 to LDL7 account for 0; MDL-C accounts for 4.4%, MDL-B accounts for 13.6%, and MDL-A accounts for 17.0%; if the preset empirical parameters α1=α2=α3=α4=α5=α6=α7=835. Without using total cholesterol data, the average particle size of low-density lipoprotein can be calculated by the formula: Mean LDL Particle Size=Σα i L i The average particle size of low-density lipoprotein is Because its value is greater than Therefore, the type of low-density lipoprotein particles is TYPE A. When using total cholesterol data, the low-density lipoprotein cholesterol content is calculated using the formula: Total LDL-C = ΣL i Chol+∑M j Chol, the corresponding low-density lipoprotein cholesterol content of the test tube is (20.6%+9.7%+2.0%+4.4%+13.6%+17.0%)×120=38.76+42=81 mg / dL.

[0090] In summary, the image processing method for lipoprotein subtyping provided in the embodiment of the present application has the following technical effects:

[0091] The original image containing lipoprotein subtype bands is obtained to provide raw data for subsequent processing. The original image is preprocessed to obtain a target image, and the preprocessing includes grayscale correction, distortion correction, rotation correction, and test tube area centering processing, which effectively eliminates the noise and distortion introduced by factors such as shooting angle and lighting conditions in the image, lays a good data foundation for subsequent analysis, and helps to improve the accuracy and robustness of the entire detection process. According to the target image, a single test tube image is extracted to obtain multiple single test tube images, and the automatic splitting and extraction of single test tube images are realized, which improves the processing efficiency and batch analysis capabilities. For multiple single test tube images, the effective areas are extracted respectively to obtain multiple ROI areas, wherein each ROI area corresponds to the very low density lipoprotein VLDL band to the high density lipoprotein HDL band in the test tube image, and the VLDL band to the HDL band is taken as the target. The ROI area is determined by an algorithm based on pixel threshold analysis, which avoids problems such as positioning errors caused by human factors, and provides high-quality data support for subsequent analysis. Analyze multiple ROI areas separately to obtain multiple analysis results, wherein each analysis result includes a lipoprotein subtyping peak diagram and an average particle size value of low-density lipoprotein, realizing the calculation of the average particle size of low-density lipoprotein in the absence of total cholesterol data. For each analysis result, the type of low-density lipoprotein particles is obtained according to the average particle size value of low-density lipoprotein, and the types of low-density lipoprotein particles include type A, type B, and intermediate types, realizing the accurate calculation of the average particle size value of low-density lipoprotein and the judgment of the type of low-density lipoprotein particles in the absence of total cholesterol values.

[0092] Embodiment 2, based on the same inventive concept as the image processing method for lipoprotein subtyping in the above embodiment, Figure 7 As shown, the embodiment of the present application provides an image processing device for lipoprotein subtyping, the device comprising:

[0093] The original image acquisition unit 11 is used to acquire the original image containing the lipoprotein subtype bands;

[0094] An image preprocessing unit 12 is used to preprocess the original image to obtain a target image, wherein the preprocessing includes grayscale correction, distortion correction, rotation correction, and test tube area centering processing;

[0095] A test tube image extraction unit 13 is used to extract a single test tube image according to the target image to obtain a plurality of single test tube images;

[0096] The effective region extraction unit 14 is used to extract effective regions from the multiple single test tube images respectively to obtain multiple ROI regions, wherein each ROI region corresponds to the very low density lipoprotein VLDL band to the high density lipoprotein HDL band in the test tube image;

[0097] The ROI region analysis unit 15 is used to analyze the multiple ROI regions respectively to obtain multiple analysis results, wherein each analysis result includes a lipoprotein subtype peak diagram and a low-density lipoprotein average particle size value;

[0098] The type acquisition unit 16 is used to acquire the type of low-density lipoprotein particles according to the average particle size of low-density lipoprotein particles in each analysis result, and the types of low-density lipoprotein particles include type A, type B and intermediate type.

[0099] Furthermore, the image preprocessing unit 12 includes the following execution steps:

[0100] Performing grayscale correction and distortion correction on the original image to obtain a first processed image;

[0101] extracting a central test tube region from the first processed image using a pixel statistics method, obtaining a regional slope according to left and right boundaries of the central test tube region, and performing rotation correction on the first processed image based on the regional slope to obtain a second processed image;

[0102] The offset between the central test tube area and the center of the original image is obtained according to the upper and lower boundaries and the left and right boundaries of the central test tube area, and the central test tube area is moved to the center position of the second processed image according to the offset to obtain the target image.

[0103] Further, the test tube image extraction unit 13 includes the following execution steps:

[0104] Extracting a central test tube region in the target image, wherein the central test tube region includes a plurality of single test tube images;

[0105] Dividing the central test tube area into a plurality of candidate test tube areas according to a preset width of the single test tube image;

[0106] For each of the candidate test tube regions, respectively determine its left and right boundaries, and correct the boundary of each of the candidate test tube regions according to the determined left and right boundaries to obtain a plurality of test tube region images;

[0107] The grayscale distribution of multiple test tube area images is judged, and images without test tubes are excluded to obtain multiple single test tube images.

[0108] Furthermore, the effective area extraction unit 14 includes the following execution steps:

[0109] For each of the single test tube images, the grayscale average value of pixels in each row of the image is calculated along the length direction of the test tube to obtain a one-dimensional grayscale array;

[0110] Setting a preset threshold, clearing the grayscale values ​​in the one-dimensional grayscale array that are higher than the preset threshold to obtain an updated one-dimensional grayscale array;

[0111] Searching for continuous zero-value intervals in the updated one-dimensional grayscale array to determine the upper boundary of the very low-density lipoprotein VLDL band and the lower boundary of the high-density lipoprotein HDL band;

[0112] According to the upper boundary of the very low density lipoprotein VLDL band and the lower boundary of the high density lipoprotein HDL band, a region located therein is extracted from the single test tube image as a ROI region.

[0113] Furthermore, the ROI region analysis unit 15 includes the following execution steps:

[0114] Establish a standardized coordinate system;

[0115] In the standardized coordinate system, the regional ranges of VLDL, IDL, LDL, and HDL are divided, and the curve area ratio of each region is calculated to form a lipoprotein subtyping peak diagram;

[0116] According to the area ratio of the LDL region curve in the lipoprotein subtyping peak diagram, the average particle size of low-density lipoprotein is calculated as follows:

[0117] Mean LDL Particle Size=∑α i L i ;

[0118] Among them, Mean LDL Particle Size is the average particle size of low-density lipoprotein, L i Represents the area proportion of each low-density lipoprotein subtype, and the low-density lipoprotein subtype includes LDL-1, LDL-2, LDL-3, LDL-4, LDL-5, LDL-6, LDL-7, α i represents the preset empirical parameters for each LDL subtype;

[0119] The average particle size values ​​of low-density lipoprotein in multiple ROI areas are calculated respectively to obtain multiple analysis results.

[0120] Furthermore, the type acquisition unit 16 includes the following execution steps:

[0121] If the average particle size of low-density lipoprotein is greater than or equal to Then the type of low-density lipoprotein particles is type A;

[0122] If the average particle size of low-density lipoprotein is less than or equal to Then the type of low-density lipoprotein particles is type B;

[0123] If the average particle size of low-density lipoprotein is greater than and less than The type of low-density lipoprotein particles is intermediate.

[0124] Furthermore, the embodiment of the present application also includes a cholesterol content acquisition unit, which includes the following execution steps:

[0125] Obtain the total cholesterol value and the area proportion of each intermediate density lipoprotein subtype in the intermediate density lipoprotein;

[0126] Based on the total cholesterol value and the area ratio of each intermediate density lipoprotein subtype, the low density lipoprotein cholesterol content is calculated according to the following formula:

[0127] Total LDL-C=∑L i Chol+∑M j Chol;

[0128] Among them, Total LDL-C is the low-density lipoprotein cholesterol content, Chol is the total cholesterol value, L i is the area proportion of each low-density lipoprotein subtype, M j The area proportion of each intermediate density lipoprotein subtype, wherein the intermediate density lipoprotein subtype includes MDL-C, MDL-B, and MDL-A;

[0129] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0130] Furthermore, the first or second mentioned above may not only represent an order relationship, but may also represent a specific concept, and / or refer to multiple elements that can be selected individually or in full. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. An image processing method for lipoprotein subtyping, characterized in that: The method comprises: Acquire the original image containing the lipoprotein subtype bands; Preprocessing the original image to obtain a target image, wherein the preprocessing includes grayscale correction, distortion correction, rotation correction, and test tube area centering processing; Extracting a single test tube image according to the target image to obtain a plurality of single test tube images; Extracting effective areas from multiple single test tube images respectively to obtain multiple ROI areas, wherein each ROI area corresponds to a very low density lipoprotein VLDL band to a high density lipoprotein HDL band in the test tube image; Analyze multiple ROI regions respectively to obtain multiple analysis results, wherein each analysis result includes a lipoprotein subtype peak diagram and a low-density lipoprotein average particle size value; For each analysis result, the type of low-density lipoprotein particles is obtained according to the average particle size value of the low-density lipoprotein therein, and the types of low-density lipoprotein particles include type A, type B and intermediate type.

2. The image processing method for lipoprotein subtyping according to claim 1, characterized in that: Preprocessing the original image to obtain a target image includes: Performing grayscale correction and distortion correction on the original image to obtain a first processed image; extracting a central test tube region from the first processed image using a pixel statistics method, obtaining a regional slope according to left and right boundaries of the central test tube region, and performing rotation correction on the first processed image based on the regional slope to obtain a second processed image; The offset between the central test tube area and the center of the original image is obtained according to the upper and lower boundaries and the left and right boundaries of the central test tube area, and the central test tube area is moved to the center position of the second processed image according to the offset to obtain the target image.

3. The image processing method for lipoprotein subtyping according to claim 2, characterized in that: According to the target image, a single test tube image is extracted to obtain multiple single test tube images, including: Extracting a central test tube region in the target image, wherein the central test tube region includes a plurality of single test tube images; Dividing the central test tube area into a plurality of candidate test tube areas according to a preset width of the single test tube image; For each of the candidate test tube regions, respectively determine its left and right boundaries, and correct the boundary of each of the candidate test tube regions according to the determined left and right boundaries to obtain a plurality of test tube region images; The grayscale distribution of multiple test tube area images is judged, and images without test tubes are excluded to obtain multiple single test tube images.

4. The image processing method for lipoprotein subtyping according to claim 1, characterized in that: For multiple single test tube images, valid areas are extracted respectively to obtain multiple ROI areas, including: For each of the single test tube images, the grayscale average value of pixels in each row of the image is calculated along the length direction of the test tube to obtain a one-dimensional grayscale array; Setting a preset threshold, clearing the grayscale values ​​in the one-dimensional grayscale array that are higher than the preset threshold to obtain an updated one-dimensional grayscale array; Searching for continuous zero-value intervals in the updated one-dimensional grayscale array to determine the upper boundary of the very low-density lipoprotein VLDL band and the lower boundary of the high-density lipoprotein HDL band; According to the upper boundary of the very low density lipoprotein VLDL band and the lower boundary of the high density lipoprotein HDL band, a region located therein is extracted from the single test tube image as a ROI region.

5. The image processing method for lipoprotein subtyping according to claim 1, characterized in that: Analyze multiple ROI areas separately to obtain multiple analysis results, including: Establish a standardized coordinate system; In the standardized coordinate system, the regional ranges of VLDL, IDL, LDL, and HDL are divided, and the curve area ratio of each region is calculated to form a lipoprotein subtyping peak diagram; According to the area ratio of the LDL region curve in the lipoprotein subtyping peak diagram, the average particle size of low-density lipoprotein is calculated as follows: Mean LDL Particle Size=Σα i L i ; Among them, Mean LDL Particle Size is the average particle size of low-density lipoprotein, L i Represents the area proportion of each low-density lipoprotein subtype, and the low-density lipoprotein subtype includes LDL-1, LDL-2, LDL-3, LDL-4, LDL-5, LDL-6, LDL-7, α i represents the preset empirical parameters for each LDL subtype; The average particle size values ​​of low-density lipoprotein in multiple ROI areas are calculated respectively to obtain multiple analysis results.

6. The image processing method for lipoprotein subtyping according to claim 5, characterized in that: For each analysis result, according to the average particle size value of the low-density lipoprotein, the type of low-density lipoprotein particles is obtained, including: If the average particle size of low-density lipoprotein is greater than or equal to Then the type of low-density lipoprotein particles is type A; If the average particle size of low-density lipoprotein is less than or equal to Then the type of low-density lipoprotein particles is type B; If the average particle size of low-density lipoprotein is greater than and less than The type of low-density lipoprotein particles is intermediate.

7. The image processing method for lipoprotein subtyping according to claim 5, characterized in that: The method further comprises: Obtain the total cholesterol value and the area proportion of each intermediate density lipoprotein subtype in the intermediate density lipoprotein; Based on the total cholesterol value and the area ratio of each intermediate density lipoprotein subtype, the low density lipoprotein cholesterol content is calculated according to the following formula: TotalLDL-C=ΣL i Chol+ΣM j Chol; Among them, TotalLDL-C is the low-density lipoprotein cholesterol content, Chol is the total cholesterol value, L i is the area proportion of each low-density lipoprotein subtype, M j The area proportion of each intermediate density lipoprotein subtype, wherein the intermediate density lipoprotein subtype includes MDL-C, MDL-B, and MDL-A.

8. An image processing device for lipoprotein subtyping, characterized in that: For implementing the image processing method for lipoprotein subtyping according to any one of claims 1 to 7, the device comprises: An original image acquisition unit, wherein the original image acquisition unit is used to acquire an original image containing lipoprotein subtype bands; An image preprocessing unit, the image preprocessing unit is used to preprocess the original image to obtain a target image, the preprocessing includes grayscale correction, distortion correction, rotation correction, and test tube area centering processing; A test tube image extraction unit, the test tube image extraction unit is used to extract a single test tube image according to the target image to obtain a plurality of single test tube images; An effective region extraction unit, the effective region extraction unit is used to extract effective regions from a plurality of single test tube images respectively to obtain a plurality of ROI regions, wherein each ROI region corresponds to a very low density lipoprotein VLDL band to a high density lipoprotein HDL band in the test tube image; A ROI region analysis unit, wherein the ROI region analysis unit is used to analyze multiple ROI regions respectively to obtain multiple analysis results, wherein each analysis result includes a lipoprotein subtyping peak diagram and a low-density lipoprotein average particle size value; A type acquisition unit is used to acquire the type of low-density lipoprotein particles according to the average particle size value of the low-density lipoprotein in each analysis result, and the low-density lipoprotein particle types include type A, type B and intermediate type.

Images