Plasma quality detection method, electronic equipment and storage medium

Abstract

The invention provides a plasma quality detection method, electronic equipment and a storage medium, and the method comprises the steps: carrying out the image cutting of a plasma data set, enabling the cut plasma data set to sequentially enter a backbone network region feature extraction structure which is alternately operated by a convolutional neural network and a full-dimensional dynamic convolution C3 module, and carrying out the multivariate information feature extraction of a sample; carrying out multidimensional multivariate information fusion, up-sampling and C3-ODConv operation on the extracted multivariate information features in a neck region of the YOLOv5 network; and inputting the obtained multi-dimensional vector into a head detection area to complete a plasma data set supervision training task. The improved YOLOv5 deep learning network is used for carrying out quality detection on normal, hemolysis and fat plasma, errors caused by manual naked eye classification can be avoided, the accuracy and consistency of automatic plasma quality detection are improved, and technical support is provided for in-vitro diagnosis of medical instruments.

Description

Technical Field

[0001] The present invention relates to the technical field of in vitro diagnostic medical devices, and in particular to a plasma quality detection method, electronic equipment and storage medium. Background Art

[0002] As a key component of blood, plasma maintains the physiological functions of blood together with red blood cells and platelets. It is mainly composed of water and plasma proteins, providing an important basic raw material for the production of blood products. However, during the preparation and use of plasma, hemolysis may introduce substances such as hemoglobin into plasma. When the hemoglobin content in plasma exceeds a certain threshold, it may cause a series of adverse reactions, thus affecting the quality and safety of plasma. Therefore, accurate detection of the hemoglobin content in plasma is particularly important. At present, blood centers mainly use manual comparison of plasma bags with standard colorimetric cards to judge the eligibility of plasma. Although this method is simple and easy, it is easily affected by external factors such as the lighting of the processing environment, resulting in deviations in the judgment results. In addition, manual judgment is also affected by multiple factors such as the experience, subjective feelings, and personal status of different staff members at the time, resulting in greater uncertainty in the judgment results. In order to improve the accuracy and reliability of plasma quality control, it is necessary to introduce more advanced detection methods and technologies, such as automated instrument analysis, to achieve accurate and rapid determination of the hemoglobin content in plasma. This will help ensure the safety and effectiveness of plasma products and provide more reliable guarantees for clinical treatment and patient rehabilitation. Summary of the invention

[0003] In order to achieve the above-mentioned purpose and other advantages of the present invention, the first object of the present invention is to provide a method for detecting plasma quality, comprising the following steps:

[0004] After image cropping, the plasma dataset enters the backbone network regional feature extraction structure that alternately runs the convolutional neural network and the full-dimensional dynamic convolution C3 module to extract multivariate information features from the samples;

[0005] The extracted multi-information features are subjected to multi-dimensional multi-information fusion, upsampling and C3-ODConv operations in the neck area of ​​the deep learning network;

[0006] The obtained multi-dimensional vector is input into the head detection area to complete the supervised training task of the plasma dataset.

[0007] Furthermore, the plasma dataset, after image cropping, sequentially enters the backbone network regional feature extraction structure that alternately operates the convolutional neural network and the full-dimensional dynamic convolution C3 module, and further includes the following steps:

[0008] In the regional feature extraction structure of the backbone network, the information is pooled in this region through the spatial pyramid separable kernel attention module.

[0009] Furthermore, the plasma dataset, after image cropping, sequentially enters the backbone network regional feature extraction structure that alternately operates the convolutional neural network and the full-dimensional dynamic convolution C3 module, and further includes the following steps:

[0010] The information is transmitted to the last full-dimensional dynamic convolution C3 module in the backbone network regional feature extraction structure through the spatial pyramid separable kernel attention module.

[0011] Furthermore, the full-dimensional dynamic convolution C3 module utilizes a multi-dimensional attention mechanism and a parallel strategy to learn the attention of the convolution kernel along the four dimensions of the kernel space in any convolution layer.

[0012] Furthermore, the convolution kernel W of the full-dimensional dynamic convolution C3 module i In the four dimensions of the kernel space α si , α ci , α fi , α wi It performs full-dimensional attention in parallel, assigning different attention values ​​to the spatial position, input channel, output channel and the parameters of the overall convolution kernel.

[0013] Furthermore, the spatial pyramid separable kernel attention module includes a spatial pyramid fast pooling module and a separable convolutional attention module, the spatial pyramid fast pooling module is used to perform pooling operations on different spatial scales, and the separable convolutional attention module is used to decompose the 2D convolution kernel of the convolution layer into a series of 1D convolution kernels.

[0014] Furthermore, the deep learning network is an improved YOLOv5 deep learning network.

[0015] The second object of the present invention is to provide an electronic device, comprising: a memory on which a program code is stored; a processor connected to the memory and implementing the above method when the program code is executed by the processor.

[0016] A third object of the present invention is to provide a computer-readable storage medium having program instructions stored thereon, wherein the program instructions implement the above method when executed.

[0017] Compared with the prior art, the present invention has the following beneficial effects:

[0018] The present invention provides a plasma quality detection method based on a full-dimensional dynamic convolutional network and spatially separated kernel attention. The quality of normal, hemolytic and fatty plasma is detected by an improved YOLOv5 deep learning network, which can avoid the errors caused by manual classification by the naked eye, improve the accuracy and consistency of automated detection of plasma quality, and provide technical support for in vitro diagnostic medical devices.

[0019] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention and implement it according to the contents of the specification, the following is a detailed description of the preferred embodiments of the present invention in conjunction with the accompanying drawings. The specific implementation of the present invention is given in detail by the following embodiments and their accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0021] Figure 1 This is a flow chart of the plasma quality detection method of Example 1;

[0022] Figure 2 This is a schematic diagram of the main network structure of Example 1;

[0023] Figure 3 Schematic diagram of a full-dimensional dynamic convolutional network of Example 1;

[0024] Figure 4 Schematic diagram of the spatial pyramid separable kernel attention module of Example 1;

[0025] Figure 5 is a schematic diagram of an electronic device of Example 2;

[0026] Figure 6 This is a schematic diagram of the storage medium of Example 3. DETAILED DESCRIPTION

[0027] The present invention is further described below in conjunction with the accompanying drawings and specific implementation methods. It should be noted that, under the premise of no conflict, the various embodiments or technical features described below can be arbitrarily combined to form a new embodiment.

[0028] Example 1

[0029] A plasma quality detection method can effectively extract multivariate information features from plasma images, improve detection accuracy, and provide strong support for clinical diagnosis and medical testing. Figure 1 , Figure 2 As shown, the method comprises the following steps:

[0030] S1, plasma datasets are cropped (Focus) and then enter the backbone network area (Backbone) feature extraction structure that alternately runs the convolutional neural network (Conv) and the full-dimensional dynamic convolution C3 module to extract multivariate information features from the samples;

[0031] In some embodiments, the plasma dataset is cropped and sequentially entered into the backbone network region feature extraction structure with the convolutional neural network and the full-dimensional dynamic convolution C3 module running alternately, and the steps further include the following steps:

[0032] In the regional feature extraction structure of the backbone network, the information is pooled in this region through the spatial pyramid separable kernel attention (SPPF-LSKA) module. The pooling operation in deep learning has the advantages of reducing information redundancy, improving the scale invariance and rotation invariance of the model, reducing the amount of model calculation, and preventing overfitting.

[0033] In this embodiment, the spatial pyramid separable kernel attention module includes a spatial pyramid fast pooling (Spatial Pyramid Pooling Fast, SPPF) module and a separable convolutional attention (Large Separable Kernel Attention, LSKA) module. The spatial pyramid fast pooling module can effectively solve the image distortion problem caused by image region cropping and scaling. By performing pooling operations on different spatial scales, it allows the network to better understand the features of different scales in the input image, which not only improves the representation ability of the network, but also avoids repeated feature extraction, thereby increasing the speed of generating candidate boxes. The separable convolutional attention module is used to decompose the 2D convolution kernel of the convolutional layer into a series of 1D convolution kernels, which improves the network's attention to different spatial positions. This network uses larger convolution kernels in the attention module to capture richer contextual information. Figure 4 As shown in the figure, the combination of SPPF and LSKA can significantly improve the performance of convolutional neural networks in image recognition tasks, avoid image distortion and repeated feature extraction problems, and improve the network's attention to different scales and spatial locations.

[0034] The information is transmitted to the last full-dimensional dynamic convolution C3 module in the backbone network regional feature extraction structure through the spatial pyramid separable kernel attention module.

[0035] Conventional convolution has only one static convolution kernel and is independent of the input sample. In contrast, dynamic convolution linearly weights multiple convolution kernels, and the weighted value is related to the input, and the convolution operation can be adaptively adjusted according to the input content. Omni-Dimensional Dynamic Convolution (ODConv) uses a multi-dimensional attention mechanism and parallel strategy to learn the attention of the convolution kernel along the four dimensions of the kernel space in any convolution layer. This mechanism allows ODConv to focus more accurately on important features in the input data, thereby improving the representation ability and accuracy of the model.

[0036] like Figure 3 As shown, the convolution kernel W of the full-dimensional dynamic convolution C3 module i In the four dimensions of the kernel space α si , α ci , α fi , α wi By giving different attention values ​​to the spatial position, input channel, output channel, and parameters of the overall convolution kernel, progressive convolution can make the convolution operation different in each dimension. This difference can better adapt to different input data and capture richer contextual information. By multiplying different attention values ​​on dimensions such as position, channel, filter, and kernel, progressive convolution can focus more accurately on important features in the input data, thereby providing better performance. By giving different attention values ​​to convolution parameters of different dimensions, progressive convolution can become a powerful tool to handle a variety of complex deep learning tasks.

[0037] In some embodiments, the deep learning network is an improved YOLOv5 deep learning network, and the improved YOLOv5 deep learning network architecture is as follows: Figure 2 As shown in the figure, full-dimensional dynamic convolution and spatial pyramid separable kernel attention provide a more powerful feature extraction vision and capability for improving the YOLOv5 network, and multi-feature information is used for iterative feature learning, while effectively overcoming the distortion problem caused by image cropping and scaling.

[0038] Keeping the other algorithm structures of the neck region (Neck) of the Yolov5 deep learning network unchanged, S2, the multi-information features extracted by the feature extraction structure of the backbone network region (Backbone) are subjected to multi-dimensional multi-information fusion, upsampling and C3-ODConv operations in the neck region of the deep learning network.

[0039] S3. Input the multidimensional vector obtained from the neck area (Neck) into the head detection area (Head) to complete the supervised training task of the plasma dataset.

[0040] Medical detection technology based on deep learning has made significant progress, especially the application of convolutional neural networks and multivariate information fusion, which greatly improves the accuracy of image recognition. The present invention proposes a plasma quality detection method based on full-dimensional dynamic convolution spatially separable kernel attention and heterogeneous information fusion reparameterization. First, multi-layer spatial feature information is obtained from the backbone network through full-dimensional dynamic convolution; secondly, the front-layer feature information is vectorized and rearranged through spatially separable kernel attention to form a more representative feature vector; then, the feature information obtained by the backbone network for the first two times is input into the neck area for heterogeneous information fusion. This fusion method can make full use of the complementarity between different features; then, a continuously reparameterized convolutional neural network is introduced into the neck area, and feature extraction is performed again and input into the head detection area; finally, the effectiveness and accuracy of the method proposed in the present invention are verified on the plasma data set. The results show that the plasma detection rate of the improved YOLOv5 network, which combines the full-dimensional dynamic convolutional spatially separable kernel attention of the backbone network and the heterogeneous information fusion reparameterization of the neck area, is 96.8%. It is able to acquire high spatial mapping features in a shorter training batch and improve the average recognition accuracy of plasma of different qualities.

[0041] Example 2

[0042] An electronic device, such as Figure 5 As shown, it includes: a memory on which program codes are stored; a processor connected to the memory, and when the program codes are executed by the processor, a method for detecting plasma quality is implemented. For a detailed description of the method, reference can be made to the corresponding description in the above method embodiment, which will not be repeated here.

[0043] Example 3

[0044] A computer readable storage medium such as Figure 6 As shown, program instructions are stored thereon, and when the program instructions are executed, a method for detecting plasma quality is implemented. For a detailed description of the method, reference may be made to the corresponding description in the above method embodiment, and no further description will be given here.

[0045] The number of devices and processing scales described here are used to simplify the description of the present invention. Applications, modifications and variations of the present invention will be obvious to those skilled in the art.

[0046] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and implementation modes. They can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

[0047] The apparatus, electronic device, non-volatile computer storage medium and method provided in the embodiments of this specification correspond to each other, and therefore, the apparatus, electronic device and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, electronic device and non-volatile computer storage medium will not be repeated here.

[0048] Those skilled in the art also know that, in addition to implementing the controller in a purely computer-readable program code, the controller can be made to implement the same function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, such a controller can be considered as a hardware component, and the devices for implementing various functions included therein can also be considered as structures within the hardware component. Or even, the devices for implementing various functions can be considered as both software units for implementing the method and structures within the hardware component.

[0049] The systems, devices or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. For the convenience of description, the above devices are described separately by functions in various units. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in the same or more software and / or hardware.

[0050] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification may be in the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the embodiments of this specification may be in the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0051] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0052] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0053] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0054] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0055] The specification may be described in the general context of computer-executable instructions executed by a computer, such as program units. Generally, program units include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communications network. In a distributed computing environment, program units may be located in local and remote computer storage media, including storage devices.

[0056] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0057] The above description is only an embodiment of this specification and is not intended to limit one or more embodiments of this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included in the scope of the claims of one or more embodiments of this specification.

Claims

1. A method for detecting plasma quality, characterized in that: The following steps are involved: After image cropping, the plasma dataset enters the backbone network regional feature extraction structure that alternately runs the convolutional neural network and the full-dimensional dynamic convolution C3 module to extract multivariate information features from the samples; The extracted multi-information features are subjected to multi-dimensional multi-information fusion, upsampling and C3-ODConv operations in the neck area of ​​the deep learning network; The obtained multi-dimensional vector is input into the head detection area to complete the supervised training task of the plasma dataset.

2. A method for detecting plasma quality according to claim 1, characterized in that: The plasma dataset is cropped and sequentially entered into the backbone network regional feature extraction structure with the convolutional neural network and the full-dimensional dynamic convolution C3 module running alternately, and the steps also include the following steps: In the regional feature extraction structure of the backbone network, the information is pooled in this region through the spatial pyramid separable kernel attention module.

3. A method for detecting plasma quality according to claim 2, characterized in that: The plasma dataset is cropped and sequentially entered into the backbone network regional feature extraction structure with the convolutional neural network and the full-dimensional dynamic convolution C3 module running alternately, and the steps also include the following steps: The information is transmitted to the last full-dimensional dynamic convolution C3 module in the backbone network regional feature extraction structure through the spatial pyramid separable kernel attention module.

4. A method for detecting plasma quality according to claim 3, characterized in that: The fully dimensional dynamic convolution C3 module utilizes a multi-dimensional attention mechanism and a parallel strategy to learn the attention of the convolution kernel along the four dimensions of the kernel space in any convolutional layer.

5. A method for detecting plasma quality according to claim 4, characterized in that: The convolution kernel W of the full-dimensional dynamic convolution C3 module i In the four dimensions of the kernel space α si , α ci , α fi , α wi It performs full-dimensional attention in parallel, assigning different attention values ​​to the spatial position, input channel, output channel and the parameters of the overall convolution kernel.

6. A method for detecting plasma quality according to claim 2, characterized in that: The spatial pyramid separable kernel attention module includes a spatial pyramid fast pooling module and a separable convolution attention module. The spatial pyramid fast pooling module is used to perform pooling operations on different spatial scales, and the separable convolution attention module is used to decompose the 2D convolution kernel of the convolution layer into a series of 1D convolution kernels.

7. A method for detecting plasma quality according to claim 1, characterized in that: The deep learning network is an improved YOLOv5 deep learning network.

8. An electronic device, characterized in that: include: A memory having program code stored therein; A processor is connected to the memory, and when the program code is executed by the processor, the method according to any one of claims 1 to 7 is implemented.

9. A computer-readable storage medium, characterized in that: Program instructions are stored thereon, and when the program instructions are executed, the method according to any one of claims 1 to 7 is implemented.

Images