Feature extraction method and system based on platelet ultrastructure imaging

Abstract

A feature extraction method and system based on platelet ultrastructure imaging. The feature extraction method comprises: acquiring stable platelets having retained activity, and performing anticoagulation treatment on the platelets; performing antibody labeling on the platelets so as to prepare a sample for platelet imaging; imaging the sample so as to obtain a platelet ultrastructure picture; and segmenting the platelet ultrastructure picture to obtain individual-platelet pictures, acquiring diameter information of the corresponding platelets, and acquiring in the individual-platelet pictures the numbers of dense granules and α granules, granule area information, subcellular distribution information and α granule subcellular distribution features of activated platelets, so as to achieve feature extraction of platelet ultrastructures. The feature extraction method and system based on platelet ultrastructure imaging can acquire contour features of platelets, also extract the subcellular structural features of α granules and dense granules of platelets to serve as biomarkers, thus achieving important significance for the research and testing of physiological functions of platelets.

Description

A feature extraction method and system based on platelet ultrastructure imaging Technical Field

[0001] The present invention belongs to the field of biotechnology, and more specifically, relates to a feature extraction method and system based on platelet ultrastructure imaging. Background Art

[0002] Acute ischemic stroke, also known as acute cerebral infarction, is most commonly caused by activated platelets forming thrombi that occlude blood vessels, triggering ischemia and hypoxia in the resulting brain tissue, ultimately leading to irreversible damage. Digital subtraction angiography (DSA) is considered the gold standard for imaging collateral vascular status. However, its invasive nature limits its application. Conversely, non-invasive CT angiography makes it a good option for assessing collateral circulation in acute stroke patients. However, there is a lack of non-contrast CT or magnetic resonance imaging (MRI)-based methods for assessing collateral status, necessitating direct comparison with DSA. Furthermore, multiple neuroimaging modalities are often limited to major stroke centers and unavailable in many parts of the world, making it impossible to track dynamic changes in collateral circulation during long-term follow-up. Therefore, a simple, rapid, and sensitive parallel assay is crucial. Flow cytometry and enzyme-linked immunosorbent assay can be used to quantify platelet activation markers, but their disadvantages are that they are time-consuming and require specialized personnel. Light transmission aggregometry is the gold standard for platelet function testing, but its sensitivity is low and equipment is not widely available. PFA-100 / 200 measures the occlusion time of platelets at a constant shear rate, and its limitations are unrelated to thrombus. Thromboelastography is commonly used in clinical practice to evaluate platelet aggregation and fibrin polymerization. However, thromboelastography requires a baseline level as a reference before treatment. Currently, there are still no particularly good measures for the prevention and early detection of acute ischemic stroke. A recent study showed that platelets can accelerate the development of atherosclerotic plaques. This means that platelets have begun to function before the plaque ruptures, and structural and functional changes are likely to have occurred before the onset of acute ischemic stroke. Therefore, further research on the ultrastructure of platelets in patients with acute ischemic stroke will provide more information on the occurrence and development of the disease, which can greatly help the early detection, timely diagnosis and reasonable treatment of acute ischemic stroke in the future.

[0003] Platelets are small, sensitive cells that circulate in the bloodstream and play a crucial role in biological responses, including inflammation, immune responses, defense against infection, arteriosclerosis, cancer metastasis, and growth. Excessive platelet activation and aggregation can form thrombi, leading to vascular occlusion and even acute ischemic stroke or acute myocardial infarction. Due to their small size and ease of activation, studying their ultrastructure has been challenging. Platelets in acute ischemic stroke have also been considered to be abnormally activated. Although platelets are only 2 to 5 μm in diameter, they contain numerous subcellular structures, including microtubules, microfilaments, mitochondria, dense granules, and α-granules. Dense granules and α-granules are the organelles that are most prominent during platelet function.

[0004] Observations and studies focusing on changes in platelet ultrastructure in patients with acute cerebral infarction, a crucial component of the pathogenesis, are scarce. Currently, the morphology of activated platelets is primarily inferred through observation under conventional optical microscopy of their division into a central granular region and a peripheral clear zone, without further investigation of their ultrastructure. Electron microscopy allows for monitoring subcellular structural information in platelets. Some studies have used transmission electron microscopy to investigate changes in the number of dense granules and α-granules in platelets, but these studies are unable to accurately assess the number of granules within a single platelet. In summary, methods for analyzing and extracting subcellular structural information from platelets are currently lacking.

[0005] Summary of the Invention

[0006] To address the above-mentioned deficiencies or improvement needs of the prior art, the present invention provides a feature extraction method and system based on platelet ultrastructural imaging. This method uses a whole blood specimen and treats the specimen with a secondary antibody to obtain optical super-resolution imaging of the subcellular ultrastructure of platelets in human blood. The image is then segmented using image segmentation processing. Finally, the imaging features are extracted and statistically analyzed to obtain the subcellular structural characteristics of platelet α-granules and dense granules. The present invention can obtain platelet contour features and simultaneously extract the subcellular structural characteristics of platelet α-granules and dense granules as biomarkers, which is of great significance for the study and detection of platelet physiological functions.

[0007] To achieve the above object, according to one aspect of the present invention, a feature extraction method based on platelet ultrastructure imaging is proposed, comprising the following steps:

[0008] S1 obtains stable platelets that retain activity and performs anticoagulation on the platelets;

[0009] S2 labels the platelets that have been anticoagulated and are in a resting state with antibodies to prepare a sample for platelet imaging;

[0010] S3 imaging the sample to obtain an image of the platelet ultrastructure;

[0011] S4 segments the platelet ultrastructure image to obtain a single platelet image, and obtains the diameter information of the corresponding platelet, obtains the number of dense granules and α granules in the single platelet image, granule area information, subcellular distribution characteristics, and subcellular distribution characteristics of α granules after platelet activation, thereby realizing feature extraction of the platelet ultrastructure.

[0012] As further preferred, step S1 includes the following steps:

[0013] S11 collects EDTA anticoagulated blood samples and centrifuges them to obtain stable platelet components that retain activity;

[0014] S12 uses Tyrode buffer containing ACD anticoagulant to preserve platelets, wherein the volume ratio of the ACD anticoagulant to the Tyrode buffer is 1:4 to 1:10.

[0015] As further preferred, in step S12, the Tyrode buffer comprises: NaCl, KCl, MgCl, NaH2PO4, NaHCO3, glucose, and HEPES buffer, and the pH value of the Tyrode buffer is 7.4.

[0016] As further preferred, step S2 includes the following steps:

[0017] S21: dilute the obtained platelet suspension and spread it into a culture dish, place it in a 37°C, 5% CO2 environment, and restore the platelet living cells to a resting state;

[0018] S22 uses fixative to fix platelets in the culture dish;

[0019] S23 uses anti-dense granule antibodies to label platelet dense granules and anti-α granule antibodies to label platelet α granules to prepare samples for platelet imaging.

[0020] As further preferred, the fixing solution is a phosphate solution, which includes 4% paraformaldehyde, 0.15% glutaraldehyde and 0.2% Triton-X100 by volume, and the pH value of the phosphate solution is 7.4.

[0021] As a further preferred embodiment, in step S24, after the platelets are labeled with anti-dense granule and anti-α granule antibodies, the culture dish is washed with buffer at 4°C, wherein the buffer is prepared with 1% goat serum and 0.05% Triton-X100 in phosphate buffer. After washing, the fluorescent dense granules and fluorescent α granules are incubated. After staining, they are fixed with 4-8% paraformaldehyde aqueous solution at room temperature, centrifuged and the supernatant is removed, washed with phosphate-buffered saline, and finally an appropriate amount of phosphate buffer is added to obtain a fixed platelet suspension.

[0022] As further preferred, step S4 includes the following steps:

[0023] S41: Segmenting the platelet ultrastructure image to obtain a single platelet image, extracting the platelet diameter distribution pattern in the single platelet image to obtain diameter information of the corresponding platelet;

[0024] S42 acquires a super-resolution image of dense granules and α-granules of a single platelet labeled with antibodies, and segments the super-resolution image according to corresponding fluorescence channel values ​​of the dense granules and α-granules to obtain a dense granule image and an α-granule image;

[0025] S43 performs quantitative statistics, particle area, particle area ratio, and subcellular distribution analysis on dense particles in dense particle images, and performs quantitative statistics, particle area, particle area ratio, and subcellular distribution analysis on α particles in α particle images.

[0026] According to another aspect of the present invention, a feature extraction system based on platelet ultrastructure imaging is provided, comprising:

[0027] A blood sample processing module is used to obtain stable platelets that retain activity, perform anticoagulation on the platelets, and label the anticoagulated platelets in a resting state with antibodies to prepare samples for platelet imaging;

[0028] The platelet super-resolution image feature extraction module is used to image the sample to obtain a platelet ultrastructure image, and then segment the platelet ultrastructure image to obtain a single platelet image, and obtain the diameter information of the corresponding platelet, and obtain the number of dense granules and α granules in the single platelet image, granule area information, subcellular distribution information, and subcellular distribution characteristics of α granules after platelet activation, thereby realizing feature extraction of the platelet ultrastructure.

[0029] As a further preferred method, obtaining the number of dense granules and α-granules in a single platelet image and the subcellular distribution characteristics of α-granules after platelet activation includes:

[0030] Acquire a super-resolution image of dense granules and α-granules labeled with antibodies of a single platelet, and segment the super-resolution image according to the RGB color channel values ​​of the dense granules and α-granules to obtain a dense granule image and an α-granule image;

[0031] The dense particles in the dense particle image are analyzed for particle number, particle area information, and subcellular distribution information; the α particles in the α particle image are analyzed for particle number, particle area information, and subcellular distribution information.

[0032] As a further preferred embodiment, the method further includes a comparison module, which is used to compare and analyze the features extracted by the platelet super-resolution image feature extraction module with a standard feature threshold.

[0033] In general, the above technical solutions conceived by the present invention have the following technical advantages compared with the existing technology:

[0034] 1. The present invention uses whole blood specimens to obtain optical super-resolution imaging of the subcellular ultrastructure of platelets in human blood. By extracting and statistically analyzing the imaging features, the subcellular structural characteristics of platelet α-granules and dense granules are obtained, which can be used as biomarkers.

[0035] 2. The present invention's subcellular structural characteristic parameters for platelet α-granules and dense granules include: the average number of granules per platelet (using anti-dense granule antibodies to label platelet dense granules and anti-α-granule antibodies to label α-granules), the average area of ​​a single granule, the total area of ​​granules per platelet, the percentage of granule area within the imaging field of view, and the subcellular distribution characteristics of α-granules after platelet activation. This addresses the technical problem of existing platelet detection technologies lacking specific platelet information recognition and limited information available for detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] FIG1 is a flow chart of a feature extraction method based on platelet ultrastructure imaging according to a preferred embodiment of the present invention;

[0037] FIG2 is an ultrastructural image of dense granules and α-granules of human platelets based on optical super-resolution imaging;

[0038] Figure 3 shows the imaging and diameter comparison of human platelets (Hy is the young healthy group, Ho is the same age (elderly healthy group), P acute ischemic stroke;

[0039] Figure 4 shows the number and subcellular distribution characteristics of dense granules in human platelets based on optical super-resolution imaging (Hy is the young healthy group, Ho is the same age group (elderly healthy group), P is acute ischemic stroke. *P<0.05, **P<0.01, ***P<0.01, ****P<0.001; scale bar 3mm (B, D), 1mm (A, C));

[0040] Figure 5 shows the number and subcellular distribution characteristics of human platelet α particles based on optical super-resolution imaging (Hy is a young healthy group, Ho is a same-age group (elderly healthy group), P is an acute ischemic stroke group. *P<0.05, **P<0.01, ***P<0.01, ****P<0.001; scale bar 3mm (B, D), 1mm (A, C));

[0041] Figure 6 shows the subcellular distribution characteristics of platelet a-granules after activation (scale bar 1 mm). DETAILED DESCRIPTION

[0042] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0043] In human blood samples, platelet dense granules are 50nm and α granules are only 200-400nm. The limiting resolution of traditional optical imaging is 200nm, which limits the study of platelet subcellular structure. Optical super-resolution microscopy technology has achieved a breakthrough in the limiting resolution of optical imaging and can obtain subcellular ultrastructural information below 200nm. In addition, the sample preparation process of optical super-resolution microscopy is more stable and convenient than that of electron microscopy. Antibodies and probes suitable for standard fluorescence microscopes can also be used for optical super-resolution imaging. Optical super-resolution microscopy technology itself can realize imaging of tiny platelets, identify platelet subcellular ultrastructure, and obtain high-throughput quantitative optical imaging data. Therefore, the present invention uses immunofluorescence staining to label platelet dense and α granules, so as to obtain platelet ultrastructural information and quantitatively analyze platelet ultrastructure, providing a scientific basis for medical detection methods.

[0044] As shown in FIG1 , an embodiment of the present invention provides a feature extraction method based on platelet ultrastructure imaging, comprising the following steps:

[0045] Step 1: Obtain stable platelets with retained activity and anticoagulate them. Specifically, collect an EDTA-anticoagulated blood sample, centrifuge it, and obtain the stable platelet component with retained activity. The platelets are then preserved in Tyrode's buffer containing an ACD anticoagulant, where the volume ratio of the ACD anticoagulant to the Tyrode's buffer is 1:4 to 1:10.

[0046] More specifically, in one embodiment of the present invention, a 2-7 mL EDTA-anticoagulated blood sample is collected. Within 6 hours of collection, the sample is centrifuged (100-500 g, 5-12 min, with both the centrifuge speed at 1) to obtain a stable platelet fraction with retained activity. The platelets are preserved in Tyrode's buffer containing the anticoagulant ACD. The buffer contains NaCl, KCl, MgCl, NaH2PO4, NaHCO3, glucose, and EPES. The pH of the solution is 7.4, and the volume ratio of ACD to Tyrode's buffer is 1:4 to 1:10.

[0047] Step 2: Anticoagulation-treated and resting platelets are labeled with antibodies to prepare samples for platelet imaging. Specifically, the obtained platelet suspension is diluted and spread on a culture dish, and placed in a 37°C, 5% CO2 environment for 2 to 6 hours to restore the resting state of live platelet cells. After the platelets settle to the bottom of the dish and recover, immunofluorescence staining is performed. A phosphate solution containing 4% paraformaldehyde, 0.15% glutaraldehyde and 0.2% TritonX-100 by volume and a pH of 7.4 is used as a fixative for fixation. After washing, punching and blocking, anti-dense granule antibodies are used to label platelet dense granules, anti-α granule antibodies are used to label α granules, and the corresponding secondary antibodies are labeled to obtain samples to be imaged.

[0048] In the present invention, platelet immunostaining specifically comprises diluting the obtained platelet suspension, plating it onto a culture dish, and incubating it at 37°C, 5% CO₂ for 2-6 hours to allow the platelets to settle to a resting state. After the platelets have settled to the bottom of the dish and recovered, immunostaining is performed. For platelet specimens requiring activation, the anticoagulant buffer is replaced and discarded, and Tyrode's buffer containing 0.5-5 U of thrombin is added for activation for 1-2 hours.

[0049] Fixation: Use a phosphate-buffered saline solution containing 3-4% paraformaldehyde, 0.1-0.2% glutaraldehyde, and 0.2% Triton-X in pH 7.4 as the fixative. Perforation: Incubate the plate with perforation solution (PBS buffer containing 0.2-2% Triton-X100 by volume) at room temperature for 10 minutes. Blocking: Use blocking solution (PBS buffer containing 10% normal goat serum, 0.05% Triton-X100, and 1% BSA) at room temperature for 1 hour. Label platelet dense granules with an anti-dense granule antibody and α-granule antibodies for α-granules to obtain the sample to be imaged. All antibodies are diluted in buffer. Remove the antibody solution and wash the dish five times at 4°C with a buffer consisting of 1% normal goat serum and 0.05% Triton-X100 in phosphate-buffered saline. After washing, incubate with a fluorescent secondary antibody. The secondary antibody is diluted in buffer. After staining, fix the platelets with a 4-8% aqueous paraformaldehyde solution at room temperature for 30 minutes. After centrifugation, remove the supernatant, wash three times with phosphate-buffered saline, and finally add an appropriate amount of phosphate buffer to obtain a fixed platelet suspension. The resulting sample was imaged under an optical super-resolution microscope to obtain images of the platelet ultrastructure. For optimal imaging, immersion oil with a refractive index of 1.518 was used for oil-immersion microscopy.

[0050] Step three: imaging the sample to obtain an image of the platelet ultrastructure.

[0051] Step 4, segmenting the platelet ultrastructure image to obtain a single platelet image, and obtaining the diameter information of the corresponding platelet, obtaining the number of dense granules and α granules in the single platelet image, granule area information, subcellular distribution characteristics, and subcellular distribution characteristics of α granules after platelet activation, thereby realizing feature extraction of the platelet ultrastructure. In this step, the fluorescence image information is extracted, the image is binarized, and the single platelet is segmented to obtain the average platelet diameter information. The dense granules and α granules in each platelet are segmented to obtain the average number of granules in each platelet, the average area of ​​a single granule, the total area of ​​granules in each platelet, the percentage of the granule area under the imaging field of view, and the subcellular distribution characteristics of α granules after platelet activation. The subcellular structural characteristics of platelet α granules and dense granules are significantly different between patients with acute ischemic stroke and healthy people of the same age, and can be used as biomarkers.

[0052] In one embodiment of the present invention, as shown in Figures 2 and 4, the specimen is imaged using an optical super-resolution microscope, and an image of the ultrastructural information of the platelet subcellular organelles is reconstructed and finally statistically analyzed. The microscope is routinely calibrated with a 100nm fluorescent ball to the lateral and axial limits of the resolution of the computer image. Incomplete, generated pseudopodia and platelets with a diameter of less than 2μm are removed. The fluorescence image information is extracted, the image is binarized, and a single platelet is segmented, and the maximum transverse diameter of the platelet is selected as the platelet diameter. A single platelet is circled from the bright field image corresponding to each fluorescence image, and the super-resolution image is binarized, thresholded, and segmented to obtain the information of individual particles within a single platelet. The number of particles in each platelet is obtained by first analyzing the average number of particles in a single platelet in a super-resolution image under a field of view, and then analyzing the average number obtained from all super-resolution images of the specimen; the average area of ​​a single particle is obtained by first analyzing the average area of ​​a single particle in an image under a field of view, and then analyzing the average area obtained from all super-resolution images of the specimen; the total area of ​​particles in each platelet is obtained by first analyzing the total area of ​​particles in a single platelet in an image under a field of view, and then analyzing the average area obtained from all super-resolution images of the specimen; the percentage of particle area in a field of view is obtained by first analyzing the percentage of area occupied by particles in a field of view, and then analyzing the percentage of particle area obtained from all super-resolution images of the specimen.

[0053] The subcellular distribution of α-granules after platelet activation was analyzed. Platelets were activated with thrombin (TH) for 5 minutes to 2 hours. Upon activation, VWF aggregates were classified into three types: small aggregates, reticular aggregates, and agglomerates. Small aggregates, reticular aggregates, and honeycomb aggregates were defined as cellular aggregates, and agglomerates were defined as parenchymal aggregates.

[0054] In this case, the number of platelet α-granules and dense granules and their subcellular structural characteristics can be used as the basis for clinical judgment. For example, when the number of platelet α-granules and dense granules follows a normal distribution, an unpaired two-tailed Student's t-test should be used. If the data do not follow a normal distribution, a two-tailed Mann-Whitney test or Kolmogorov-Smirnov test should be used. A P value < 0.05 is considered statistically significant.

[0055] As shown in FIG2 and FIG3, in the present invention, the image processing process is specifically as follows:

[0056] (41) The platelet ultrastructure image is segmented to obtain a single platelet image, and the platelet diameter distribution pattern in the single platelet image is extracted to obtain the diameter information and area information of the corresponding platelet. Specifically, first, the outline of the platelet in the platelet ultrastructure image is extracted to obtain the platelet outline information, and the platelet ultrastructure image is segmented based on the platelet outline information to obtain a single platelet image. At the same time, the platelet diameter distribution pattern in the single platelet image is extracted based on the platelet outline information, and the diameter information, particle area information, particle area relative proportion information, and subcellular distribution information of the corresponding platelet are further obtained. In the present invention, the number of platelets in a sample is also counted by counting the number of platelets in each field of view. Furthermore, in a specific embodiment of the present invention, a neural network model, such as a convolutional neural network model or a support vector machine model, is used to input platelets from an existing image into the convolutional neural network model, train the convolutional neural network model, and thereby obtain a trained convolutional neural network model for platelet contour extraction. This model is then used to extract the contours of platelets from an image of platelet ultrastructure. An image segmentation threshold is then set based on the extracted contours, and the image is segmented according to the segmentation threshold.

[0057] (42) Obtain a super-resolution image of the dense granules and α-granules of a single platelet labeled with antibodies, and segment the super-resolution image according to the RGB color channel values ​​of the dense granules and α-granules to obtain a dense granule image and an α-granule image. In the present invention, the dense granules and α-granules of the platelets are labeled with different antibodies and stained differently. For example, dense granules labeled with CD63 antibodies appear fluorescent green in the image, and α-granules labeled with VWF antibodies appear red in the image. Based on the different RGB color channel values ​​of the two colors, the super-resolution image is segmented to obtain a dense granule image and an α-granule image. More specifically, in the present invention, three sets of RGB color channel screening thresholds are set to achieve segmentation of the image background color, stained dense granules, and stained α-granules, and to eliminate the noise image, thereby obtaining the target image, i.e., the dense granule image and the α-granule image.

[0058] (43) As shown in Figures 5 and 6, the dense particles in the dense particle image were subjected to particle number statistics, particle area information analysis, and subcellular distribution analysis, while the α particles in the α particle image were subjected to particle number statistics, particle area information analysis, and subcellular distribution analysis.

[0059] Specifically, in one embodiment of the present invention, dense granule images of a large number of platelets are collected, and the images are used as input to a support vector machine model. The number and distribution position of the dense granules in the corresponding images are used as output. The support vector machine model is trained to obtain a first support vector machine model of dense granule number and position statistics. This model is then used to extract the number and position information of dense granules in the dense granule images, and then the position information of the dense granules is fitted and analyzed according to the contours of the corresponding platelets to obtain the distribution status of the dense granules in the platelets.

[0060] Specifically, in one embodiment of the present invention, a large number of platelet α particle images are collected, and the images are used as input to a support vector machine model. The number and distribution position of the α particles in the corresponding images are used as output. The support vector machine model is trained to obtain a second support vector machine model of the number and position statistics of the α particles. This model is then used to extract the number and position information of the α particles in the α particle images, and then the position information of the α particles is fitted and analyzed according to the contours of the corresponding platelets to obtain the distribution status of the α particles in the platelets.

[0061] Based on the distribution of dense granules and α granules in platelets, the average number obtained from all super-resolution images of the specimen was analyzed. The average area of ​​a single particle was obtained by first analyzing the average area of ​​a single particle in an image under a field of view, and then analyzing the average area obtained from all super-resolution images of the specimen. The total area of ​​particles in each platelet was obtained by first analyzing the total area of ​​particles in a single platelet in an image under a field of view, and then analyzing the average area obtained from all super-resolution images of the specimen. The percentage of particle area in a field of view was obtained by first analyzing the percentage of the area occupied by particles in a field of view, and then analyzing the percentage of the particle area obtained from all super-resolution images of the specimen.

[0062] In the present invention, it is also used to obtain standard platelet image features, particle-related information and distribution conditions of dense granules in platelets, and particle-related information and distribution conditions of α granules in platelets, and the above features are used as comparison features to compare the features extracted from the above sample information.

[0063] According to another aspect of the present invention, a feature extraction system based on platelet ultrastructure imaging is provided. The system is used to implement the method involved in any of the above embodiments, specifically comprising:

[0064] A blood sample processing module is used to obtain stable platelets that retain activity, perform anticoagulation on the platelets, and label the anticoagulated platelets in a resting state with antibodies to prepare samples for platelet imaging;

[0065] The platelet super-resolution image feature extraction module is used to image the sample to obtain a platelet ultrastructure image, and then segment the platelet ultrastructure image to obtain a single platelet image, and obtain the diameter information of the corresponding platelet, and obtain the number of dense granules and α granules in the single platelet image, granule area information, granule area ratio information, subcellular distribution characteristics, and subcellular distribution characteristics of α granules after platelet activation, thereby realizing feature extraction of the platelet ultrastructure.

[0066] In one embodiment of the system of the present invention, obtaining the number of dense granules and α-granules, granule area information, subcellular distribution characteristics, and subcellular distribution characteristics of α-granules after platelet activation in a single platelet image includes:

[0067] Acquire a super-resolution image of dense granules and α-granules labeled with antibodies of a single platelet, and segment the super-resolution image according to the RGB color channel values ​​of the dense granules and α-granules to obtain a dense granule image and an α-granule image;

[0068] The dense particles in the dense particle image are subjected to particle number statistics, particle area information analysis and distribution analysis, and the α particles in the α particle image are subjected to particle number statistics, particle area information analysis and distribution analysis.

[0069] In one embodiment of the system of the present invention, a comparison module is further included, which is used to compare and analyze the features extracted by the platelet super-resolution image feature extraction module with the standard feature threshold.

[0070] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A feature extraction method based on platelet ultrastructure imaging, characterized in that: The following steps are involved: S1 obtains stable platelets with retained activity and performs anticoagulation on the platelets; S2 performs antibody labeling on the platelets that have been anticoagulated and are in a resting state to prepare a sample for platelet imaging; S3 imaging the sample to obtain an image of the platelet ultrastructure; S4 segments the platelet ultrastructure image to obtain a single platelet image, and obtains the diameter information of the corresponding platelet, obtains the number of dense granules and α granules, granule area information, subcellular distribution characteristics, and subcellular distribution characteristics of α granules after platelet activation in the single platelet image, thereby realizing the feature extraction of the platelet ultrastructure.

2. A feature extraction method based on platelet ultrastructure imaging according to claim 1, characterized in that: Step S1 includes the following steps: S11 collects EDTA anticoagulated blood samples and obtains stable platelet components with retained activity after centrifugation; S12 uses Tyrode buffer containing ACD anticoagulant to preserve platelets, wherein the volume ratio of the ACD anticoagulant to the Tyrode buffer is 1:4 to 1:

10.

3. A feature extraction method based on platelet ultrastructure imaging according to claim 1, characterized in that: In step S12, the Tyrode buffer comprises: NaCl, KCl, MgCl, NaH2PO4, NaHCO3, glucose, and HEPES buffer, and the pH value of the Tyrode buffer is 7.

2.

4. A feature extraction method based on platelet ultrastructure imaging according to claim 1, characterized in that: Step S2 includes the following steps: S21 dilute the obtained platelet suspension and spread it into a culture dish, place it in a 37°C, 5% CO2 environment, and restore the platelet living cells to a resting state; S22 uses fixative to fix platelets in the culture dish; S23 uses anti-dense granule antibody to label platelet dense granules and anti-α granule antibody to label platelet α granules to prepare samples for platelet imaging.

5. A feature extraction method based on platelet ultrastructure imaging according to claim 4, characterized in that: The fixing solution is a phosphate solution, which includes 4% paraformaldehyde, 0.15% glutaraldehyde and 0.2% Triton-X100 in volume percentage, and the pH value of the phosphate solution is 7.

4.

6. A feature extraction method based on platelet ultrastructure imaging according to claim 4, characterized in that: In step S24, after the platelets are labeled with anti-dense granules and anti-α granules antibodies, the culture dish is washed with a buffer solution at 4°C, wherein the buffer solution is prepared with 1% goat serum and 0.05% Triton-X100 in phosphate buffer. After washing, the fluorescent dense granules and fluorescent α granules are incubated. After staining, they are fixed at room temperature with a 4-8% paraformaldehyde phosphate buffer solution, centrifuged and the supernatant is removed, washed with phosphate buffered saline, and finally an appropriate amount of phosphate buffer is added to obtain a fixed platelet suspension.

7. A feature extraction method based on platelet ultrastructure imaging according to claim 1, characterized in that: Step S4 includes the following steps: S41: segmenting the platelet ultrastructure image to obtain a single platelet image, extracting the platelet diameter distribution law in the single platelet image to obtain the diameter information and area information of the corresponding platelet; S42 acquires a super-resolution image of dense granules and α-granules antibody-labeled on a single platelet, and segments the super-resolution image according to corresponding fluorescence channel information of the dense granules and α-granules to acquire a dense granule image and an α-granule image; S43 performs particle number statistics, particle area information analysis, and subcellular distribution analysis on the dense particles in the dense particle image, and performs particle number statistics, particle area information analysis, and subcellular distribution analysis on the α particles in the α particle image.

8. A feature extraction system based on platelet ultrastructure imaging, characterized in that: include: A blood specimen processing module is used to obtain stable platelets that retain activity, perform anticoagulation on the platelets, and perform antibody labeling on the anticoagulated platelets that are in a resting state to prepare samples for platelet imaging; The platelet super-resolution image feature extraction module is used to image the sample to obtain a platelet ultrastructure image, and then segment the platelet ultrastructure image to obtain a single platelet image, and obtain the diameter information of the corresponding platelet, and obtain the number of dense granules and α granules in the single platelet image, granule area information, subcellular distribution characteristics, and subcellular distribution characteristics of α granules after platelet activation, thereby realizing the feature extraction of the platelet ultrastructure.

9. A feature extraction system based on platelet ultrastructure imaging according to claim 8, characterized in that: The method of obtaining the number of dense granules and α granules, granule area information, subcellular distribution characteristics, and subcellular distribution characteristics of α granules after platelet activation in a single platelet image includes: Acquire a super-resolution image of dense granules and α-granules of a single platelet labeled with antibodies, and segment the super-resolution image according to RGB color channel values ​​of the dense granules and α-granules to obtain a dense granule image and an α-granule image; The dense granules in the dense granule image were statistically analyzed for quantity and distribution, and the α granules in the α granule image were statistically analyzed for quantity, granule area, granule area ratio and subcellular distribution.

10. A feature extraction system based on platelet ultrastructure imaging according to claim 8, characterized in that: It also includes a comparison module, which is used to compare and analyze the features extracted by the platelet super-resolution image feature extraction module with the standard feature threshold.

Images