Image processing method and device for judging hemolysis of suspended red blood cells
By constructing and training the hemolysis detection model, and analyzing blood images using image processing technology, the problem of insufficient objective and low accuracy of the hemolysis detection results of suspended red blood cells is solved, and high accuracy and real-time hemolysis detection is achieved.
Patent Information
- Application Number
- CN202411761181.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-03
- Publication Date
- 2025-03-18
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The detection results of suspended red blood cell hemolysis in the prior art are affected by subjective factors of the operator, resulting in the detection results being inconsistent enough and the accuracy is not high.
The image processing method is used to obtain the blood image set, and the preset hemolysis detection model is trained. The trained model is used to process the collected blood actual image to obtain the hemolysis detection result information. The model includes a feature detection network and a fusion detection network. Through a two-step design of feature detection and fusion detection, it achieves high-accurate hemolytic detection.
It improves the accuracy and real-timeness of hemolysis detection, reduces dependence on operators, and enhances the objectivity and reliability of detection results.
Smart Images
Figure CN119649419A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of blood quality detection and image processing, and in particular to an image processing method and device for judging hemolysis of suspended red blood cells. Background Art
[0002] During the use of blood, such as when the transfusion medicine department of a hospital sends stored blood to the clinic, the quality of the blood must be checked before using the pre-collected and stored blood. An important indicator for judging blood quality is whether hemolysis of suspended red blood cells occurs. The normal static state of suspended red blood cells and the state of hemolysis of red blood cells will cause the blood to present completely different color characteristics. In the current blood quality inspection process before blood use, the color of the plasma layer in the suspended red blood cells in the static state is often observed with the naked eye to judge whether hemolysis of red blood cells occurs. This method is often affected by the subjective factors of the operator, resulting in the judgment result of hemolysis of suspended red blood cells not being objective enough and the accuracy rate is not high. Summary of the invention
[0003] The present invention mainly solves the problem that the current detection of suspended red blood cell hemolysis is often affected by the subjective factors of the operator, resulting in the detection result being not objective and the accuracy being low. The present invention discloses an image processing method and device for judging suspended red blood cell hemolysis.
[0004] In a first aspect of the embodiments of the present application, an image processing method for determining hemolysis of suspended red blood cells is disclosed, comprising:
[0005] S1, obtaining a blood image set; the blood image set includes a plurality of blood images and corresponding label information;
[0006] S2, using the blood image set to train a preset hemolysis detection model to obtain a trained hemolysis detection model;
[0007] S3, using the trained hemolysis detection model, processing the collected actual blood image to obtain hemolysis detection result information; the hemolysis detection result information is used to characterize the red blood cell hemolysis information of the collected actual blood image.
[0008] The trained hemolysis detection model is used to process the collected actual blood image to obtain hemolysis detection result information, including:
[0009] Preprocessing the collected actual blood image to obtain a preprocessed image;
[0010] The pre-processed image is processed using the trained hemolysis detection model to obtain hemolysis detection result information.
[0011] The step of preprocessing the collected actual blood image to obtain a preprocessed image includes:
[0012] Normalizing the collected actual blood image to obtain a normalized image;
[0013] The normalized image is subjected to boundary definition processing to obtain a preprocessed image.
[0014] The hemolysis detection model includes: a feature detection network and a fusion detection network;
[0015] The feature detection network is used to process the blood image to obtain image feature information; the fusion detection network is used to perform hemolysis prediction processing on the image feature information to obtain hemolysis detection result information of the blood image;
[0016] The feature detection network includes a first input module, a first convolution module, a depth-separable convolution module, a first dimension-raising convolution module, a second dimension-raising convolution module, a third dimension-raising convolution module, a fourth dimension-raising convolution module, a second convolution module, a first pooling module, a third convolution module and a first fully connected module;
[0017] The input end of the first input module of the feature detection network is used to receive the blood image; the output end of the first input module of the feature detection network is connected to the input end of the first convolution module of the feature detection network; the output end of the first convolution module of the feature detection network is connected to the input end of the depthwise separable convolution module of the feature detection network; the output end of the depthwise separable convolution module of the feature detection network is connected to the input end of the first dimensionality-raising convolution module of the feature detection network; the output end of the first dimensionality-raising convolution module of the feature detection network is connected to the input end of the second dimensionality-raising convolution module of the feature detection network; the output end of the second dimensionality-raising convolution module of the feature detection network is connected to the input end of the second dimensionality-raising convolution module of the feature detection network. The output end of the third dimensionality-raising convolution module of the feature detection network is connected to the input end of the fourth dimensionality-raising convolution module of the feature detection network; the output end of the fourth dimensionality-raising convolution module of the feature detection network is connected to the input end of the second convolution module of the feature detection network; the output end of the second convolution module of the feature detection network is connected to the input end of the first pooling module of the feature detection network; the output end of the first pooling module of the feature detection network is connected to the input end of the third convolution module of the feature detection network; the output end of the third convolution module of the feature detection network is connected to the input end of the first fully connected module of the feature detection network;
[0018] The fusion detection network includes a second input module, a fourth convolution module, a fifth convolution module, a sixth convolution module, a second pooling module, a seventh convolution module and a second fully connected module;
[0019] The input end of the second input module of the fusion detection network is connected to the output end of the first fully connected module of the feature detection network; the output end of the second input module of the fusion detection network is connected to the input end of the fourth convolution module of the fusion detection network; the output end of the fourth convolution module of the fusion detection network is connected to the input end of the fifth convolution module of the fusion detection network; the output end of the fifth convolution module of the fusion detection network is connected to the input end of the sixth convolution module of the fusion detection network; the output end of the sixth convolution module of the fusion detection network is connected to the input end of the second pooling module of the fusion detection network; the output end of the second pooling module of the fusion detection network is connected to the input end of the seventh convolution module of the fusion detection network; the output end of the seventh convolution module of the fusion detection network is connected to the input end of the second fully connected module of the fusion detection network; the output end of the second fully connected module of the fusion detection network is used to output the predicted hemolysis detection result information of the blood image.
[0020] The first convolution module, the depthwise separable convolution module, the first dimension-raising convolution module, the second dimension-raising convolution module, the third dimension-raising convolution module, the fourth dimension-raising convolution module, the second convolution module, the first pooling module and the third convolution module of the feature detection network are all implemented by N parallel three-dimensional filters; the three-dimensional filters in the front and back adjacent modules in the feature detection network are connected according to the corresponding serial numbers;
[0021] The fourth convolution module, the fifth convolution module, the sixth convolution module, the second pooling module and the seventh convolution module of the fusion detection network are all implemented by N parallel three-dimensional filters; the three-dimensional filters in the front and rear adjacent modules in the fusion detection network are connected according to the corresponding serial numbers.
[0022] The method of using the blood image set to train a preset hemolysis detection model to obtain a trained hemolysis detection model includes:
[0023] Initialize the number of training iterations;
[0024] inputting the blood images in the blood image set as input data into the hemolysis detection model;
[0025] Using the hemolysis detection model, processing the input data to obtain hemolysis detection result information;
[0026] Performing difference calculation processing on the obtained hemolysis test result information and the label information corresponding to the input data to obtain a difference value;
[0027] Determine whether the difference value satisfies a convergence condition, and obtain a first determination result;
[0028] When the first judgment result is no, judging whether the training iteration number value is equal to the training number threshold, and obtaining a second judgment result;
[0029] When the second judgment result is no, determining that the model training state does not meet the training termination condition;
[0030] When the second judgment result is yes, determining that the model training state satisfies the training termination condition;
[0031] When the first judgment result is yes, determining that the model training state satisfies a training termination condition;
[0032] When the model training state does not meet the training termination condition, the parameter updating model is used to update the parameters of the hemolysis detection model, the training iteration number value is increased by 1, and the execution is triggered to input the blood image in the blood image set as input data into the hemolysis detection model;
[0033] When the model training state satisfies the training termination condition, the training process of the hemolysis detection model is completed to obtain a trained hemolysis detection model.
[0034] The parameter updating model is:
[0035]
[0036] θ←θ+v.
[0037] In the formula, x (i) is the i-th blood image in the blood image set, y (i) is the label information of the i-th blood image in the blood image set, is the loss function, v is the parameter update value, θ is the parameter to be updated, η is the initial parameter learning rate, α is the momentum angle parameter, 0≤α≤1, It means to find the partial derivative of variable θ, f(x (i) ; θ) represents the hemolysis detection result information calculated by the hemolysis detection model for the ith blood image in the blood image set, f(·) is the calculation function corresponding to the hemolysis detection model; exp represents the power operation of the constant e; η and α are preset constant values; ω1 and ω2 are preset weighting coefficients.
[0038] In a second aspect of an embodiment of the present invention, an image processing device for determining hemolysis of suspended red blood cells is disclosed, the device comprising:
[0039] A memory storing executable program code;
[0040] a processor coupled to the memory;
[0041] The processor calls the executable program code stored in the memory to execute the image processing method for determining hemolysis of suspended red blood cells.
[0042] According to a third aspect of an embodiment of the present invention, a computer storable medium is disclosed, wherein the computer storable medium stores computer instructions, and when the computer instructions are called by a computer, they are used to execute the image processing method for determining hemolysis of suspended red blood cells.
[0043] According to a fourth aspect of the embodiments of the present invention, an information data processing terminal is disclosed. The information data processing terminal is used to implement the image processing method for determining hemolysis of suspended red blood cells.
[0044] The beneficial effects of the present invention are:
[0045] The present invention constructs a hemolysis detection model and uses historical hemolysis data to train the hemolysis detection model to obtain a hemolysis detection model with a high detection probability, thereby improving the accuracy and real-time performance of the model's hemolysis detection, and solving the problem that the current detection of suspended red blood cell hemolysis is often affected by the operator's subjective factors, resulting in the detection results being not objective and having a low accuracy rate.
[0046] The hemolysis detection model constructed by the present invention adopts a two-step design of feature detection and fusion detection, which ensures that each part can be independently operated and upgraded, thereby improving the flexibility of the model. At the same time, low-complexity modules are used in the modules of the model, ensuring the rapidity and real-time operation of the model.
[0047] The present invention constructs a circulant matrix and uses a digital enhancement method to perform data enhancement processing on the detection results of a single image. By comprehensively utilizing information from different sources, it can provide a more comprehensive data view than a single data, thereby increasing the decision accuracy and reliability of the system. Data from different frames complement each other's deficiencies, reducing the deviations or errors that may be introduced by relying on a single data source. In addition, data enhancement processing can rely on other modes to continue to operate normally when one mode is disturbed or fails, ensuring the continuity and robustness of the system prediction. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] Figure 1 1 is a flow chart for implementing the method of the present invention;
[0049] Figure 2 A schematic diagram of the connection of three-dimensional filters in the front and back adjacent modules in a feature detection network or a fusion detection network. DETAILED DESCRIPTION
[0050] In order to better understand the content of the present invention, an embodiment is given here.
[0051] Figure 1 is a flow chart for implementing the method of the present invention, Figure 2 A schematic diagram of the connection of three-dimensional filters in the front and back adjacent modules in a feature detection network or a fusion detection network.
[0052] In view of the problem that the current detection of suspended red blood cell hemolysis is often affected by the subjective factors of the operator, resulting in the problem that the detection results are not objective and the accuracy is not high, the present invention discloses an image processing method and device for judging suspended red blood cell hemolysis.
[0053] In a first aspect of the embodiments of the present application, an image processing method for determining hemolysis of suspended red blood cells is disclosed, comprising:
[0054] S1, obtaining a blood image set; the blood image set includes a plurality of blood images and corresponding label information; the label information is used to characterize the occurrence information of red blood cell hemolysis of the corresponding blood image;
[0055] S2, using the blood image set to train a preset hemolysis detection model to obtain a trained hemolysis detection model;
[0056] S3, using the trained hemolysis detection model to process the collected actual blood image to obtain hemolysis detection result information; the hemolysis detection result information is used to characterize the red blood cell hemolysis information of the collected actual blood image;
[0057] The value of the information on occurrence of hemolysis of red blood cells is whether hemolysis occurs or not.
[0058] The trained hemolysis detection model is used to process the collected actual blood image to obtain hemolysis detection result information, including:
[0059] Preprocessing the collected actual blood image to obtain a preprocessed image;
[0060] Using the trained hemolysis detection model, the pre-processed image is processed to obtain hemolysis detection result information;
[0061] The step of preprocessing the collected actual blood image to obtain a preprocessed image includes:
[0062] Normalizing the collected actual blood image to obtain a normalized image;
[0063] The normalized image is subjected to boundary definition processing to obtain a preprocessed image.
[0064] The hemolysis detection model includes: a feature detection network and a fusion detection network;
[0065] The feature detection network is used to process the blood image to obtain image feature information; the fusion detection network is used to perform hemolysis prediction processing on the image feature information to obtain hemolysis detection result information of the blood image.
[0066] The feature detection network includes a first input module, a first convolution module, a depth-separable convolution module, a first dimension-raising convolution module, a second dimension-raising convolution module, a third dimension-raising convolution module, a fourth dimension-raising convolution module, a second convolution module, a first pooling module, a third convolution module and a first fully connected module;
[0067] The input end of the first input module of the feature detection network is used to receive the blood image; the output end of the first input module of the feature detection network is connected to the input end of the first convolution module of the feature detection network; the output end of the first convolution module of the feature detection network is connected to the input end of the depthwise separable convolution module of the feature detection network; the output end of the depthwise separable convolution module of the feature detection network is connected to the input end of the first dimensionality-raising convolution module of the feature detection network; the output end of the first dimensionality-raising convolution module of the feature detection network is connected to the input end of the second dimensionality-raising convolution module of the feature detection network; the output end of the second dimensionality-raising convolution module of the feature detection network is connected to the input end of the second dimensionality-raising convolution module of the feature detection network. The output end of the third dimensionality-raising convolution module of the feature detection network is connected to the input end of the fourth dimensionality-raising convolution module of the feature detection network; the output end of the fourth dimensionality-raising convolution module of the feature detection network is connected to the input end of the second convolution module of the feature detection network; the output end of the second convolution module of the feature detection network is connected to the input end of the first pooling module of the feature detection network; the output end of the first pooling module of the feature detection network is connected to the input end of the third convolution module of the feature detection network; the output end of the third convolution module of the feature detection network is connected to the input end of the first fully connected module of the feature detection network;
[0068] The fusion detection network includes a second input module, a fourth convolution module, a fifth convolution module, a sixth convolution module, a second pooling module, a seventh convolution module and a second fully connected module;
[0069] The input end of the second input module of the fusion detection network is connected to the output end of the first fully connected module of the feature detection network; the output end of the second input module of the fusion detection network is connected to the input end of the fourth convolution module of the fusion detection network; the output end of the fourth convolution module of the fusion detection network is connected to the input end of the fifth convolution module of the fusion detection network; the output end of the fifth convolution module of the fusion detection network is connected to the input end of the sixth convolution module of the fusion detection network; the output end of the sixth convolution module of the fusion detection network is connected to the input end of the second pooling module of the fusion detection network; the output end of the second pooling module of the fusion detection network is connected to the input end of the seventh convolution module of the fusion detection network; the output end of the seventh convolution module of the fusion detection network is connected to the input end of the second fully connected module of the fusion detection network; the output end of the second fully connected module of the fusion detection network is used to output the predicted hemolysis detection result information of the blood image.
[0070] The convolution module can be implemented by using 3D multi-channel convolution;
[0071] The depthwise separable convolution module can be implemented by connecting a channel splitting submodule and a single-channel convolution submodule. Specifically, it can be implemented by using the depthwise separable convolution module in the MobileNet network.
[0072] The dimension-raising convolution module can be implemented by using the dimension-raising convolution module in the ResNet network;
[0073] The pooling module can be implemented by using a maximum pooling operation.
[0074] Specifically, when the convolution module, depthwise separable convolution module, dimensionality-increasing convolution module and pooling module are implemented by using multiple three-dimensional filters in parallel, the data characteristics of the matrices of the corresponding three-dimensional filters are different. The data characteristics of the matrices include the mean, variance, eigenvalues, etc. of the matrices.
[0075] The output end of the first fully connected module of the feature detection network is used to output the image feature information.
[0076] The output end of the second fully connected module of the fusion detection network is used to output the hemolysis detection result information of the blood image.
[0077] The first convolution module, the depthwise separable convolution module, the first dimension-raising convolution module, the second dimension-raising convolution module, the third dimension-raising convolution module, the fourth dimension-raising convolution module, the second convolution module, the first pooling module and the third convolution module of the feature detection network are all implemented by N parallel three-dimensional filters; the three-dimensional filters in the front and back adjacent modules in the feature detection network are connected according to the sequence number;
[0078] The fourth convolution module, the fifth convolution module, the sixth convolution module, the second pooling module and the seventh convolution module of the fusion detection network are all implemented by using N parallel three-dimensional filters; the three-dimensional filters in the front and back adjacent modules in the fusion detection network are connected according to the sequence number;
[0079] The three-dimensional filters in the front and rear adjacent modules in the feature detection network or the fusion detection network are connected according to the serial number. Specifically, the first three-dimensional filter of the previous convolution module is connected to the first three-dimensional filter of the next convolution module, the second three-dimensional filter of the previous convolution module is connected to the second three-dimensional filter of the next convolution module, the third three-dimensional filter of the previous convolution module is connected to the third three-dimensional filter of the next convolution module, and specifically, the last three-dimensional filter of the previous convolution module is connected to the last three-dimensional filter of the next convolution module. Taking the connection relationship between the first dimensional up-convolution module and the second dimensional up-convolution module in the feature detection network as an example, assuming that the first dimensional up-convolution module and the second dimensional up-convolution module both include three three-dimensional filters, the connection relationship between the two is as follows: Figure 2 shown.
[0080] The method of using the blood image set to train a preset hemolysis detection model to obtain a trained hemolysis detection model includes:
[0081] Initialize the number of training iterations;
[0082] inputting the blood images in the blood image set as input data into the hemolysis detection model;
[0083] Using the hemolysis detection model, processing the input data to obtain hemolysis detection result information;
[0084] Performing difference calculation processing on the obtained hemolysis test result information and the label information corresponding to the input data to obtain a difference value;
[0085] Determine whether the difference value satisfies a convergence condition, and obtain a first determination result;
[0086] When the first judgment result is no, judging whether the training iteration number value is equal to the training number threshold, and obtaining a second judgment result;
[0087] When the second judgment result is no, determining that the model training state does not meet the training termination condition;
[0088] When the second judgment result is yes, determining that the model training state satisfies the training termination condition;
[0089] When the first judgment result is yes, determining that the model training state satisfies a training termination condition;
[0090] When the model training state does not meet the training termination condition, the parameter updating model is used to update the parameters of the hemolysis detection model, the training iteration number value is increased by 1, and the execution is triggered to input the blood image in the blood image set as input data into the hemolysis detection model;
[0091] When the model training state satisfies the training termination condition, the training process of the hemolysis detection model is completed to obtain a trained hemolysis detection model.
[0092] The parameter updating model is:
[0093]
[0094] θ←θ+v.
[0095] In the formula, x (i) is the i-th blood image in the blood image set, y (i) is the label information of the i-th blood image in the blood image set, is the loss function, v is the parameter update value, θ is the parameter of the module to be updated, η is the initial parameter learning rate, α is the momentum angle parameter, 0≤α≤1, It means to find the partial derivative of variable θ, f(x (i) ; θ) represents the hemolysis detection result information calculated by the hemolysis detection model for the i-th blood image in the training data set, f(·) is the calculation function corresponding to the hemolysis detection model; exp represents the power operation of the constant e; η and α are preset constant values; ω1 and ω2 are preset weighting coefficients.
[0096] The difference value satisfies the convergence condition, which means that the difference value is less than a preset convergence threshold; the difference value does not satisfy the convergence condition, which means that the difference value is not less than the preset convergence threshold.
[0097] The difference calculation process may be implemented using a loss function.
[0098] The loss function may adopt a cross entropy loss function.
[0099] The boundary limitation processing is to determine whether the value of each data of the normalized image is within a preset value range, and delete the data that is not within the preset value range from the normalized image to obtain a preprocessed image.
[0100] The trained hemolysis detection model is used to process the collected actual blood image to obtain hemolysis detection result information, including:
[0101] The actual blood image obtained by the acquisition includes image data of several frames;
[0102] Acquiring current frame image data of the actual blood image;
[0103] Using the trained hemolysis detection model, the image data of the current frame is processed to obtain hemolysis detection result information of N frames subsequent to the current frame;
[0104] Using the trained hemolysis detection model, the image data of the current frame and the previous N frames are processed to obtain N+1 frame hemolysis detection result information;
[0105] A circulant matrix is constructed using the hemolysis detection result information of the current frame and the N frames before and after it; the i-th row of the circulant matrix is the N+1 frame hemolysis detection result information obtained by the image data of the i-1 frame after the current frame and the N frames before the i-1 frame;
[0106] Perform data enhancement processing on the circulatory matrix to obtain the hemolysis detection result information of the current frame data;
[0107] The data enhancement processing includes:
[0108] Extract the sub-diagonal vector of the circulant matrix;
[0109] The sampling sequence of the normalized Gaussian distribution function is used to perform inner product operation with the sub-diagonal vector to obtain the hemolysis detection result information of the current frame data.
[0110] Determine whether the hemolysis detection result information of the current frame data is greater than a preset relative probability threshold to obtain the hemolysis detection result of the current frame data; if greater, the hemolysis detection result of the current frame data is that hemolysis occurs; if not greater, the hemolysis detection result of the current frame data is that hemolysis does not occur.
[0111] Taking the current frame as the 10th frame as an example, the expression of the constructed circulant matrix is:
[0112]
[0113] The numbers in the matrix represent the hemolysis detection result information of the current frame data of the sequence frame, and the data in the matrix represent the relative probability value of hemolysis.
[0114] The present invention constructs a circulant matrix and uses a digital enhancement method to perform data enhancement processing on the detection results of a single image. By comprehensively utilizing information from different sources, it can provide a more comprehensive data view than a single data, thereby increasing the decision accuracy and reliability of the system. Data from different frames complement each other's deficiencies, reducing the deviations or errors that may be introduced by relying on a single data source. In addition, data enhancement processing can rely on other modes to continue to operate normally when one mode is disturbed or fails, ensuring the continuity and robustness of the system prediction.
[0115] In a second aspect of an embodiment of the present invention, an image processing device for determining hemolysis of suspended red blood cells is disclosed, the device comprising:
[0116] A memory storing executable program code;
[0117] a processor coupled to the memory;
[0118] The processor calls the executable program code stored in the memory to execute the image processing method for determining hemolysis of suspended red blood cells.
[0119] According to a third aspect of an embodiment of the present invention, a computer storable medium is disclosed, wherein the computer storable medium stores computer instructions, and when the computer instructions are called by a computer, they are used to execute the image processing method for determining hemolysis of suspended red blood cells.
[0120] According to a fourth aspect of the embodiments of the present invention, an information data processing terminal is disclosed. The information data processing terminal is used to implement the image processing method for determining hemolysis of suspended red blood cells.
[0121] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims
1. An image processing method for determining hemolysis of suspended red blood cells, characterized in that: include: S1, obtain a blood image set; The blood image set includes a plurality of blood images and corresponding label information; S2, using the blood image set to train a preset hemolysis detection model to obtain a trained hemolysis detection model; S3, using the trained hemolysis detection model, processing the collected actual blood image to obtain hemolysis detection result information; The hemolysis detection result information is used to characterize the red blood cell hemolysis information of the collected actual blood image.
2. The image processing method for determining hemolysis of suspended red blood cells according to claim 1, characterized in that: The trained hemolysis detection model is used to process the collected actual blood image to obtain hemolysis detection result information, including: Preprocessing the collected actual blood image to obtain a preprocessed image; The pre-processed image is processed using the trained hemolysis detection model to obtain hemolysis detection result information.
3. The image processing method for determining hemolysis of suspended red blood cells according to claim 2, characterized in that: The step of preprocessing the collected actual blood image to obtain a preprocessed image includes: Normalizing the collected actual blood image to obtain a normalized image; The normalized image is subjected to boundary definition processing to obtain a preprocessed image.
4. The image processing method for determining hemolysis of suspended red blood cells according to claim 3, characterized in that: The hemolysis detection model includes: a feature detection network and a fusion detection network; The feature detection network is used to process the blood image to obtain image feature information; the fusion detection network is used to perform hemolysis prediction processing on the image feature information to obtain hemolysis detection result information of the blood image; The feature detection network includes a first input module, a first convolution module, a depth-separable convolution module, a first dimension-raising convolution module, a second dimension-raising convolution module, a third dimension-raising convolution module, a fourth dimension-raising convolution module, a second convolution module, a first pooling module, a third convolution module and a first fully connected module; The input end of the first input module of the feature detection network is used to receive the blood image; the output end of the first input module of the feature detection network is connected to the input end of the first convolution module of the feature detection network; the output end of the first convolution module of the feature detection network is connected to the input end of the depthwise separable convolution module of the feature detection network; the output end of the depthwise separable convolution module of the feature detection network is connected to the input end of the first dimensionality-raising convolution module of the feature detection network; the output end of the first dimensionality-raising convolution module of the feature detection network is connected to the input end of the second dimensionality-raising convolution module of the feature detection network; the output end of the second dimensionality-raising convolution module of the feature detection network is connected to the input end of the second dimensionality-raising convolution module of the feature detection network. The output end of the third dimensionality-raising convolution module of the feature detection network is connected to the input end of the fourth dimensionality-raising convolution module of the feature detection network; the output end of the fourth dimensionality-raising convolution module of the feature detection network is connected to the input end of the second convolution module of the feature detection network; the output end of the second convolution module of the feature detection network is connected to the input end of the first pooling module of the feature detection network; the output end of the first pooling module of the feature detection network is connected to the input end of the third convolution module of the feature detection network; the output end of the third convolution module of the feature detection network is connected to the input end of the first fully connected module of the feature detection network; The fusion detection network includes a second input module, a fourth convolution module, a fifth convolution module, a sixth convolution module, a second pooling module, a seventh convolution module and a second fully connected module; The input end of the second input module of the fusion detection network is connected to the output end of the first fully connected module of the feature detection network; the output end of the second input module of the fusion detection network is connected to the input end of the fourth convolution module of the fusion detection network; the output end of the fourth convolution module of the fusion detection network is connected to the input end of the fifth convolution module of the fusion detection network; the output end of the fifth convolution module of the fusion detection network is connected to the input end of the sixth convolution module of the fusion detection network; the output end of the sixth convolution module of the fusion detection network is connected to the input end of the second pooling module of the fusion detection network; the output end of the second pooling module of the fusion detection network is connected to the input end of the seventh convolution module of the fusion detection network; the output end of the seventh convolution module of the fusion detection network is connected to the input end of the second fully connected module of the fusion detection network; the output end of the second fully connected module of the fusion detection network is used to output the predicted hemolysis detection result information of the blood image.
5. The image processing method for determining hemolysis of suspended red blood cells according to claim 4, characterized in that: The first convolution module, the depthwise separable convolution module, the first dimension-raising convolution module, the second dimension-raising convolution module, the third dimension-raising convolution module, the fourth dimension-raising convolution module, the second convolution module, the first pooling module and the third convolution module of the feature detection network are all implemented by N parallel three-dimensional filters; the three-dimensional filters in the front and back adjacent modules in the feature detection network are connected according to the corresponding serial numbers; The fourth convolution module, the fifth convolution module, the sixth convolution module, the second pooling module and the seventh convolution module of the fusion detection network are all implemented by N parallel three-dimensional filters; the three-dimensional filters in the front and rear adjacent modules in the fusion detection network are connected according to the corresponding serial numbers.
6. The image processing method for determining hemolysis of suspended red blood cells according to claim 5, characterized in that: The method of using the blood image set to train a preset hemolysis detection model to obtain a trained hemolysis detection model includes: Initialize the number of training iterations; inputting the blood images in the blood image set as input data into the hemolysis detection model; Using the hemolysis detection model, processing the input data to obtain hemolysis detection result information; Performing difference calculation processing on the obtained hemolysis test result information and the label information corresponding to the input data to obtain a difference value; Determine whether the difference value satisfies a convergence condition, and obtain a first determination result; When the first judgment result is no, judging whether the training iteration number value is equal to the training number threshold, and obtaining a second judgment result; When the second judgment result is no, determining that the model training state does not meet the training termination condition; When the second judgment result is yes, determining that the model training state satisfies the training termination condition; When the first judgment result is yes, determining that the model training state satisfies a training termination condition; When the model training state does not meet the training termination condition, the parameter updating model is used to update the parameters of the hemolysis detection model, the training iteration number value is increased by 1, and the execution is triggered to input the blood image in the blood image set as input data into the hemolysis detection model; When the model training state satisfies the training termination condition, the training process of the hemolysis detection model is completed to obtain a trained hemolysis detection model.
7. The image processing method for determining hemolysis of suspended red blood cells according to claim 6, characterized in that: The parameter updating model is: θ←θ+v. In the formula, x (i) is the i-th blood image in the blood image set, y (i) is the label information of the i-th blood image in the blood image set, is the loss function, v is the parameter update value, θ is the parameter to be updated, η is the initial parameter learning rate, α is the momentum angle parameter, 0≤α≤1, It means to find the partial derivative of variable θ, f(x (i) ; θ) represents the hemolysis detection result information calculated by the hemolysis detection model for the ith blood image in the blood image set, f(·) is the calculation function corresponding to the hemolysis detection model; exp represents the power operation of the constant e; η and α are preset constant values; ω1 and ω2 are preset weighting coefficients.
8. An image processing device for determining hemolysis of suspended red blood cells, characterized in that: The device comprises: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the image processing method for determining hemolysis of suspended red blood cells according to any one of claims 1 to 7.
9. A computer storable medium, characterized in that: The computer storable medium stores computer instructions, and when the computer instructions are called by a computer, they are used to execute the image processing method for determining hemolysis of suspended red blood cells according to any one of claims 1 to 7.
10. An information data processing terminal, characterized in that: The information data processing terminal is used to implement the image processing method for determining hemolysis of suspended red blood cells according to any one of claims 1 to 7.