Platelet performance index evaluation method and device

Abstract

The invention discloses a platelet performance index evaluation method and device. The method comprises the following steps: collecting platelets to be detected through a platelet performance index evaluation method; acquiring cloud feature data of the to-be-detected platelets by using an image video device, wherein the cloud feature data comprises cloud feature image data and / or cloud feature video data; and evaluating platelet performance indexes by using the cloud feature data of the to-be-detected platelets. The platelet performance indexes can be accurately and rapidly detected at low cost, and the problems that a traditional platelet performance detection method is long in consumed time, high in cost, large in result error and the like are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of blood testing, and in particular to a method and device for evaluating platelet performance indicators. Background Art

[0002] Platelets are small pieces of cytoplasm formed by the degranulation of megakaryocytes. They play a vital role in the process of hemostasis through functions such as adhesion, aggregation, release and contraction. In clinical treatment, platelet transfusion is widely used to prevent or treat hemorrhagic diseases caused by trauma, surgery, leukemia, bone marrow suppression after radiotherapy or chemotherapy, aplastic anemia, infection, etc., as well as functional abnormalities caused by reduced platelet count. Timely transfusion of platelets can buy more treatment time for patients, thereby improving prognosis and even saving lives. However, in recent years, ineffective platelet transfusions have often occurred, seriously delaying the treatment of patients. After in-depth research, it was found that the quality defects of platelets transfused to patients are one of the main reasons. Therefore, it is particularly important to evaluate the performance of platelets before transfusion.

[0003] At present, clinical platelet performance testing mainly relies on traditional laboratory methods, such as platelet aggregation test, thromboelastogram, dynamic monitoring of coagulation and platelet function, etc. Although the existing platelet performance testing methods can meet some clinical needs to a certain extent, they still have the following problems:

[0004] (1) The results are not accurate enough. Traditional platelet performance testing requires multiple steps. If there is a problem in one of the steps, it will affect the accuracy of the entire result and fail to accurately reflect the patient's true coagulation status, thereby misleading subsequent treatment. Taking thromboelastography as an example, the reaction cup must be manually loaded first, and then the blood sample, kaolin and calcium ions are added separately. Then the reaction cup is pushed to the detection part and the start button is pressed to start the timing. The equipment should be absolutely still during the entire detection process. Slight vibrations will affect the accuracy of the results or cause the cup to fall and require re-testing.

[0005] (2) It takes a long time. Traditional platelet performance testing methods usually take more than an hour for a sample from pre-test preparation to the issuance of test results, which seriously delays patient treatment, especially for patients with acute massive bleeding who need immediate platelet transfusion, and emergency surgery, acute coronary syndrome and other situations that require rapid evaluation of platelet function. For example, when a hypocoagulable patient receives procoagulant therapy, it is very likely that hypocoagulable will turn into hypercoagulable in a short period of time, resulting in coagulation disorder. Even the latest traditional test results can only reflect the patient's hypocoagulable state one hour ago, and cannot detect hypercoagulable conditions in real time, which seriously misleads subsequent treatment. It is not only not conducive to the patient's recovery, but even worse, it may cause the patient to lose his life.

[0006] (3) Excessive cost. Traditional testing requires professional personnel, equipment, reagents, consumables, etc., and the cost is high when used.

[0007] In summary, traditional platelet performance testing results are not accurate enough, time-consuming, and costly, and cannot meet the actual needs of accurately, quickly, and cheaply evaluating the performance of platelets before transfusion. Summary of the invention

[0008] The present invention overcomes the deficiencies of the prior art and provides a platelet performance index evaluation method and device, which can accurately, quickly and inexpensively detect platelet performance indicators, and solve the problems of traditional platelet performance test results being inaccurate, time-consuming and costly.

[0009] According to one aspect of the present invention, a method for evaluating platelet performance indicators is proposed, the method comprising:

[0010] Collect platelets for testing;

[0011] Using an image and video device to collect cloud-like characteristic data of the platelets to be detected, wherein the cloud-like characteristic data includes cloud-like characteristic image data and / or cloud-like characteristic video data;

[0012] The cloud-like characteristic data of the platelets to be tested are used to evaluate platelet performance indicators.

[0013] In a possible implementation, the evaluating the platelet performance index by using the cloud-like characteristic data of the platelets to be detected includes:

[0014] Performing frequency domain analysis and gradient analysis on each frame of the cloud-like characteristic data of the platelets to be detected, and obtaining a comprehensive evaluation index of the cloud-like characteristic data of the platelets to be detected;

[0015] Recording the duration of the cloud-like characteristic of the platelets to be detected, and obtaining the duration of the cloud-like characteristic data of the platelets to be detected;

[0016] Obtaining significant features of the cloud-like characteristic data of the platelets to be detected according to the comprehensive evaluation index and duration of the cloud-like characteristic data of the platelets to be detected;

[0017] The platelet performance index is evaluated based on the correspondence between the significant features of the cloud-like characteristic data of the platelets to be detected and the platelet performance index.

[0018] In a possible implementation, the comprehensive evaluation index of the cloud-like characteristic data of the platelets to be detected includes spectrum energy ratio, gradient amplitude mean and fringe direction score.

[0019] In a possible implementation, the evaluating of the platelet performance index according to the correspondence between the significant features of the cloud-like feature data of the platelets to be detected and the platelet performance index includes:

[0020] Determining the most significant feature of the cloud-like characteristic data of the platelets to be detected;

[0021] The quality of the platelet performance index is evaluated based on the correspondence between the most significant feature of the cloud-like characteristic data of the platelets to be detected and the level of the platelet performance index.

[0022] In a possible implementation, the platelet performance indicators are divided into different levels according to the platelet coagulation function.

[0023] In a possible implementation, the evaluating the platelet performance index by using the cloud-like characteristic data of the platelets to be detected further includes:

[0024] The cloud-like characteristic data of the platelets to be detected are input into the trained platelet performance index evaluation model PEM to obtain platelet performance index evaluation data.

[0025] In a possible implementation, the platelet performance index evaluation model PEM is trained based on platelet performance index data and platelet cloud feature data.

[0026] In a possible implementation, the platelet performance index evaluation model PEM is trained based on platelet performance index data and platelet cloud feature data, including:

[0027] Collect platelets for training;

[0028] Using an image and video device to collect cloud-like characteristic data of training platelets, the cloud-like characteristic data includes cloud-like characteristic image data and / or cloud-like characteristic video data;

[0029] Testing the performance index data of platelets for training;

[0030] Preprocessing the cloud-like characteristic data of the training platelets to obtain the cloud-like characteristic preprocessed data of the training platelets that meets the input requirements of the platelet performance index evaluation model PEM;

[0031] Inputting the cloud-like characteristic preprocessing data of the training platelets into the platelet performance index evaluation model PEM to obtain the training platelet performance index prediction data;

[0032] Based on the training platelet performance index detection data and prediction data, a platelet performance index evaluation model PEM is trained.

[0033] In a possible implementation, the preprocessing of the cloud feature data of the training platelets to obtain the cloud feature preprocessed data of the training platelets that meets the input requirements of the platelet performance index evaluation model PEM includes:

[0034] Extracting key frame images from the cloud-like feature video data of the training platelets;

[0035] The key frame images are clustered using the K-means clustering algorithm to obtain the cloud-like characteristic key frame images of the training platelets at different cluster centers.

[0036] Performing denoising, resolution adjustment, image enhancement and normalization processing on the cloud-like characteristic key frame images of the training platelets at the different categories of cluster centers to obtain the cloud-like characteristic key frame images of the training platelets;

[0037] The key frame image of the cloudy characteristics of the training platelets is preprocessed data of the cloudy characteristics of the training platelets that meets the input requirements of the platelet performance evaluation model PEM.

[0038] In a possible implementation, the platelet performance index evaluation model PEM includes a convolutional neural network CNN and a recurrent neural network RNN.

[0039] In a possible implementation, the cloud-like feature preprocessing data of the training platelets is input into the platelet performance index evaluation model PEM to obtain the training platelet performance index prediction data; based on the training platelet performance index detection data and the prediction data, the platelet performance index evaluation model PEM is trained, including:

[0040] P1: Inputting the key frame image of the cloudy feature of the training platelet into the convolutional neural network CNN of the platelet performance index evaluation model PEM to obtain the key frame image features of the cloudy feature of the training platelet;

[0041] P2: Inputting the key frame image features of the cloud-like features of the training platelets into the recurrent neural network RNN, and outputting the predicted data of the performance index of the training platelets;

[0042] P3: Calculating the loss value of the platelet performance index evaluation model PEM based on the platelet performance index detection data and prediction data for training;

[0043] P4: using the loss value of the platelet performance index evaluation model PEM and the gradient back propagation algorithm to update the weight parameters and bias parameters of the platelet performance index evaluation model PEM;

[0044] P5: Repeat the process of P1-P4 until the preset number of training rounds, and obtain the platelet performance index evaluation model PEM through training.

[0045] In one possible implementation, the platelet performance index data include: adhesion rate, maximum aggregation rate, platelet factor 4 (PF4) concentration, blood clot contraction rate, mouse carotid artery thrombosis time, mouse tail bleeding time, the proportion of platelets transfused into the body, phosphatidylserine (PS) positivity rate, P-selectin (CD62P) positivity rate, and one or more of the average fluorescence intensity of platelet glycoprotein Ibα (CD42b).

[0046] According to another aspect of the present invention, a platelet performance index evaluation device is provided, the device comprising:

[0047] A collection device for collecting platelets to be tested;

[0048] An image and video acquisition device, used for acquiring cloud-like characteristic data of the platelets to be detected, wherein the cloud-like characteristic data includes cloud-like characteristic image data and / or cloud-like characteristic video data;

[0049] An evaluation module is used to evaluate the platelet performance index using the cloud-like characteristic data of the platelets to be detected.

[0050] According to another aspect of the present invention, an electronic device is provided. The device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the method described above is implemented.

[0051] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method described above is implemented.

[0052] The platelet performance index evaluation method of the present invention collects platelets to be tested; uses image and video equipment to collect cloud-like characteristic data of the platelets to be tested, wherein the cloud-like characteristic data includes cloud-like characteristic image data and / or cloud-like characteristic video data; and uses the cloud-like characteristic data of the platelets to be tested to evaluate the platelet performance index. The platelet performance index can be accurately, quickly and at low cost, solving the problems of inaccurate, time-consuming and high-cost results of traditional platelet performance testing.

[0053] Other optional features and technical effects of the embodiments of the present invention are partially described below, and partially can be understood by reading this document. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. The elements shown are not limited to the proportions shown in the accompanying drawings. The same or similar reference numerals in the accompanying drawings represent the same or similar elements, wherein:

[0055] Figure 1 A flow chart of a method for evaluating platelet performance indicators according to an embodiment of the present invention is shown;

[0056] Figure 2 A schematic diagram of a cloud-like characteristic image of platelets according to an embodiment of the present invention is shown;

[0057] Figure 3 A flowchart showing step S3 of a platelet performance index evaluation method according to an embodiment of the present invention is shown;

[0058] Figure 4 A schematic diagram of a frequency domain spectrum of a cloud-like characteristic image of platelets according to an embodiment of the present invention is shown;

[0059] Figure 5 A schematic diagram showing the gradient amplitude of a cloud-like characteristic image of platelets according to an embodiment of the present invention is shown;

[0060] Figure 6 A flowchart of a platelet performance index evaluation model PEM training method according to another embodiment of the present invention is shown;

[0061] Figure 7a A scatter plot showing the actual value and predicted value of the platelet performance indicator-adhesion rate according to one embodiment of the present invention;

[0062] Figure 7b A scatter plot showing the actual value and predicted value of the maximum aggregation rate, a platelet performance indicator, according to an embodiment of the present invention;

[0063] Figure 7c A scatter plot showing the actual value and predicted value of the platelet performance indicator-PF4 concentration according to one embodiment of the present invention;

[0064] Figure 7d A scatter plot showing the actual value and predicted value of the platelet performance indicator-blood clot contraction rate according to one embodiment of the present invention;

[0065] Figure 7e A scatter plot showing the actual value and predicted value of the platelet performance indicator - mouse carotid artery thrombosis time according to one embodiment of the present invention;

[0066] Figure 7f A scatter plot showing the actual value and predicted value of the platelet performance indicator - mouse tail bleeding time according to one embodiment of the present invention;

[0067] Figure 7g A scatter plot showing the actual value and predicted value of the platelet performance indicator - the proportion of platelets transfused into the body according to an embodiment of the present invention;

[0068] Figure 7h A scatter plot showing the actual value and predicted value of the platelet performance indicator-PS positive rate according to one embodiment of the present invention;

[0069] Figure 7i A scatter plot showing the actual value and predicted value of the platelet performance indicator CD62P positive rate according to one embodiment of the present invention;

[0070] Figure 7j A scatter plot showing the actual value and predicted value of the platelet performance indicator CD42b mean fluorescence intensity according to one embodiment of the present invention;

[0071] Figure 8 A structural diagram of a platelet performance index evaluation device based on the cloud-like characteristics of platelets according to an embodiment of the present invention is shown;

[0072] Fig. 9 A schematic diagram of an electronic device according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0073] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific implementation methods and drawings. Here, the exemplary implementation methods of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0074] As used herein, the term "including" and its variations mean open inclusion, i.e., "including but not limited to". Unless otherwise stated, the term "or" means "and / or". The term "based on" means "based at least in part on". The terms "an example embodiment" and "an embodiment" mean "at least one example embodiment". The term "another embodiment" means "at least one additional embodiment". The terms "first", "second", etc. may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0075] In addition, the steps shown in the flowchart of the accompanying drawings can be executed in a computer such as a set of computer executable instructions. Also, although a logical sequence is shown in the flowchart, in some cases, the steps shown or described can be performed in a sequence different from that here.

[0076] Figure 1 A flow chart of a method for evaluating platelet performance indicators according to an embodiment of the present invention is shown.

[0077] like Figure 1As shown, the method may include:

[0078] Step S1: Collecting platelets to be tested.

[0079] The platelets to be tested can be collected using a blood collection device, wherein the blood collection device can be a blood cell separator, a fully automatic blood cell apheresis machine, a blood cell analyzer, etc., which are not limited here. For example, according to GB18467-2011 "Health Examination Requirements for Blood Donors", a blood cell separator (Amicus 4R4580, Fenwal, USA) can be used to collect platelets from healthy volunteers. In addition, the platelets to be tested can also be sampled platelets, which are not limited here.

[0080] The collected platelets to be tested can be used to prepare leukocyte-depleted single-donor platelets, single-donor platelets, concentrated platelets, multi-person mixed concentrated platelets, etc. to be tested, and the image acquisition equipment can be used to capture the cloud-like characteristics of leukocyte-depleted single-donor platelets, single-donor platelets, concentrated platelets, multi-person mixed concentrated platelets, and leukocyte-depleted single-donor platelets.

[0081] The following is an example of leukocyte-reduced single-donor platelets to be tested. According to actual needs, you can choose leukocyte-reduced single-donor platelets, single-donor platelets, concentrated platelets, multi-person mixed concentrated platelets, etc., without limitation. For example, leukocyte-reduced single-donor platelets are prepared by collecting platelets, that is, each volunteer donates 2 therapeutic doses (each therapeutic dose is 225-275mL, containing no less than 2.5×10 11 The leukocyte-depleted single-donor platelets (platelets) are placed in a platelet shaking storage box for continuous shaking storage. The temperature range of the incubator is 20° C.-24° C., and no limitation is made here.

[0082] Step S2: using image and video equipment to collect cloud-like characteristic data of the platelets to be detected, wherein the cloud-like characteristic data includes cloud-like characteristic image data and / or cloud-like characteristic video data.

[0083] Figure 2 A schematic diagram of a cloud-like characteristic image based on platelets to be detected according to an embodiment of the present invention is shown.

[0084] The image and video equipment may be a video recorder, a camcorder, a still camera, a webcam, as well as mobile phones, computers and other equipment with photo and video recording functions, and are not limited to any one of them here.

[0085] Platelet concentrates are composed of a liquid medium and platelets, in which the platelets are suspended and aggregated. When properly shaken, the platelet concentrates will exhibit a cloudy appearance, which is referred to as the cloudy characteristic of platelets.

[0086] The leukocyte-depleted single-donor platelet storage bags obtained and stored in step S1 are collected in a standardized manner. The standardized collection process is to set the ambient temperature to about 22°C, and use a light source to illuminate from below the platelets; fix the platelets on a shaking table, set the maximum deviation distance of the shaking amplitude to 15 cm, and the shaking frequency to 4 Hz, and start shaking. When the shaking starts, turn on the image and video equipment, and stop shaking after 6 seconds; after the shaking stops, use the video acquisition equipment to collect platelet images or videos, and the collection time is a preset duration, and the preset duration can be set to a suitable duration based on the platelet-like texture characteristics to be detected, preferably 5 seconds; stop recording the video after the cloudy texture of the platelets dissipates. The cloudy characteristic data of the platelets to be detected includes cloudy characteristic image data and / or cloudy characteristic video data, wherein the cloudy characteristic texture of the platelets to be detected is as follows: Figure 2 By standardizing the cloud-like characteristic data of the platelets to be detected, the error caused by different collection processes between the cloud-like characteristic data of the platelets to be detected can be minimized.

[0087] Step S3: using the cloud-like characteristic data of the platelets to be tested to evaluate the platelet performance index.

[0088] Among them, platelet performance index data may include: adhesion rate, maximum aggregation rate, platelet factor 4 (PF4) concentration, blood clot contraction rate, mouse carotid artery thrombosis time, mouse tail bleeding time, proportion of platelets transfused into the body, phosphatidylserine (PS) positivity rate, P-selectin (CD62P) positivity rate, and platelet glycoprotein Ibα (CD42b) average fluorescence intensity.

[0089] For example, when a blood vessel is damaged, platelets will respond immediately: attach to the damaged part through adhesion; quickly aggregate to form a thrombus to block the damaged part of the blood vessel; and further enhance the hemostatic effect through release and contraction. However, platelets will undergo storage damage during in vitro storage, which is specifically manifested in: the glycoprotein structure on the surface of platelets changes, resulting in impaired adhesion function; platelets undergo apoptosis, resulting in reduced release of bioactive substances; platelets become activated, which then induces rapid clearance by the body; the final result is that platelets have hemostatic dysfunction and a shortened survival period in vivo. Therefore, the adhesion, aggregation, release, and contraction functions of platelets are the key to their hemostatic effect. The stronger these functions are, and the lighter the platelet storage damage is, the stronger the hemostatic function of platelets is, which is specifically manifested in the shorter carotid artery thrombosis time and tail-cut bleeding time of mice. Therefore, evaluating the quality of the above-mentioned platelet performance indicators has important guiding significance in the precise transfusion of platelets in clinical medicine.

[0090] The platelet performance index evaluation method of the present invention can accurately, quickly and at low cost detect platelet performance indexes, and solves the problems of traditional platelet performance detection methods such as long time consumption, high cost and large result errors.

[0091] Embodiment 1

[0092] Figure 3 A flow chart showing step S3 of a method for evaluating platelet performance indicators according to an embodiment of the present invention is shown.

[0093] In one example, as shown in 3, step S3 may include:

[0094] Step S311: performing frequency domain analysis and gradient analysis on each frame of the cloud-like characteristic data of the platelets to be detected, to obtain a comprehensive evaluation index of the cloud-like characteristic data of the platelets to be detected;

[0095] Step S312: recording the duration of the cloud-like characteristic of the platelets to be detected, and obtaining the duration of the cloud-like characteristic data of the platelets to be detected;

[0096] Step S313: obtaining the significant features of the cloud-like characteristic data of the platelets to be detected according to the comprehensive evaluation index and duration of the cloud-like characteristic data of the platelets to be detected;

[0097] Step S314: Evaluate the platelet performance index based on the correspondence between the significant features of the cloud-like feature data of the platelets to be detected and the platelet performance index.

[0098] Figure 4 and Figure 5 A schematic diagram of the frequency domain spectrum and a schematic diagram of the gradient amplitude of a cloudy characteristic image of platelets to be detected according to an embodiment of the present invention are respectively shown.

[0099] The comprehensive evaluation index of the cloud-like characteristic data of the platelets to be detected in step S311 includes the spectrum energy ratio, the gradient amplitude mean and the fringe direction score. Then the comprehensive evaluation index S of the cloud-like characteristic data of the platelets to be detected is:

[0100] S=αR freq +βG mean +γH s (1)

[0101] Among them, R freq is the spectrum energy ratio, G mean is the mean value of gradient amplitude, H s is the stripe direction score; α is the frequency domain feature weight, which can be taken as 0.5; β is the gradient feature weight, which can be taken as 0.3; γ is the directional feature weight, which can be taken as 0.2.

[0102] Figure 4 A schematic diagram of the frequency domain spectrum of a cloud-like characteristic image of a platelet to be detected according to an embodiment of the present invention is shown.

[0103] Calculate the spectral energy ratio R freq :

[0104] Spectral energy ratio R freq The ratio of the high-frequency component energy (related to the fringe characteristics) in the cloudy characteristic map of the platelet to be detected to the total energy (the ratio of the high-frequency component). The frequency domain characteristics of the cloudy characteristic map of the platelet to be detected can be used to measure the ratio of the high-frequency component energy (related to the fringe characteristics) in the cloudy characteristic map of the platelet to be detected to the total energy. The specific calculation process is as follows:

[0105] Step L1: Convert the input RGB image of the cloud feature map of the platelet to be detected into Figure 2 The grayscale image I(x, y) of the cloudy characteristic map of the platelets to be detected is converted from the spatial domain to the frequency domain (frequency domain) using discrete Fourier transform:

[0106]

[0107] Wherein, M x N is the grayscale image size of the cloud-like features of the platelets to be detected, and M and N are positive integers.

[0108] Step L2: The fft.fftshift function in the numpy library shown in formula (3) can be used to move the low-frequency part of the frequency domain feature of the cloud-like feature map of the platelets to be detected to the center of the spectrum:

[0109] F′(u,v)=F(u,v)·(-1) u+v (3)

[0110] Step L3: Extract the amplitude of the frequency component of the frequency domain feature of the cloud-like feature map of the platelet to be detected, ignoring the phase information:

[0111]

[0112] Wherein, Re(F′(u, v)) and Im(F′(u, v)) are the real part and the imaginary part of the Fourier transform, respectively.

[0113] Step L4: Using equation (5), take the logarithm of the amplitude of the frequency component of the frequency domain feature of the cloud-like characteristic graph of the platelet to be detected, and obtain Figure 4 The frequency domain spectrum is shown. The logarithmic value of the amplitude of the frequency component is used to compress the dynamic range, which can enhance the display effect.

[0114] S(u, v)=log(1+|F′(u, v)|) (5)

[0115] Step L5: Calculate the high-frequency energy of the frequency domain characteristics of the cloud-like characteristic graph of the platelets to be detected. For example, the frequency components outside the frequency radius R from the center of the spectrum can be selected as high-frequency components (i.e., high-frequency areas), and the frequency radius r (u, v) from the center of the spectrum is defined as:

[0116]

[0117] The frequency component with a frequency radius r(u, v)>R from the center of the spectrum is expressed as a high-frequency component (high-frequency region), thereby obtaining high-frequency energy:

[0118]

[0119] Step L6: Calculate the energy sum of all high-frequency components in the frequency domain characteristics of the cloud-like characteristic graph of the platelet to be detected:

[0120]

[0121] Step L7: Calculate the spectrum energy ratio R freq (Percentage of high-frequency energy):

[0122]

[0123] Figure 5 A schematic diagram of the gradient amplitude of a cloudy characteristic image of platelets to be detected according to an embodiment of the present invention is shown.

[0124] Calculate the mean gradient amplitude G mean :

[0125] The gradient characteristics of the frequency domain image of the cloud-like characteristic of the platelets to be detected can be used to measure the significance of the pixel intensity change of the cloud-like characteristic image of the platelets to be detected. The significance of the pixel intensity change is related to the clarity of the stripe edge. Calculate the mean value of the gradient amplitude G mean The specific process is as follows:

[0126] Step Q1: Use the Sobel operator (image gradient extraction operator) to calculate the horizontal gradient G of the grayscale image of the cloud-like feature of the platelet to be detected x (x, y) and the vertical gradient G y (x, y):

[0127]

[0128]

[0129] Among them, * represents the convolution operation.

[0130] Step Q2: Horizontal gradient G of the grayscale image of the cloud-like feature of the platelet to be detected calculated in step Q1 x (x, y) and the vertical gradient G y (x, y), and the gradient amplitude of each pixel of the grayscale image of the cloud-like feature of the platelet to be detected is calculated according to formula (12):

[0131]

[0132] Step Q3: Calculate the mean value G of the gradient amplitude of all pixels of the grayscale image of the cloud-like feature of the platelet to be detected mean , which is used to represent the average degree of pixel intensity change in the overall grayscale image of the cloud-like features of the platelets to be detected. Gradient amplitude mean G mean for:

[0133]

[0134] The gradient amplitude mean G mean Mapping to the color range of the cloud-like characteristic image of the platelet to be detected gives Figure 5 The gradient amplitude map of the cloudy characteristic image of the platelets to be detected is shown.

[0135] Calculate the distribution histogram entropy of the gradient direction:

[0136] Step M1: Obtain the gradient direction θ(x, y) of each pixel in the grayscale image I(x, y) of the cloud-like characteristic image of the platelets to be detected. The gradient direction θ(x, y) is expressed in radians and has a value range of [-π, π]. Then the gradient direction θ(x, y) is:

[0137]

[0138] Step M2: discretize the gradient direction θ(x, y) of each pixel point in the grayscale image I(x, y) of the cloudy characteristic image of the platelets to be detected, and evenly divide the gradient direction θ(x, y) of each pixel point in the continuous grayscale image I(x, y) of the cloudy characteristic image of the platelets to be detected into N direction intervals, where N is a positive integer.

[0139]

[0140] Among them, i∈[1,N] represents the i-th direction interval.

[0141] Step M4: Count the frequency of each gradient direction in the grayscale image I(x,y) of the cloudy characteristic image of the platelets to be detected, and construct a distribution histogram of the gradient direction of the grayscale image I(x,y) of the cloudy characteristic image of the platelets to be detected. Each item of the gradient direction histogram H represents the number of pixels belonging to the direction interval in the grayscale image I(x,y) of the cloudy characteristic image of the platelets to be detected:

[0142]

[0143] Among them, δ is the Kronecker delta function (Kronecker function, an inner chain binary function) used to determine whether the gradient direction θ(x, y) falls in the i-th direction interval, and bin(θ) is the interval number corresponding to the direction θ.

[0144] Step M5: Normalize the distribution histogram of the gradient direction of the grayscale image I(x,y) of the cloud-like characteristic image of the platelets to be detected into a probability distribution:

[0145]

[0146] Among them, j∈[1,N] represents the j-th direction interval.

[0147] Step M6: Calculate the entropy of the gradient direction histogram of the grayscale image I(x,y) of the cloud-like characteristic image of the platelet to be detected according to the Shannon entropy (information entropy) formula:

[0148]

[0149] Step M7: Calculate the stripe direction score:

[0150] H s =1-H / log(N) (19)

[0151] The distribution of the grayscale image gradient direction of the cloudy characteristic image of the platelets to be detected can be measured by using the entropy of the grayscale image gradient direction distribution of the cloudy characteristic image of the platelets to be detected, which can effectively reflect the significance of the image stripes. For example, a higher entropy value usually indicates that the grayscale image gradient direction distribution of the cloudy characteristic image of the platelets to be detected is more uniform, and the stripe directions of the cloudy characteristic image of the platelets to be detected are more diverse and dispersed; a lower entropy value indicates that the grayscale image gradient direction distribution of the cloudy characteristic image of the platelets to be detected is concentrated, and the stripe directions of the cloudy characteristic image of the platelets to be detected are consistent. That is, by calculating the distribution entropy of the grayscale image gradient direction of the cloudy characteristic image of the platelets to be detected, the directional complexity of the stripes in the cloudy characteristic image of the platelets to be detected can be evaluated.

[0152] Through the above steps, the comprehensive evaluation index S of the cloud-like characteristic data of the platelets to be detected can be calculated, the duration T of the cloud-like characteristic data of the platelets to be detected can be recorded, and the comprehensive evaluation index S of the cloud-like characteristic data of the platelets to be detected and its duration T can be multiplied to obtain the significance index of the cloud-like characteristic data of the platelets to be detected for each frame. In one example, the platelet performance index can be evaluated based on the correspondence between the significance features of the cloud-like characteristic data of the platelets to be detected and the platelet performance index, that is, the maximum significance feature of the cloud-like characteristic data of the platelets to be detected is determined, and the quality of the platelet performance index can be evaluated based on the correspondence between the maximum significance feature of the cloud-like characteristic data of the platelets to be detected and the level of the platelet performance index.

[0153] For example, platelet performance indicators can be divided into different levels according to the platelet coagulation function. For example, the platelet performance indicators can be divided into five levels from level 1 to level 5 according to the platelet coagulation function. Level 5 means that the platelet performance is the best and the platelet coagulation function is the best; level 4, level 3, level 2, and level 1 indicate that the platelet performance gradually decreases; level 1 indicates that the platelet performance is the worst and the platelet coagulation function is also the worst. Accurate evaluation of platelet performance indicators has important guiding significance in the clinical precise platelet transfusion. For example, when a doctor applies to the blood transfusion department for concentrated platelets to effectively stop bleeding for an acute bleeding patient, if the platelet performance indicator level is level 5, the platelet performance is the best, and the blood transfusion department can effectively stop bleeding by issuing 2 units of concentrated platelets; conversely, if the platelet performance indicator level is level 1, the platelet performance is the worst, in order to achieve the same treatment effect, the blood transfusion department issues 10 units of concentrated platelets. In addition, it also plays an important role in guiding the use of antiplatelet drugs including aspirin and clopidogrel. For example, a potential stroke patient has a platelet performance index level of 5 before taking aspirin. If he takes 100 mg of aspirin every day, his platelet performance index level should drop below 5. If the platelet performance index level does not decrease, it means that the treatment dose is not effective and the dose should be increased; if the platelet performance index level does not decrease after the aspirin dose is increased to 300 mg, it means that aspirin is not effective for the patient and other antiplatelet drugs should be switched.

[0154] According to a large number of experimental analyses, the correspondence between the significant features of the cloud-like characteristic data of the platelets to be detected and the platelet performance index can be obtained. For example, the significant feature value of the cloud-like characteristic data of the platelets to be detected is in [0, 5), and the corresponding platelet performance index level is 1; the significant feature value of the cloud-like characteristic data of the platelets to be detected is in [5, 8), and the corresponding platelet performance index level is 2; the significant feature value of the cloud-like characteristic data of the platelets to be detected is in [8, 13), and the corresponding platelet performance index level is 3; the significant feature value of the cloud-like characteristic data of the platelets to be detected is in [13, 17), and the corresponding platelet performance index level is 4; the significant feature value of the cloud-like characteristic data of the platelets to be detected is greater than or equal to 17, and the corresponding platelet performance index level is 5.

[0155] When the same platelet sample is continuously photographed or recorded using an image video device, the cloud-like feature of the platelet sample lasts for a certain period of time T. At this time, the comprehensive evaluation index S of each frame image in the platelet cloud-like video can be calculated using formula (1), and the largest comprehensive evaluation index S is selected. The largest comprehensive evaluation index S is multiplied by the cloud-like feature duration T of the corresponding platelet sample to obtain the maximum significant feature index of the platelet sample. At this time, the cloud-like feature (texture, edge, etc.) of the platelet sample frame image is the most significant. According to the size of the significant feature index of the platelet sample and its corresponding relationship with the platelet performance index level, the level of the platelet performance index can be found, and the platelet performance is judged by the platelet performance index level. The quality of platelet performance indicators can be evaluated efficiently and quickly, which has an important guiding role in clinical platelet precision transfusion and antiplatelet drug application.

[0156] Experimental verification:

[0157] The correlation coefficient R is a statistic that measures the correlation between the platelet performance index and the significant features of the cloud-like characteristic data of the platelets to be tested, and represents the degree of fit of the correlation between the platelet performance index and the significant features of the cloud-like characteristic data of the platelets to be tested.

[0158] Correlation coefficient

[0159] in, δX is the standard deviation of variable X, and δY is the standard deviation of variable Y. In this embodiment, variable X is a platelet performance index, and variable Y is a significant feature of the cloud-like characteristic data of the platelets to be detected.

[0160] The value range of the correlation coefficient R is between -1 and 1. The larger the absolute value of the correlation coefficient R, the stronger the correlation between the variables X and Y, and the better the model fitting degree of the correlation coefficient R. As can be seen from Table 1, the R values ​​are all above 0.8, and some R values ​​are above 0.9, indicating that the correlation between the platelet performance index and the significant characteristics of the cloud-like characteristic data of the platelets to be tested is stronger.

[0161] Table 1: Correlation coefficients R between the significant features of the cloud-like characteristic data of the platelets to be tested and the platelet performance indicators

[0162]

[0163] Embodiment 2

[0164] According to another aspect of the present invention, evaluating platelet performance indexes using cloud-like characteristic data of platelets to be detected may include: inputting the cloud-like characteristic data of platelets to be detected into a trained platelet performance index evaluation model PEM to obtain platelet performance index evaluation data. The platelet performance index evaluation model (PEM) is used to evaluate platelet performance to guide clinical accurate platelet transfusion, application of antiplatelet drugs, and other anticoagulant or procoagulant treatments.

[0165] Among them, the platelet performance index evaluation model PEM is trained based on a large amount of platelet performance index data and platelet cloud feature data. The platelet performance index evaluation model PEM includes a convolutional neural network CNN and a recurrent neural network RNN. The convolutional neural network B-CNET is a selected existing convolutional neural network (CNN) backbone network model, which can extract the platelet cloud feature from the platelet cloud feature image or video frame; the convolutional neural network B-CNET includes a sequentially connected input layer, 4 convolutional layers, 2 pooling layers, a batch normalization layer, a convolution 3D layer, and a fully connected layer. The recurrent neural network B-RNET is a selected existing recurrent neural network (RNN) backbone network model, which can process the dynamic characteristics of platelet cloud over time; the recurrent neural network B-RNET includes an input layer, a hidden layer, and an output layer.

[0166] In this embodiment, the platelet performance index evaluation method may include: using an image and video device to collect cloud-like characteristic data of the platelets to be detected, wherein the cloud-like characteristic data includes cloud-like characteristic image data and / or cloud-like characteristic video data; preprocessing the cloud-like characteristic data of the platelets to be detected to obtain cloud-like characteristic preprocessed data of the platelets that meet the input requirements of the platelet performance index evaluation model PEM; inputting the cloud-like characteristic preprocessed data of the platelets into the platelet performance index evaluation model PEM to obtain platelet performance index data. The platelet performance index prediction data obtained based on the platelet performance index evaluation model PEM further evaluates the quality of the platelet performance index.

[0167] Figure 6 A flow chart of a platelet performance evaluation model PEM training method according to an embodiment of the present invention is shown.

[0168] In one possible implementation, Figure 6 As shown, the platelet performance index evaluation model PEM is trained based on platelet performance index data and platelet cloud feature data, and may include:

[0169] S61: Collecting platelets for training. The collection of platelets can be implemented using the solution in the first embodiment, which will not be described in detail here.

[0170] S62: Using an image and video device to collect the cloud feature data of the training platelets, the cloud feature data including the cloud feature image data and / or the cloud feature video data. The collection of the cloud feature data of the platelets can be implemented using the solution in the first embodiment, which will not be described in detail here.

[0171] S63: Detect performance indicator data of platelets used for training.

[0172] Platelet performance index data can be detected in the following ways, such as adhesion rate, maximum aggregation rate, platelet factor 4 (PF4) concentration, blood clot contraction rate, mouse carotid artery thrombosis time, mouse tail bleeding time, the proportion of platelets transfused into the body, phosphatidylserine (PS) positivity rate, P-selectin (CD62P) positivity rate, and platelet glycoprotein Ibα (CD42b) average fluorescence intensity.

[0173] Platelet adhesion rate test. Take 0.5 ml of the leukocyte-reduced single-donor platelets to be tested, and use a blood cell analyzer (XP-100, Sysmex, Japan) to count the platelet concentration (before contact). Take 1.5 ml of the leukocyte-reduced single-donor platelets and place them in a spherical bottle. Place the spherical bottle on a rotating device and rotate it at a speed of 3 r / min for 15 minutes to allow the platelets to fully contact the bottle wall. Take 0.5 ml of the leukocyte-reduced single-donor platelets in the spherical bottle and use a blood cell analyzer to count the platelet concentration (after contact). Platelet adhesion rate = (platelet concentration before contact - platelet concentration after contact) / platelet concentration before contact × 100%.

[0174] Platelet maximum aggregation rate detection. The maximum aggregation rate of platelets can be detected using a hemagglutinator (CS-2400, Sysmex, Japan). According to the instructions of the platelet aggregation function (ADP) kit (Sysmex, Japan), the maximum aggregation rate of platelets was detected respectively.

[0175] PF4 concentration detection: Take 1 ml of leukocyte-reduced platelets to be tested, centrifuge at 1000 g for 5 min, obtain the supernatant, and detect the PF4 concentration according to the instructions of the PF4 ELISA test kit (Boster, China).

[0176] Blood clot shrinkage rate test. Take 0.5 ml of the leukocyte-reduced platelet to be tested, add calcium chloride to coagulate the plasma to form a clot. After removing the plasma clot, read the volume of the remaining serum (i.e. the volume of the serum precipitated). Blood clot shrinkage rate = (leukocyte-reduced platelet volume - precipitated serum volume) / leukocyte-reduced platelet volume × 100%.

[0177] Determination of carotid artery thrombosis time in mice. NOD-SCID mice aged 10 to 12 weeks were housed in a sterile environment. Anti-mouse CD42b monoclonal antibody (0.5 μL / g, Emfret, Germany) was injected into the tail vein to remove the autologous platelets in the mice. After 24 h, 1.25% tribromoethanol (30 μL / mg, Aibei Biotechnology, China) was used to induce anesthesia. Rhodamine 6G solution (0.5 mg / mL, 8 μL / g, Sigma, USA) was injected intraperitoneally. Leukocyte-reduced platelets (1×10 9 / mL, 18μL / g). The sublingual gland was separated to fully expose the common carotid artery. Filter paper soaked in FeCl3 solution (200mg / mL, Aladdin, China) was covered on the exposed common carotid artery for 2min. A stereofluorescence microscope (AXIO Zoom.V16, Zeiss, Germany) was used to observe the process of thrombosis in the common carotid artery and record its formation time.

[0178] The tail-clip bleeding time of mice was measured. Anti-mouse CD42b monoclonal antibody was injected through the tail vein to eliminate the autologous platelets in NOD-SCID mice. After 24 hours, the leukocyte-reduced platelets (1×10 9 / mL, 18μL / g). Cut the tail tip 3mm away from the tail tip, and quickly put the mouse with the tail cut off into a centrifuge tube filled with preheated saline (maintained at 37℃). Start timing from the time the mouse tail starts bleeding, and stop timing when the bleeding stops and there is no more bleeding within 1 minute.

[0179] In vivo survival test of transfused platelets. Leukocyte-depleted apheresis platelets to be tested were injected into 12-week-old NOD-SCID mice (1×10 9 / mL, 14μL / g), and 50μL of peripheral blood was collected 2h later. After centrifugation at 250g for 10min, 1mL of ACK lysis buffer (Boster, China) was added to the collected peripheral blood pellet to lyse the red blood cells. After 1min, 2mL of phosphate buffer was added and centrifuged again at 250g for 10min, and the pellet was fixed with 4% paraformaldehyde for 5min. 1mL of flow cytometry staining buffer was added and centrifuged again at 250g for 10min for washing. Platelets were labeled with APC-conjugated anti-human CD42a and FITC-conjugated anti-mouse CD42a antibodies and analyzed using a flow cytometer (FACS Canto, BD, USA).

[0180] PS positive platelet rate detection. Take the leukocyte-reduced single-donor platelets to be tested (1 μL, concentration 1×10 9 / mL), refer to the instructions of the apoptosis detection kit (BD, USA), add 5 μL Annexin V-FITC and mix well, incubate at room temperature in the dark for 15 min, wash three times with buffer and analyze on a flow cytometer (FACS Canto, BD, USA).

[0181] CD62P positive platelet rate detection. Take the leukocyte-reduced single-donor platelets to be tested (1 μL, concentration 1×10 9 / mL), 5 μL FITC-conjugated anti-human CD62P (BD, USA) was added and mixed, incubated at room temperature in the dark for 15 min, washed three times with flow cytometry buffer and analyzed by flow cytometer (FACS Canto, BD, USA).

[0182] Platelet CD42b mean fluorescence intensity detection. Take the leukocyte-reduced single-donor platelets to be tested (1 μL, concentration 1×10 9 / mL), add 5 μL APC-conjugated anti-human CD42b (BD, USA) and mix well, incubate at room temperature in the dark for 15 min, wash three times with flow cytometry buffer and analyze on a flow cytometer (FACS Canto, BD, USA).

[0183] The platelet performance index detection method described above can be used to detect platelet performance index evaluation data sets at different time points. For example, 100 platelet performance index data stored on the first day, 100 platelet performance index data stored on the second day, 100 platelet performance index data stored on the third day, 100 platelet performance index data stored on the fourth day, and 100 platelet performance index data stored on the fifth day can be collected to form a platelet performance index data set of 500 cases.

[0184] Similarly, the method in step S2 can be used to collect the cloud-like characteristic images and / or video data of platelets corresponding to the above-mentioned 500 platelet performance index evaluation data sets.

[0185] The 500 platelet performance index data sets and the corresponding 500 platelet cloud feature images and / or video data together constitute a platelet performance index evaluation data set. The platelet performance evaluation data set is divided into training data T and test data C for the platelet performance index evaluation model PEM according to any ratio from 7:3 to 9:1.

[0186] S64: Preprocessing the cloud-like characteristic data of the training platelets to obtain the cloud-like characteristic preprocessed data of the training platelets that meets the input requirements of the platelet performance evaluation model PEM.

[0187] Firstly, by extracting key frame images from the cloudy feature video data of platelets, the processed cloudy feature video of platelets can have a suitable time series length, thereby reducing redundant information and video frame rate.

[0188] Extracting the key frame may be to perform cluster analysis on N cloud-like feature images within a preset time length using a K-means clustering algorithm, and obtain the most representative cloud-like feature image of the cluster center of different categories as the key frame image, and C key frame images of different categories constitute a cloud-like feature video V of platelets of a time series with a fixed length of C, wherein the number of cluster categories may be set to C. C is a positive integer, C≥1, and is not limited here.

[0189] Secondly, the resolution of the key frame images in the cloud feature video V is adjusted to ensure that the key frame image data size of all platelet cloud features input to the convolutional neural network B-CNET of the platelet performance index evaluation model PEM is a fixed size, such as 1024*768. Other resolutions can also be selected as needed, without any limitation. And the key frame images of the platelet cloud features after adjusting the resolution are denoised and image enhanced. And during the training process, the key frame image data of the platelet cloud feature is randomly enhanced, rotated, translated, scaled, flipped, contrast adjusted, brightness adjusted, etc., so that it can more easily capture the cloud feature image of the training platelet or the cloud feature (texture feature) in the video frame data image.

[0190] Finally, the pixel values ​​of the key frame image data of the platelet cloudy feature are normalized (for example, maximum-minimum processing, etc.) to scale the pixel values ​​of the key frame image data of the platelet cloudy feature to the range of 0 to 1 or standardized so that the distribution of the pixel values ​​of the key frame image data of the platelet cloudy feature obeys the normal distribution. In this way, the platelet cloudy feature preprocessing data that meets the input requirements of the platelet performance index evaluation model PEM is obtained.

[0191] Through step S641, the cloudy characteristic images or video frame data images of platelets with different time series lengths can be processed to meet the input feature requirements of the platelet performance index evaluation model PEM, thereby improving the training efficiency of the platelet performance index evaluation model PEM.

[0192] S65: inputting the cloud-like feature preprocessing data of the training platelets into the platelet performance index evaluation model PEM to obtain the training platelet performance index prediction data;

[0193] S66: Based on the platelet performance index detection data and prediction data for training, train the platelet performance index evaluation model PEM.

[0194] Before training the platelet performance index evaluation model PEM, the deep learning framework PyTorch can be used to define the convolutional neural network B-CNET and the recurrent neural network B-RNET in the platelet performance index evaluation model PEM, and initialize the convolutional neural network B-CNET and the recurrent neural network B-RNET. The learning rate of the platelet performance index evaluation model PEM is initialized to 0.01, the number of training epochs is 50, and the batch_size of the training samples is 32.

[0195] The platelet performance evaluation model PEM is trained using the training data T of the platelet performance evaluation data set that meets the normal distribution condition constructed in step S63. The specific process is as follows:

[0196] P1: Input the key frame image of the cloudy feature of the training platelets into the convolutional neural network B-CNET of the platelet performance index evaluation model PEM, and output the key frame image features of the cloudy feature of the platelets to be detected, such as the edge, texture, color, contour, corner point, shape feature and angular features of the key frame image of the cloudy feature of the platelets to be detected, to form a key frame image feature parameter matrix of the cloudy feature of the platelets to be detected.

[0197] P2: Input the key frame image features of the cloud-like features of the platelets to be detected into the recurrent neural network B-RNET, and output the platelet performance index prediction data. That is, the key frame image feature parameter matrix of the cloud-like features of the platelets to be detected is input into the recurrent neural network B-RNET for feature association, and the key frame image feature parameter matrix of the cloud-like features of the platelets to be detected is mapped and classified into the platelet performance index prediction data through the fully connected layer of the platelet performance index evaluation model PEM, and the platelet performance index prediction data is output.

[0198] P3: Based on the platelet performance index detection data and prediction data, calculate the loss value of the platelet performance index evaluation model PEM.

[0199] The loss value of the platelet performance index evaluation model PEM can be calculated according to the L1 loss function of formula (21):

[0200]

[0201] in, is the true value of the mth platelet performance index of the nth sample of the platelet performance index evaluation model PEM, is the platelet performance index prediction result corresponding to the platelet performance index evaluation model PEM, M and N are both positive integers, M is the number of platelet performance indexes, and N is the number of training samples of the platelet performance index evaluation model PEM.

[0202] P4: Use the loss value of the platelet performance index evaluation model PEM and the gradient back propagation algorithm to update the weight parameters and bias parameters of the platelet performance index evaluation model PEM.

[0203] The weight parameter W and bias parameter b of the platelet performance evaluation model PEM can be updated by the SGD optimizer and the gradient back propagation algorithm, specifically:

[0204]

[0205] Among them, W and b are the weight and bias parameters of the platelet performance index evaluation model PEM before updating, and W' and b' are the weight and bias parameters of the platelet performance index evaluation model PEM after updating. and are the gradients of the loss L with respect to the weight parameter W and the bias parameter b, respectively, and η is the learning rate.

[0206] P5: Repeat the process of P1-P4 until the preset number of training epochs, and train to obtain the platelet performance index evaluation model PEM.

[0207] After steps S61-S66, a trained platelet performance index evaluation model PEM can be obtained. Finally, the cloud-like characteristic data of the platelets to be detected are input into the platelet performance index evaluation model PEM to obtain platelet performance index evaluation data.

[0208] According to another aspect of the present invention, a platelet performance index evaluation device is proposed, which may include: an image and video device for collecting cloud-like characteristic data of platelets to be detected, wherein the cloud-like characteristic data includes cloud-like characteristic image data and / or cloud-like characteristic video data; a preprocessing module for preprocessing the cloud-like characteristic data of platelets to be detected to obtain cloud-like characteristic preprocessed data of platelets that meet the input requirements of a platelet performance index evaluation model PEM; a model output module for inputting the cloud-like characteristic preprocessed data of platelets into the platelet performance index evaluation model PEM and outputting platelet performance index data; an evaluation module for evaluating the quality of the platelet performance index based on the platelet performance index prediction data obtained by the platelet performance index evaluation model PEM.

[0209] The platelet performance index evaluation model PEM training method of the platelet performance index cloud feature of the present invention inputs the platelet performance index cloud feature data in the test data C in the platelet performance index evaluation data set into the platelet performance index evaluation model PEM, and outputs the platelet performance index prediction value. The platelet performance index prediction value output by the platelet performance index evaluation model PEM is compared with the platelet performance index detection result (measured value, true value) corresponding to the platelet performance index cloud feature data in the test data C in the platelet performance evaluation data set to verify the accuracy of the platelet performance index evaluation model PEM.

[0210] Verification effect:

[0211] The experimental running environment is Ubuntu 18.04 and Python 3.7, the hardware configuration is Intel Core i7-8700K CPU, Nvidia RTX 2080Ti GPU and 32GB RAM, and the deep learning framework used is Pytorch 1.4.0.

[0212] Figures 7a-7j The scatter plots respectively show the actual value and the predicted value of the platelet performance indicator according to an embodiment of the present invention.

[0213] Input multiple frames of cloud-like feature data of the platelets to be detected into the platelet performance index evaluation model PEM, and the platelet performance index evaluation model PEM outputs multiple platelet performance index prediction values ​​and their corresponding true values ​​as scatter plots, such as Figures 7a-7j The horizontal axis in the figure is the measured value of the platelet performance index, and the vertical axis is the predicted value of the platelet performance index output by the platelet performance index evaluation model PEM. Figures 7a-7j It can be seen that most of the scattered points are concentrated around the straight line x=y, indicating that the closer the predicted value and measured value of the platelet performance index output by the platelet performance index evaluation model PEM are, the better the performance of the platelet performance index evaluation model PEM is. It also shows that the cloudy characteristics of platelets are strongly correlated with the platelet performance index.

[0214] Taking the carotid thrombosis time of mice in the test data C in the platelet performance evaluation data set as a benchmark, the correlation coefficient between the carotid thrombosis time of mice in the test data C in the platelet performance evaluation data set and other indicators in the platelet performance index is calculated by formula (20), as shown in the true value row in Table 2, and at the same time, the correlation trend between the carotid thrombosis time of mice and other indicators in the platelet performance index in the test data C in the platelet performance evaluation data set can be obtained; the cloud-like characteristic data of the platelets in the test data C in the platelet performance evaluation data set is input into the platelet performance index evaluation model PEM to obtain the platelet performance index test value, and taking the predicted carotid thrombosis time of mice as a benchmark, the correlation coefficient between the predicted value of the carotid thrombosis time of mice and the predicted values ​​of other indicators in the platelet performance index is calculated by formula (20), as shown in the test value row in Table 2, and the correlation trend between the predicted value of the carotid thrombosis time of mice and the predicted values ​​of other indicators in the platelet performance index can also be obtained.

[0215] Table 2 Correlation between platelet performance indicators

[0216]

[0217]

[0218] As can be seen from Table 2, the difference between the correlation coefficient between the carotid artery thrombosis time of mice in the test data C in the platelet performance evaluation data set and other indicators in the platelet performance index (the true value in Table 2) and the correlation coefficient between the carotid artery thrombosis time of mice and other indicators in the platelet performance index output by the platelet performance index evaluation model PEM (the predicted value in Table 2) is basically within 0.2, indicating that the correlation trend between the carotid artery thrombosis time of mice in the test data C in the platelet performance evaluation data set and other indicators in the platelet performance index is basically consistent with the correlation trend between the carotid artery thrombosis time of mice and other indicators in the platelet performance index predicted by the platelet performance index evaluation model PEM.

[0219] Table 3: PEM accuracy evaluation index of platelet performance index evaluation model

[0220]

[0221]

[0222] in, represents the predicted value of the platelet performance index evaluation model PEM, y i Indicates the true value of platelet performance indicators.

[0223] MAE is the mean absolute error evaluation index (Mean Absolute Error) and is a commonly used continuous data evaluation method. The mean absolute error evaluation index MAE is used to evaluate the prediction accuracy of the platelet performance index evaluation model PEM. When the MAE value is 0, it means that the predicted value of the platelet performance index evaluation model PEM is exactly the same as the true value of the platelet performance index; the smaller the MAE value, the closer the predicted value of the platelet performance index evaluation model PEM is to the true value of the platelet performance index, indicating that the prediction accuracy of the platelet performance index evaluation model PEM is higher. As can be seen from Table 2, the MAE value of each platelet performance index is very small relative to the normal value of the platelet performance index (the MAE value of the mean fluorescence intensity of CD42b is 1979.78, and the normal value is about 20,000 to 30,000, which is very small in comparison), indicating that the prediction accuracy of the platelet performance index evaluation model PEM is very high, and at the same time, it shows that the cloud-like characteristics of platelets are highly correlated with platelet performance.

[0224] The variable X in the correlation coefficient R in Table 3 is the predicted value of the platelet performance index evaluation model PEM, and the variable Y is the true value of the platelet performance index, which is used to indicate the degree of fit of the platelet performance index evaluation model PEM to the true value of the platelet performance index. The value range of the correlation coefficient R is between -1 and 1. The larger the absolute value of the correlation coefficient R, the better the fit of the platelet performance index evaluation model PEM to the true value of the platelet performance index. As can be seen from Table 3, the R values ​​are all above 0.8, and some R values ​​are above 0.9, indicating that the platelet performance index evaluation model PEM has a high degree of fit to the true value of the platelet performance index, and the cloud-like characteristics of platelets are strongly correlated with the platelet performance index. Therefore, the cloud-like characteristic data of platelets can be used to evaluate the performance index of platelets.

[0225] The platelet performance index evaluation method of the present invention can detect the performance index of platelets according to the cloud-like characteristics of platelets, and solve the problems of long time consumption, high cost, large result errors, etc. of traditional platelet performance detection methods. Through the regression prediction of the cloud-like characteristic image video of platelets, the manual operation steps are reduced, and the cloud-like characteristic video data of platelets can be quickly processed and analyzed through the deep neural network, so as to realize real-time or near real-time platelet performance evaluation, significantly shorten the detection time, and improve the detection efficiency; no professional personnel and special reagents and consumables are required, which greatly reduces the detection cost, thereby reducing the medical expenses of patients.

[0226] The following is an embodiment of the device of the present application, which can be used to execute the embodiment of the method of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the method of the present application.

[0227] Figure 8 FIG. 4 shows a structural diagram of a platelet performance index evaluation device based on the cloud-like characteristics of platelets according to an embodiment of the present invention; Figure 8 As shown, the platelet performance index evaluation device may include:

[0228] A collection device 801 is used to collect platelets to be tested;

[0229] An image and video acquisition device 802 is used to acquire cloud-like characteristic data of the platelets to be detected, wherein the cloud-like characteristic data includes cloud-like characteristic image data and / or cloud-like characteristic video data;

[0230] Evaluation module 803, for evaluating the platelet performance index using the cloud-like characteristic data of the platelets to be detected

[0231] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0232] Fig. 9 Schematic diagram of the structure of the electronic device 3 provided in the embodiment of the present application. Fig. 9 As shown, the electronic device 3 of this embodiment includes: a processor 301, a memory 302, and a computer program 303 stored in the memory 302 and executable on the processor 301. When the processor 301 executes the computer program 303, the steps in the above-mentioned method embodiments are implemented. Alternatively, when the processor 301 executes the computer program 303, the functions of the modules / units in the above-mentioned device embodiments are implemented.

[0233] Exemplarily, the computer program 303 may be divided into one or more modules / units, which are stored in the memory 302 and executed by the processor 301 to complete the present application. The one or more modules / units may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program 303 in the electronic device 3.

[0234] The electronic device 3 may be a desktop computer, a notebook, a PDA, a cloud server, or other electronic device. The electronic device 3 may include, but is not limited to, a processor 301 and a memory 302. Those skilled in the art will appreciate that Fig. 9 It is only an example of the electronic device 3 and does not constitute a limitation of the electronic device 3. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0235] The processor 301 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0236] The memory 302 may be an internal storage unit of the electronic device 3, for example, a hard disk or memory of the electronic device 3. The memory 302 may also be an external storage device of the electronic device 3, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 3. Further, the memory 302 may also include both an internal storage unit and an external storage device of the electronic device 3. The memory 302 is used to store computer programs and other programs and data required by the electronic device. The memory 302 may also be used to temporarily store data that has been output or is to be output.

[0237] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0238] The exemplary systems and methods of the present invention have been specifically shown and described with reference to the above embodiments, which are merely examples of the best modes for implementing the present systems and methods. It will be appreciated by those skilled in the art that various changes may be made to the embodiments of the systems and methods described herein when implementing the present systems and / or methods without departing from the spirit and scope of the present invention as defined in the appended claims.

Claims

1. A method for evaluating platelet performance indicators, characterized in that: The method comprises: Collect platelets for testing; Using an image and video device to collect cloud-like characteristic data of the platelets to be detected, wherein the cloud-like characteristic data includes cloud-like characteristic image data and / or cloud-like characteristic video data; The cloud-like characteristic data of the platelets to be tested are used to evaluate platelet performance indicators.

2. The platelet performance index evaluation method according to claim 1, characterized in that: The method of evaluating the platelet performance index by using the cloud-like characteristic data of the platelets to be detected includes: Performing frequency domain analysis and gradient analysis on each frame of the cloud-like characteristic data of the platelets to be detected, and obtaining a comprehensive evaluation index of the cloud-like characteristic data of the platelets to be detected; Recording the duration of the cloud-like characteristic of the platelets to be detected, and obtaining the duration of the cloud-like characteristic data of the platelets to be detected; Obtaining significant features of the cloud-like characteristic data of the platelets to be detected according to the comprehensive evaluation index and duration of the cloud-like characteristic data of the platelets to be detected; The platelet performance index is evaluated based on the correspondence between the significant features of the cloud-like characteristic data of the platelets to be detected and the platelet performance index.

3. The platelet performance index evaluation method according to claim 2, characterized in that: The comprehensive evaluation indexes of the cloud-like characteristic data of the platelets to be detected include spectrum energy ratio, gradient amplitude mean value and stripe direction score.

4. The platelet performance index evaluation method according to claim 2, characterized in that: The method of evaluating the platelet performance index according to the correspondence between the significant features of the cloud-like characteristic data of the platelets to be detected and the platelet performance index comprises: Determining the most significant feature of the cloud-like characteristic data of the platelets to be detected; The quality of the platelet performance index is evaluated based on the correspondence between the most significant feature of the cloud-like characteristic data of the platelets to be detected and the level of the platelet performance index.

5. The platelet performance index evaluation method according to claim 4, characterized in that: The platelet performance indicators are divided into different levels according to the platelet coagulation function.

6. The platelet performance index evaluation method according to claim 1, characterized in that: The method of evaluating the platelet performance index by using the cloud-like characteristic data of the platelets to be detected also includes: The cloud-like characteristic data of the platelets to be detected are input into the trained platelet performance index evaluation model PEM to obtain platelet performance index evaluation data.

7. The platelet performance index evaluation method according to claim 6, characterized in that: The platelet performance index evaluation model PEM is trained based on platelet performance index data and platelet cloud feature data.

8. The platelet performance index evaluation method according to claim 7, characterized in that: The platelet performance index evaluation model PEM is trained based on platelet performance index data and platelet cloud feature data, including: Collect platelets for training; Using an image and video device to collect cloud-like characteristic data of training platelets, the cloud-like characteristic data includes cloud-like characteristic image data and / or cloud-like characteristic video data; Testing the performance index data of training platelets; Preprocessing the cloud-like characteristic data of the training platelets to obtain the cloud-like characteristic preprocessed data of the training platelets that meets the input requirements of the platelet performance index evaluation model PEM; Inputting the cloud-like characteristic preprocessing data of the training platelets into the platelet performance index evaluation model PEM to obtain the training platelet performance index prediction data; Based on the training platelet performance index detection data and prediction data, a platelet performance index evaluation model PEM is trained.

9. The platelet performance index evaluation method according to claim 8, characterized in that: The preprocessing of the cloud-like characteristic data of the training platelets to obtain the cloud-like characteristic preprocessed data of the training platelets that meets the input requirements of the platelet performance index evaluation model PEM includes: Extracting key frame images from the cloud-like feature video data of the training platelets; The key frame images are clustered using the K-means clustering algorithm to obtain the cloud-like characteristic key frame images of the training platelets at different cluster centers. Performing denoising, resolution adjustment, image enhancement and normalization processing on the cloud-like characteristic key frame images of the training platelets at the different categories of cluster centers to obtain the cloud-like characteristic key frame images of the training platelets; The key frame image of the cloudy characteristics of the training platelets is preprocessed data of the cloudy characteristics of the training platelets that meets the input requirements of the platelet performance evaluation model PEM.

10. The platelet performance index evaluation method according to claim 8, characterized in that: The platelet performance index evaluation model PEM includes a convolutional neural network CNN and a recurrent neural network RNN.

11. The platelet performance index evaluation method according to claim 10, characterized in that: The cloud-like characteristic preprocessing data of the training platelets is input into the platelet performance index evaluation model PEM to obtain the training platelet performance index prediction data; Based on the platelet performance index detection data and prediction data for training, the platelet performance index evaluation model PEM is trained, including: P1: Inputting the key frame image of the cloudy feature of the training platelet into the convolutional neural network CNN of the platelet performance index evaluation model PEM to obtain the key frame image features of the cloudy feature of the training platelet; P2: Inputting the key frame image features of the cloud-like features of the training platelets into the recurrent neural network RNN, and outputting the predicted data of the performance index of the training platelets; P3: Calculating the loss value of the platelet performance index evaluation model PEM based on the platelet performance index detection data and prediction data for training; P4: using the loss value of the platelet performance index evaluation model PEM and the gradient back propagation algorithm to update the weight parameters and bias parameters of the platelet performance index evaluation model PEM; P5: Repeat the process of P1-P4 until the preset number of training rounds, and obtain the platelet performance index evaluation model PEM through training.

12. The platelet performance index evaluation method according to claim 1, characterized in that: The platelet performance index data include: adhesion rate, maximum aggregation rate, platelet factor 4 (PF4) concentration, blood clot contraction rate, mouse carotid artery thrombosis time, mouse tail bleeding time, the proportion of platelets transfused into the body, phosphatidylserine (PS) positivity rate, P-selectin (CD62P) positivity rate, and one or more of the average fluorescence intensity of platelet glycoprotein Ibα (CD42b).

13. A platelet performance index evaluation device, characterized in that: The device comprises: A collection device for collecting platelets to be tested; An image and video acquisition device, used for acquiring cloud-like characteristic data of the platelets to be detected, wherein the cloud-like characteristic data includes cloud-like characteristic image data and / or cloud-like characteristic video data; An evaluation module is used to evaluate the platelet performance index using the cloud-like characteristic data of the platelets to be detected.

14. An electronic device, characterized in that: The device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 12 when executing the program.

15. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 12 is implemented.

Images