Image detection-based method for evaluating the chromatographic effect of blood-separated plasma samples

Abstract

The present invention discloses a method for evaluating the chromatographic effect of a blood-separated plasma sample based on image detection. First, a high-resolution camera or a video camera is used to collect a blood separation image in real time, and a suitable light source is configured to eliminate errors caused by changes in the color or transparency of the blood. Furthermore, the collected image is preprocessed (such as grayscale processing) to convert the original blood stratification effect into a clear grayscale image. Furthermore, after the blood is centrifuged to form a red blood cell layer, a white blood cell-platelet layer, and a plasma layer, and then the top layer of the blood, i.e., the plasma layer, or an area containing a small portion of the white blood cell-platelet layer, is quickly located and extracted through coarse segmentation. Furthermore, through fine segmentation, the white blood cell-platelet layer is removed from the plasma layer to obtain an image with clear edges and containing only the plasma layer. Finally, an evaluation is performed based on the impurity content of the plasma layer in the image, and a purity analysis result is ultimately given.

Description

Technical Field

[0001] The present invention relates to the field of blood analysis, and in particular to a method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection. Background Art

[0002] Blood component separators are widely used in clinical treatments such as plasma exchange, immunotherapy, and hemodialysis. They aim to separate different components of the blood (such as plasma, red blood cells, white blood cells, and platelets) to obtain pure plasma or other blood components for patient treatment. Plasma plays a vital role in these treatments, especially in regulating the immune system and providing proteins and antibodies. However, the purity of plasma during the blood separation process directly affects the treatment effect and patient safety. Therefore, after the plasma is separated using a blood component separator, the plasma purity needs to be tested and evaluated. Only when the test and evaluation meet the set conditions can it be used in clinical treatment.

[0003] Currently, blood component separators use physical methods such as centrifugation and filtration to separate blood, and monitor the separation results in real time using optical, electronic, or sensor technologies. These monitoring technologies are used to assess the effectiveness of the separation, ensure plasma purity, and prevent contamination by impurities such as red and white blood cells. However, existing technologies still lack accuracy in assessing separation effectiveness, particularly in terms of real-time monitoring of plasma quality, purity, and contaminant detection during the separation process.

[0004] Due to the limitations of detection technology, doctors often rely on relatively crude indicators to judge plasma purity and separation effectiveness when using a separator to separate blood components, lacking detailed quality assessment tools. This can result in blood separation results in clinical treatments failing to achieve the expected purity, which in turn affects treatment effectiveness. For example, if a certain proportion of red blood cells or white blood cells remains in the plasma during the plasma exchange process, it may trigger an immune response or other complications. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection, which solves the above-mentioned technical problems pointed out in the prior art.

[0006] The present invention provides a method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection, comprising the following steps:

[0007] Acquiring a blood separation image in real time, and preprocessing the blood separation image to obtain a first blood separation image;

[0008] performing coarse segmentation on the first blood separation image to obtain a second blood separation image;

[0009] performing fine segmentation on the second blood separation image to obtain a third blood separation image;

[0010] Purity assessment processing and analysis are performed based on the third blood separation image to obtain a purity analysis result.

[0011] Preferably, the step of finely segmenting the second blood separation image to obtain the third blood separation image comprises the following steps:

[0012] constructing a two-dimensional image matrix M(i, j) based on the grayscale values ​​of each pixel in the second blood separation image, wherein each element in the two-dimensional image matrix M(i, j) represents the grayscale value corresponding to each pixel position (i, j) in the second blood separation image; preprocessing the two-dimensional image matrix M(i, j) to obtain a first two-dimensional image matrix M'(i, j);

[0013] Perform feature extraction based on the first two-dimensional image matrix M'(i, j) to obtain multidimensional feature data F n (i,j);

[0014] Based on the multidimensional feature data F n (i, j) clustering each pixel point in the second blood separation image to obtain a plurality of clusters; constructing an image of the area to be processed corresponding to each cluster based on all pixel points in the bounding box of each cluster;

[0015] The image of the area to be processed corresponding to the highest vertical position among the images of the area to be processed is selected as the head area image; and a plurality of images of the area to be processed are merged through a region growing operation based on the head area image to obtain a third blood separation image.

[0016] Preferably, the multidimensional feature data F n (i, j) includes grayscale distribution feature data F1 (i, j), texture feature data F2 (i, j) and edge feature data F3 (i, j).

[0017] Preferably, the step of merging a plurality of to-be-processed region images through a region growing operation based on the head region image to obtain a third blood separation image comprises the following steps:

[0018] Extract the texture feature data in the head area image; and calculate the gray value mean μ of the pixel points in the head area image i' And the standard deviation of the gray value of pixels in the head area image μ n; At the same time, the energy parameter factor of the head area image is obtained; Based on the energy parameter factor, the comprehensive energy quality score Q of the head area image is calculated by a preset energy function n ;

[0019] Based on the texture feature data of the current head area image and the grayscale value mean μ of the pixels in the head area image i' , the standard deviation of the gray value of pixels in the head area image μ n And the comprehensive energy quality score Q of the head area image n The image of the area to be processed is combined with the image of the head area to obtain a third blood separation image.

[0020] Preferably, the energy parameter factor includes gray uniformity U h , Regional Connectivity C c and shape regularity R s ;

[0021] The comprehensive energy quality score Q n The calculation method is:

[0022]

[0023] Where Q n is the comprehensive energy quality score; U h is the grayscale uniformity; C c is regional connectivity; R s is the shape regularity; is the ideal value of grayscale uniformity; I(i,j) is the grayscale value of the image at position (i,j); μ i' is the mean of the image grayscale values; μ n is the standard deviation of the image grayscale values.

[0024] Preferably, the texture feature data based on the current head region image and the mean grayscale value μ of the pixels in the head region image are i' , the standard deviation of the gray value of pixels in the head area image μ n And the comprehensive energy quality score Q of the head area image n Merging the image of the area to be processed with the image of the head area to obtain a third blood separation image includes the following steps:

[0025] Start traversing from the image to be processed that is adjacent to the current head region image among all the image to be processed region images: calculate the texture similarity S between the image to be processed region image and the head region image t (R i' ); Based on the texture similarity, the mean grayscale value μ of the pixels in the head area imagei' And the standard deviation of the gray value of pixels in the head area image μ n The grayscale value of each pixel in the image of the area to be processed is used to calculate the comprehensive similarity S(R i' );

[0026] Determine the comprehensive similarity S(R i' ) is greater than or equal to a preset minimum comprehensive similarity threshold; if so, the image of the area to be processed is determined as the image to be merged; if not, the image of the area to be processed is placed in the set of images of the area to be supplemented;

[0027] constructing a first head region image based on the image to be merged and the head region image; and obtaining an energy parameter factor of the first head region image;

[0028] Based on the energy parameter factor of the first head region image, a comprehensive energy quality score Q' of the first head region image is calculated by a preset energy function. n ; Determine the comprehensive energy quality score Q' of the first head region image n Is it greater than or equal to the comprehensive energy quality score Q of the head area image? n If so, the first head region image is determined to be a new head region image, and the new head region image is returned to the above operation to continue traversing until all the region images to be processed are traversed, and the final new head region image is output as the third blood separation image to be determined;

[0029] Image optimization processing is performed based on the connected domain analysis of the third blood separation image to be determined and the image set of the area to be supplemented to obtain the third blood separation image.

[0030] Preferably, the comprehensive similarity S(R i' ) is calculated as:

[0031]

[0032] Where w g 、w t are weight coefficients; μ i' is the mean gray value of the pixels in the head area image μ i' ;μ n is the standard deviation of the grayscale value of pixels in the head area image μ n ; σ n is the standard deviation of the grayscale values ​​of pixels in the image of the current area to be processed; S t (R i’) is the texture similarity between the image of the area to be processed and the image of the head area.

[0033] Preferably, the image optimization processing based on the connected domain analysis of the third blood separation image to be determined and the image set of the area to be supplemented to obtain the third blood separation image includes the following steps:

[0034] Obtaining a hollow area through connected domain analysis based on the third blood separation image to be determined;

[0035] Establishing an association relationship based on the position of the hollow area image and the position of each of the to-be-processed area images in the to-be-supplemented area image set, obtaining the to-be-supplemented area image corresponding to the hollow area;

[0036] The image of the area to be supplemented is supplemented to a position corresponding to the hollow area to obtain a third blood separation image.

[0037] Preferably, the method of obtaining the hollow area through connected domain analysis based on the third blood separation image to be determined includes the following steps:

[0038] extracting an outer contour of the third blood separation image to be determined;

[0039] A closed contour inside the outer contour is identified based on the outer contour to obtain a hollow area.

[0040] Another aspect of the present invention provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection.

[0041] Compared with the prior art, the embodiments of the present invention have at least the following technical advantages:

[0042] From the analysis of the above-mentioned chromatographic effect evaluation method of blood separation plasma sample based on image detection provided by the present invention, it can be known that in specific application, the blood separation image is first collected in real time by a high-resolution camera or a camera to ensure high and clear image quality, and a suitable light source (such as transmitted light or reflected light) is configured to eliminate the error caused by the color or transparency change of the blood, so as to obtain a more accurate image; and the collected image is preprocessed (such as grayscale processing) to convert the original blood stratification effect into a clear grayscale image, which is convenient for subsequent segmentation and purity analysis; further, after the blood is centrifuged, a red blood cell layer, a white blood cell-platelet layer and a plasma layer are formed, and then the top layer of the blood is quickly located and extracted by coarse segmentation, that is, The plasma layer, or the area containing a small amount of white blood cell-platelet layer, can be quickly located to simplify the calculation amount and provide a data basis for subsequent fine segmentation; furthermore, the second blood separation image may still contain part of the white blood cell-platelet layer, which will cause errors in the purity assessment of the plasma layer. Through fine segmentation, the white blood cell-platelet layer is removed from the plasma layer to obtain an image with clear edges and only containing the plasma layer; finally, by adopting fuzzy comprehensive evaluation method, multi-criteria decision evaluation method, statistical evaluation method or expert decision evaluation method, the impurity content of the plasma layer in the image is evaluated, and finally a purity analysis result is given, which provides a quantitative basis for the blood separation effect, helps doctors judge the quality of separated blood components, and further improves the treatment effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of the main process of a method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection;

[0044] Figure 2 A schematic diagram of a blood separation image simulation in a method for evaluating the chromatographic effect of blood separation plasma samples based on image detection;

[0045] Figure 3 A schematic diagram of a coarse segmentation image simulation in a method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection;

[0046] Figure 4 A schematic diagram of image simulation of a plasma layer in a method for evaluating the chromatographic effect of blood-plasma samples based on image detection;

[0047] Figure 5 Schematic diagram of the operating steps for obtaining a third blood separation image by fine segmentation in a method for evaluating the chromatographic effect of a blood separation plasma sample based on image detection;

[0048] Figure 6 Schematic diagram of the operation steps for merging and obtaining a third blood separation image in a method for evaluating the chromatographic effect of blood separation plasma samples based on image detection;

[0049] Figure 7 Schematic diagram of the operation steps for merging and obtaining a third blood separation image in a method for evaluating the chromatographic effect of blood separation plasma samples based on image detection;

[0050] Figure 8 A schematic diagram of a hollow region simulation in a method for evaluating the chromatographic effect of blood-plasma samples based on image detection;

[0051] Figure 9 The present invention is a schematic diagram of the operation steps for optimizing the processing to obtain a third blood separation image in a method for evaluating the chromatographic effect of blood separation plasma samples based on image detection. DETAILED DESCRIPTION

[0052] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0053] The present invention will be further described in detail below through specific embodiments in conjunction with the accompanying drawings.

[0054] like Figure 1 As shown, an embodiment of the present invention provides a method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection, comprising the following steps:

[0055] Step S10: acquiring a blood separation image in real time, and preprocessing the blood separation image to obtain a first blood separation image;

[0056] It should be noted that the blood separation image in the embodiment of the present application is a high-quality image (such as a high-resolution camera or a camera head) obtained by using a high-resolution camera or a camera head with an appropriate light source (such as transmitted light or reflected light) to capture the blood component separator tube during the separation process. Figure 2 As shown, the blood separation image at this time is the initial blood separation image, which is an image in which the components in the blood are separated into layers due to the centrifugal force of the blood component separator and the characteristics of the components in the blood, such as density. The use of the above-mentioned light source can avoid errors caused by the color or transparency of the blood; the above-mentioned preprocessing includes grayscale processing to obtain a grayscale image (i.e., the above-mentioned first blood separation image), which converts the layered effect of the above-mentioned initial blood separation image into a clear grayscale image to facilitate subsequent analysis.

[0057] Step S20: performing rough segmentation on the first blood separation image to obtain a second blood separation image;

[0058] It should be noted that blood is composed of multiple components, including red blood cells, white blood cells, platelets, plasma, etc. Among them, red blood cells account for the majority of the total blood volume and are mainly responsible for oxygen transportation. Since red blood cells contain a large amount of hemoglobin, their density is relatively high. Under the centrifugal force of the blood component separator, red blood cells will settle to the bottom of the test tube and form the bottom layer (i.e., the red blood cell layer); white blood cells are responsible for immune response and are fewer in number. Platelets participate in the coagulation process and also account for a smaller part of the total blood volume. The density of white blood cells and platelets is close, but both are much lower than that of red blood cells. They are in the blood. The centrifugal force of the blood separator usually forms a thin layer above the red blood cells. This layer is called the leukocyte layer or leukocyte-platelet layer, also known as the "lamellar layer." It contains most of the white blood cells and a small number of platelets. Plasma is the liquid component of blood. It contains water, proteins, ions, waste products, nutrients, hormones, etc., and is mainly responsible for transporting various substances. Plasma has the lowest density and is mainly composed of water and dissolved substances. Under the centrifugal force of the blood separator, plasma will float to the top, forming the top transparent liquid layer (the plasma layer).

[0059] The coarse segmentation in the above embodiment of the present application refers to the coarse segmentation of the top layer (i.e., the plasma layer) in the layered image of the first blood separation image, and obtaining a coarse segmentation image containing only the plasma layer or a small portion of the white blood cell-platelet layer and the plasma layer (e.g., Figure 3 As shown in FIG. 1 , i.e., the second blood separation image described above), in the subsequent operation process, the separation purity of the plasma layer is mainly evaluated to obtain the analysis result. The higher the purity, the better the separation effect, and the separation result is more helpful for clinical treatment, thereby improving the treatment effect.

[0060] The above-mentioned coarse segmentation is essentially a fast positioning method, the main purpose of which is to quickly locate and quickly extract the plasma layer area, providing a data basis for subsequent fine segmentation and purity evaluation. Its segmentation accuracy requirements are low, so the execution speed is fast and does not take up too much computing power; the above-mentioned fast positioning can be determined by preliminary positioning based on physical properties, area division based on color features, spatial position constraints or rough boundaries, which are all common knowledge of technicians in this field and will not be repeated in this application.

[0061] Step S30: performing fine segmentation on the second blood separation image to obtain a third blood separation image;

[0062] It should be noted that the second blood separation image is obtained by coarse segmentation of the first blood separation image. Therefore, it may still contain images of a partial white blood cell-platelet layer, which will cause a large evaluation error in the subsequent evaluation and analysis of the plasma layer (that is, the more images with partial white blood cell-platelet layer, the more likely the evaluation result will be affected as too low purity), thereby affecting subsequent treatment. Therefore, the second blood separation image needs to be further finely segmented to obtain an image with clear edges and containing only the plasma layer, such as Figure 4 shown.

[0063] Step S40: performing purity evaluation processing and analysis based on the third blood separation image to obtain a purity analysis result.

[0064] It should be noted that the purity evaluation process analysis is essentially an analysis of the amount of impurities contained in the plasma layer. The fewer impurities, the higher the purity of the plasma layer, and the better the further purity analysis result. The purity evaluation process analysis can adopt fuzzy comprehensive evaluation method, multi-criteria decision evaluation method, statistical evaluation method or expert decision evaluation method, etc., which will not be described in detail in this application.

[0065] The above-mentioned embodiment of the present application first uses a high-resolution camera or a camera to collect blood separation images in real time to ensure high image quality and clarity, and configures a suitable light source (such as transmitted light or reflected light) to eliminate errors caused by changes in blood color or transparency, thereby obtaining a more accurate image; and, the collected image is pre-processed (such as grayscale processing) to convert the original blood stratification effect into a clear grayscale image, which is convenient for subsequent segmentation and purity analysis; further, after the blood is centrifuged, a red blood cell layer, a white blood cell-platelet layer and a plasma layer are formed, and then the top layer of the blood, that is, the plasma layer, or the layer containing a small amount of white blood cells-platelets is quickly located and extracted through coarse segmentation. region, by quickly locating the plasma layer, simplifying the amount of calculation, and providing a data basis for subsequent fine segmentation; further, the second blood separation image may still contain part of the white blood cell-platelet layer, which will cause errors in the purity assessment of the plasma layer. Through fine segmentation, the white blood cell-platelet layer is removed from the plasma layer, and an image with clear edges and only containing the plasma layer is obtained; finally, by adopting fuzzy comprehensive evaluation method, multi-criteria decision evaluation method, statistical evaluation method or expert decision evaluation method, the impurity content of the plasma layer in the image is evaluated, and finally a purity analysis result is given, which provides a quantitative basis for the blood separation effect, helps doctors judge the quality of separated blood components, and further improves the treatment effect.

[0066] Specifically, if Figure 5 As shown, in step S30, fine segmentation is performed on the second blood separation image to obtain a third blood separation image, which includes the following steps:

[0067] Step S31: constructing a two-dimensional image matrix M(i, j) based on the grayscale values ​​of each pixel in the second blood separation image, wherein each element in the two-dimensional image matrix M(i, j) represents the grayscale value corresponding to each pixel position (i, j) in the second blood separation image; preprocessing the two-dimensional image matrix M(i, j) to obtain a first two-dimensional image matrix M'(i, j);

[0068] It should be noted that the above-mentioned preprocessing of the two-dimensional image matrix M(i, j) refers to performing noise filtering and edge enhancement processing on the two-dimensional image matrix M(i, j), wherein noise filtering refers to applying Gaussian filtering or median filtering to smooth the image and reduce the interference of random noise on the segmentation result; edge enhancement processing refers to using a gradient enhancement algorithm (such as the Sobel operator) to improve the clarity of the image edge, which is helpful for subsequent edge detection and segmentation;

[0069] Step S32: Perform feature extraction based on the first two-dimensional image matrix M'(i, j) to obtain multi-dimensional feature data F n (i,j);

[0070] The multidimensional feature data F n (i, j) includes grayscale distribution feature data F1(i, j), texture feature data F2(i, j) and edge feature data F3(i, j);

[0071] It should be noted that, in the multidimensional feature data in the above-mentioned embodiment of the present application, the grayscale distribution feature data F1(i, j) is obtained by analyzing the mean, variance, skewness and kurtosis of the grayscale values ​​in the region to capture statistical differences at different levels; the texture feature data F2(i, j) is obtained by using the gray level co-occurrence matrix (GLCM) to extract texture indicators such as contrast, correlation, energy and homogeneity to assist in distinguishing different tissue structures; the edge feature data F3(i, j) is obtained by identifying potential segmentation boundaries through edge detection results to provide spatial position information;

[0072] Step S33: Based on the multi-dimensional feature data F n (i, j) clustering each pixel point in the second blood separation image to obtain a plurality of clusters; constructing an image of the area to be processed corresponding to each cluster based on all pixel points in the bounding box of each cluster;

[0073] Step S34: Select the image of the area to be processed corresponding to the highest vertical position in the image of the area to be processed (the captured image of the test tube detection of the separator) as the head area image; based on the head area image, merge multiple images of the area to be processed through a region growing processing operation to obtain a third blood separation image.

[0074] It should be noted that the above-mentioned embodiment of the present application selects the image of the area to be processed with the highest vertical position among all the images of the area to be processed as the head area image. This is based on the stratification principle of blood components under the centrifugal force of the blood component separator. The highest layer is the plasma layer, and the third blood separation image in the embodiment of the present application refers to the image of the plasma layer. Therefore, the image of the area to be processed with the highest vertical position among all the images of the area to be processed must be a partial image of the plasma layer. Furthermore, the embodiment of the present application performs regional growth based on the head area image to gradually expand the range of the head area image, thereby finally obtaining the third blood separation image corresponding to the complete plasma layer.

[0075] Specifically, if Figure 6 As shown, in step S34, a plurality of to-be-processed region images are merged through a region growing operation based on the head region image to obtain a third blood separation image, which includes the following operation steps:

[0076] Step S341: extracting texture feature data from the head region image; and calculating the mean grayscale value μ of the pixels in the head region image. i' And the standard deviation of the gray value of pixels in the head area image μ n ; At the same time, the energy parameter factor of the head area image is obtained; Based on the energy parameter factor, the comprehensive energy quality score Q of the head area image is calculated by a preset energy function n ;

[0077] The energy parameter factors include gray uniformity U h , Regional Connectivity C c and shape regularity R s ;

[0078] The comprehensive energy quality score Q n The calculation method is:

[0079]

[0080] Where Q n is the comprehensive energy quality score; U h is the grayscale uniformity; C c is regional connectivity; R s is the shape regularity (rectangularity); is the ideal value of grayscale uniformity; I(i,j) is the grayscale value of the image at position (i,j); μ i' is the mean of the image grayscale values; μ n is the standard deviation of the image grayscale values;

[0081] It should be noted that, in the above embodiment of the present application, the gray uniformity Uh It is the consistency of the grayscale value in the region, reflecting whether the grayscale distribution in the region is uniform. The smaller the grayscale uniformity value is, the more consistent the grayscale in the region is and the higher the uniformity is. c Measures the degree of connectivity of a region in space, indicating whether the region is a complete, uncut connected block. Connectivity is usually evaluated by the connectivity of pixels within the region. Commonly used connectivity definitions are 4-connectivity (each pixel is connected to its four neighboring pixels above, below, left, and right) or 8-connectivity (each pixel is connected to its neighboring pixels above, below, left, and right, as well as four diagonal directions). The calculation method can be measured by the number of connected components within the region. If a region has only one connected component, the region is continuous; if there are multiple connected components, it means that the region is divided into several parts. If there are multiple connected components within the region, the continuity of the region is poor. Shape regularity R s It is a measure of the regularity of the shape of the region, and is usually used to evaluate whether the region is close to certain standard shapes (such as circle, rectangle, etc.). In the embodiment of the present application, each layer in the obtained blood separation image is displayed as a two-dimensional rectangle in the test tube. Therefore, the shape regularity R in the embodiment of the present application is s That is the rectangularity. The closer the rectangularity is to 1, the more regular the shape is.

[0082] Step S342: Based on the texture feature data of the current head region image and the grayscale value mean μ of the pixels in the head region image i' , the standard deviation of the gray value of pixels in the head area image μ n And the comprehensive energy quality score Q of the head area image n The image of the area to be processed is combined with the image of the head area to obtain a third blood separation image.

[0083] It should be noted that the above-mentioned embodiment of the present application first extracts important feature information related to blood separation (texture features, grayscale mean, standard deviation, energy parameter factors, etc.) from the image, quantifies the image quality, and provides a valuable data basis for subsequent image processing and analysis. It also provides a quantitative quality assessment basis for subsequent image merging, processing and analysis through a comprehensive energy quality score, which helps to ensure that the analysis of the blood separation image is more accurate; further, the image of the area to be processed is merged with the head area image that has been processed and the comprehensive energy quality score has been calculated, so as to obtain a complete third blood separation image (i.e., the image of the plasma layer mentioned above), providing favorable data support for clinical medical use.

[0084] Specifically, if Figure 7 As shown, in step S342, based on the texture feature data of the current head region image and the gray value mean μ of the pixel points in the head region image, i', the standard deviation of the gray value of pixels in the head area image μ n And the comprehensive energy quality score Q of the head area image n Merging the image of the area to be processed with the image of the head area to obtain a third blood separation image includes the following steps:

[0085] Step S3421: Start traversing from the image to be processed that is adjacent to the current head region image among all the image to be processed region images: calculate the texture similarity S between the image to be processed region image and the head region image t (R i' ); Based on the texture similarity, the mean grayscale value μ of the pixels in the head area image i' And the standard deviation of the gray value of pixels in the head area image μ n The grayscale value of each pixel in the image of the area to be processed is used to calculate the comprehensive similarity S(R i' );

[0086] The comprehensive similarity S(R i ) is calculated as:

[0087]

[0088] Where w g 、w t are weight coefficients; μ i' is the mean gray value of the pixels in the head area image μ i' ;μ n is the standard deviation of the grayscale value of pixels in the head area image μ n ; σ n is the standard deviation of the grayscale values ​​of pixels in the image of the current area to be processed; S t (R i’ ) is the texture similarity between the image of the area to be processed and the image of the head area;

[0089] Step S3422: Determine the comprehensive similarity S(R i' ) is greater than or equal to a preset minimum comprehensive similarity threshold; if so, the image of the area to be processed is determined as the image to be merged; if not, the image of the area to be processed is placed in the set of images of the area to be supplemented;

[0090] Step S3423: constructing a first head region image based on the image to be merged and the head region image; obtaining an energy parameter factor of the first head region image;

[0091] Step S3424: Calculate the comprehensive energy quality score Q' of the first head region image based on the energy parameter factor of the first head region image through a preset energy function. n ; Determine the comprehensive energy quality score Q' of the first head region image n Is it greater than or equal to the comprehensive energy quality score Q of the head area image? n If so, the first head region image is determined to be a new head region image, and the new head region image is returned to the above step S341 for continued traversal until all the region images to be processed are traversed, and the final new head region image is output as the third blood separation image to be determined;

[0092] Step S3425: performing image optimization processing based on the connected domain analysis of the third blood separation image to be determined and in combination with the set of images of the region to be supplemented, to obtain a third blood separation image.

[0093] It should be noted that the above-mentioned embodiment of the present application first traverses all images of the area to be processed and compares each image with the current head area image, and determines whether the image of the area to be processed is suitable for merging with the head area image by calculating the comprehensive similarity between the current image of the area to be processed and the head area image; further, by comparing the comprehensive similarity value with a preset minimum threshold, the image of the area to be processed and the head area image are merged or classified as a "area to be supplemented"; further, a new first head area image is constructed through the merging operation; and the comprehensive energy quality score of the image is calculated by obtaining the energy parameter factor of the merged first head area image; further, the current comprehensive energy quality score is compared with the quality of the original head area image to determine whether to adopt the new image, thereby ensuring that a head area image with higher quality is finally selected and ensuring that the quality of the merged image is not lower than the original image; further, by performing a connected domain analysis on the third blood separation image to be determined, and combining it with the set of images of the area to be supplemented, it is supplemented and optimized, the overall structure is improved, and the final blood separation image is generated.

[0094] During the specific implementation of the above-mentioned embodiment of the present application, since the blood separation image will not be completely separated, and the red blood cells will quickly fall to the bottom of the test tube due to the centrifugal force of the blood component separator due to their high density, and the white blood cell-platelet layer has a lower density than the red blood cells, so its falling speed is slower. There may be a small amount of partial white blood cell-platelet layer components inside the plasma layer. Therefore, there may be a cluster composed of pixel points of the white blood cell-platelet layer inside the cluster cluster obtained by clustering. That is to say, there may be a white blood cell-platelet layer image in the image of the area to be processed adjacent to the head area image. Since the white blood cell-platelet layer image will not be merged into the head area image during the growth process of the above-mentioned area, the third blood separation image finally obtained will be hollow inside, as shown in FIG. Figure 8 As shown, the purity evaluation of the subsequent plasma layer is affected. Therefore, in the above-mentioned process of region growing and merging the image of the area to be processed, it is also necessary to merge the small amount of impurities contained in the plasma layer range (i.e., the white blood cell-platelet layer image) into the head area image, so as to obtain a more realistic plasma layer image in the blood separation image corresponding to the current test tube. The specific operations are shown in steps S34251-S34253.

[0095] Specifically, if Figure 9 As shown, in step S3425, image optimization processing is performed based on the connected domain analysis of the third blood separation image to be determined and the image set of the area to be supplemented to obtain the third blood separation image, including the following operation steps:

[0096] Step S34251: Obtaining a hollow area through connected domain analysis based on the third blood separation image to be determined;

[0097] Step S34252: establishing an association relationship between the position of the hollow area image and the position of each of the to-be-processed area images in the to-be-supplemented area image set, and obtaining the to-be-supplemented area image corresponding to the hollow area;

[0098] Step S34253: Supplement the image of the area to be supplemented to the position corresponding to the hollow area to obtain a third blood separation image.

[0099] It should be noted that the above-mentioned embodiment of the present application first performs a connected domain analysis on the third blood separation image to be determined, identifies and extracts the hollow area in the image, and then matches and associates the position of the hollow area image with the position of each to-be-processed area image in the set of to-be-supplemented area images, finds the to-be-processed area image suitable for filling the hollow area, and each to-be-supplemented area image has its corresponding position. The association process ensures that the image can be accurately matched according to the spatial position, avoiding the problem of filling errors or position confusion; finally, the to-be-supplemented area image is accurately filled into the hollow area to obtain the optimized third blood separation image.

[0100] Specifically, in step S34251, a connected domain analysis is performed based on the third blood separation image to be determined to obtain a hollow area, which includes the following steps:

[0101] Step S342511: extracting the outer contour of the third blood separation image to be determined;

[0102] Step S342512: Based on the outer contour, a closed contour inside the outer contour is identified to obtain a hollow area.

[0103] It should be noted that the above-mentioned embodiment of the present application first extracts the outer contour from the third blood separation image to be determined through Canny edge detection, Sobel operator or contour detection algorithm. The outer contour refers to the outermost boundary line of the entire third blood separation image to avoid data loss or misoperation in subsequent processing; further, based on the extracted external contour, the closed contour inside the external contour is further identified. The closed contour refers to an area with complete and closed boundaries formed within the external contour, which usually represents a blank or missing part in the image; by identifying the closed contour, the areas in the image that have not been effectively separated or covered can be effectively located. These areas are hollow areas that need further supplementation or processing, which require further repair or filling. Therefore, identifying these areas is crucial for subsequent image processing.

[0104] In summary, the present invention proposes a method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection. First, a high-resolution camera or camera is used to collect blood separation images in real time to ensure high and clear image quality, and a suitable light source (such as transmitted light or reflected light) is configured to eliminate errors caused by changes in blood color or transparency, thereby obtaining a more accurate image; and the collected image is preprocessed (such as grayscale processing) to convert the original blood stratification effect into a clear grayscale image to facilitate subsequent segmentation and purity analysis; further, after the blood is centrifuged to form a red blood cell layer, a white blood cell-platelet layer and a plasma layer, and then the top layer of the blood, i.e., the plasma layer, or the layer containing a small amount of blood is quickly located and extracted through coarse segmentation. The area of ​​the white blood cell-platelet layer is separated, and the plasma layer is quickly located to simplify the calculation amount, providing a data basis for subsequent fine segmentation; furthermore, the second blood separation image may still contain part of the white blood cell-platelet layer, which will cause errors in the purity assessment of the plasma layer. Through clustering and combining with the region growing method for fine segmentation, the white blood cell-platelet layer is removed from the plasma layer, and an image with clear edges and only containing the plasma layer is obtained; finally, by adopting fuzzy comprehensive evaluation method, multi-criteria decision evaluation method, statistical evaluation method or expert decision evaluation method, the impurity content of the plasma layer in the image is evaluated, and finally a purity analysis result is given, which provides a quantitative basis for the blood separation effect, helps doctors judge the quality of the separated blood components, and further improves the treatment effect;

[0105] During further processing, important feature information related to blood separation (texture features, grayscale mean, standard deviation, energy parameter factors, etc.) is extracted from the image to quantify the image quality, providing a valuable data basis for subsequent image processing and analysis. The comprehensive energy quality score also provides a quantitative quality assessment basis for subsequent image merging, processing, and analysis, helping to ensure more accurate analysis of the blood separation image. Furthermore, the image of the area to be processed is merged with the head area image that has been processed and calculated with a comprehensive energy quality score, thereby obtaining a complete third blood separation image (i.e., the image of the plasma layer mentioned above), providing favorable data support for clinical medical applications.

[0106] In the further operation process, by traversing all the images of the area to be processed and comparing each image with the current head area image, the comprehensive similarity between the current image of the area to be processed and the head area image is calculated to determine whether the image of the area to be processed is suitable for merging with the head area image; further, by comparing the comprehensive similarity value with a preset minimum threshold, the image of the area to be processed and the head area image are merged or classified as a "area to be supplemented"; further, a new first head area image is constructed through the merging operation; and the comprehensive energy quality score of the image is calculated by obtaining the energy parameter factor of the merged first head area image; further, the current comprehensive energy quality score is compared with the quality of the original head area image to determine whether to adopt the new image, thereby ensuring that a higher quality head area image is finally selected and ensuring that the quality of the merged image is not lower than the original image; further, the hollow area is identified by performing a connected domain analysis on the third blood separation image to be determined, and combined with the set of images of the area to be supplemented, supplementation and optimization are performed to improve the overall structure and generate the final blood separation image.

[0107] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them. A person skilled in the art may modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection, characterized in that: The steps are as follows: Acquiring a blood separation image in real time, and preprocessing the blood separation image to obtain a first blood separation image; performing coarse segmentation on the first blood separation image to obtain a second blood separation image; performing fine segmentation on the second blood separation image to obtain a third blood separation image; performing purity assessment processing and analysis based on the third blood separation image to obtain a purity analysis result; The step of finely segmenting the second blood separation image to obtain a third blood separation image includes the following steps: constructing a two-dimensional image matrix M(i, j) based on the grayscale values ​​of each pixel in the second blood separation image, wherein each element in the two-dimensional image matrix M(i, j) represents the grayscale value corresponding to each pixel position (i, j) in the second blood separation image; preprocessing the two-dimensional image matrix M(i, j) to obtain a first two-dimensional image matrix M'(i, j); Perform feature extraction based on the first two-dimensional image matrix M'(i, j) to obtain multidimensional feature data F n (i,j); Based on the multidimensional feature data F n (i, j) clustering each pixel point in the second blood separation image to obtain a plurality of clusters; constructing an image of the area to be processed corresponding to each cluster based on all pixel points in the bounding box of each cluster; The image of the area to be processed corresponding to the highest vertical position among the images of the area to be processed is selected as the head area image; and a plurality of images of the area to be processed are merged through a region growing operation based on the head area image to obtain a third blood separation image.

2. The method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection according to claim 1, characterized in that: The multidimensional feature data F n (i, j) includes grayscale distribution feature data F1 (i, j), texture feature data F2 (i, j) and edge feature data F3 (i, j).

3. The method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection according to claim 2, characterized in that: The step of merging a plurality of to-be-processed region images through a region growing operation based on the head region image to obtain a third blood separation image comprises the following steps: Extract the texture feature data in the head area image; and calculate the gray value mean μ of the pixel points in the head area image i' And the standard deviation of the gray value of pixels in the head area image μ n ; At the same time, obtaining the energy parameter factor of the head area image; The comprehensive energy quality score Q of the head region image is calculated based on the energy parameter factor through a preset energy function. n ; Based on the texture feature data of the current head area image and the grayscale value mean μ of the pixels in the head area image i' , the standard deviation of the gray value of pixels in the head area image μ n And the comprehensive energy quality score Q of the head area image n The image of the area to be processed is combined with the image of the head area to obtain a third blood separation image.

4. The method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection according to claim 3, characterized in that: The energy parameter factors include gray uniformity U h , Regional Connectivity C c and shape regularity R s ; The comprehensive energy quality score Q n The calculation method is: Where Q n is the comprehensive energy quality score; U h is the grayscale uniformity; C c is regional connectivity; R s is the shape regularity; is the ideal value of grayscale uniformity; I(i,j) is the grayscale value of the image at position (i,j); μ i' is the mean of the image grayscale values; μ n is the standard deviation of the image grayscale values.

5. The method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection according to claim 4, characterized in that: The texture feature data based on the current head area image and the gray value mean μ of the pixels in the head area image i' , the standard deviation of the gray value of pixels in the head area image μ n And the comprehensive energy quality score Q of the head area image n Merging the image of the area to be processed with the image of the head area to obtain a third blood separation image includes the following steps: Start traversing from the image to be processed that is adjacent to the current head region image among all the image to be processed region images: calculate the texture similarity S between the image to be processed region image and the head region image t (R i' ); Based on the texture similarity, the mean grayscale value μ of the pixels in the head area image i' And the standard deviation of the gray value of pixels in the head area image μ n The grayscale value of each pixel in the image of the area to be processed is used to calculate the comprehensive similarity S(R i' ); Determine the comprehensive similarity S(R i' ) is greater than or equal to the preset minimum threshold of comprehensive similarity; If yes, the image of the region to be processed is determined as the image to be merged; if no, the image of the region to be processed is placed in the set of images of the region to be supplemented; constructing a first head region image based on the image to be merged and the head region image; and obtaining an energy parameter factor of the first head region image; Based on the energy parameter factor of the first head region image, a comprehensive energy quality score Q' of the first head region image is calculated by a preset energy function. n ; Determine the comprehensive energy quality score Q' of the first head region image n Is it greater than or equal to the comprehensive energy quality score Q of the head area image? n If so, the first head region image is determined to be a new head region image, and the new head region image is returned to the above operation to continue traversing until all the region images to be processed are traversed, and the final new head region image is output as the third blood separation image to be determined; Image optimization processing is performed based on the connected domain analysis of the third blood separation image to be determined and the image set of the area to be supplemented to obtain the third blood separation image.

6. The method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection according to claim 5, characterized in that: The comprehensive similarity S(R i' ) is calculated as: Where w g 、w t are weight coefficients; μ i' is the mean gray value of pixels in the head area image; μ n is the standard deviation of the grayscale value of pixels in the head area image; σ n is the standard deviation of the grayscale values ​​of pixels in the image of the current area to be processed; S t (R i' ) is the texture similarity between the image of the area to be processed and the image of the head area.

7. The method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection according to claim 6, characterized in that: The method of performing image optimization processing based on the connected domain analysis of the third blood separation image to be determined and the image set of the region to be supplemented to obtain the third blood separation image includes the following steps: Obtaining a hollow area through connected domain analysis based on the third blood separation image to be determined; Establishing an association relationship based on the position of the hollow area image and the position of each of the to-be-processed area images in the to-be-supplemented area image set, obtaining the to-be-supplemented area image corresponding to the hollow area; The image of the area to be supplemented is supplemented to a position corresponding to the hollow area to obtain a third blood separation image.

8. The method for evaluating the chromatographic effect of blood-separated plasma samples based on image detection according to claim 7, characterized in that: The method of obtaining the hollow area by connected domain analysis based on the third blood separation image to be determined includes the following steps: extracting an outer contour of the third blood separation image to be determined; A closed contour inside the outer contour is identified based on the outer contour to obtain a hollow area.

9. A storage medium, characterized in that: The storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for evaluating the chromatographic effect of a blood-separated plasma sample based on image detection as described in any one of claims 1 to 8.

Images