Artery positioning detection method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides an artery positioning detection method and device, electronic equipment and a storage medium, and the method comprises the steps: iteratively executing at least one round of the following steps in a process of continuously moving an image collection device at a target part until the integrity degree meets a preset condition: obtaining an image sequence collected by the image collection device in the round, stacking the image sequence into a stacked image, inputting the stacked image into a positioning network, and obtaining an image state output by the positioning network; and taking the last frame of image in the image sequence obtained when the integrity degree meets a preset condition as a target image, detecting the artery blood vessel in the target image, and determining a detection result indicating the health risk of the artery blood vessel. And identifying the image to assist the user in positioning the artery blood vessel, and detecting the artery blood vessel when the integrity degree meets a preset condition.

Description

Technical Field

[0001] The present disclosure relates to the technical field of medical image analysis, and in particular to an artery positioning detection method, device, electronic device and storage medium. Background Art

[0002] With the aging of the population and the improvement of health awareness, the number of deaths and serious cases caused by cardiovascular and cerebrovascular diseases has increased rapidly worldwide, and the demand for portable medical devices and medical imaging technology has continued to grow.

[0003] Currently, imaging devices are usually combined with artificial intelligence algorithms to acquire, analyze and process medical images to provide clinicians with more accurate and comprehensive diagnostic information, which helps in early disease detection and treatment monitoring.

[0004] Imaging equipment has a high operating threshold, and the application of artificial intelligence algorithms is mainly focused on diagnostic functions, which rely on the medical expertise and skills of operators to ensure the quality and accuracy of acquired images, as well as the accuracy of image interpretation for diagnosis. Therefore, the above technologies are mainly used in professional medical environments and are not suitable for users without a medical background. Summary of the invention

[0005] The embodiments of the present disclosure provide an artery positioning detection method, device, electronic device and storage medium, which are used to solve the technical problem in the prior art that users without medical background have difficulty in using imaging equipment.

[0006] According to one aspect of an embodiment of the present disclosure, a method for detecting artery location is provided, comprising: In the process of continuously moving the image acquisition device at the target part, iteratively performing at least one round of the following steps until the degree of integrity meets the preset conditions: obtaining an image sequence acquired by the image acquisition device in this round, stacking the image sequence into a stacked image, inputting the stacked image into a positioning network, and obtaining an image state output by the positioning network, wherein the image state represents the degree of integrity of the vessel wall of the arterial vessel identified according to the corresponding stacked image; The last frame of the image sequence obtained when the degree of completeness meets the preset conditions is used as the target image, the arterial blood vessels in the target image are detected, and the detection results are determined. The detection results are used to indicate the health risks of the arterial blood vessels.

[0007] According to another aspect of the embodiments of the present disclosure, there is provided an artery positioning detection device, comprising: The positioning and identification module is used to iteratively perform at least one round of the following steps in the process of continuously moving the image acquisition device at the target part until the degree of integrity meets the preset conditions: obtaining an image sequence acquired by the image acquisition device in this round, stacking the image sequence into a stacked image, inputting the stacked image into the positioning network, and obtaining an image state output by the positioning network, wherein the image state represents the degree of integrity of the vascular wall of the arterial blood vessel identified according to the corresponding stacked image; The artery detection module is used to use the last frame image in the image sequence obtained when the completeness meets the preset conditions as the target image, detect the arterial blood vessels in the target image, and determine the detection result, which is used to indicate the health risk of the arterial blood vessels.

[0008] According to another aspect of an embodiment of the present disclosure, an electronic device is provided. The electronic device includes a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the steps of the method provided in any of the above embodiments.

[0009] According to another aspect of the embodiments of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method provided in any of the above embodiments are implemented.

[0010] According to one aspect of an embodiment of the present disclosure, a computer program product is provided, including a computer program, and the computer program is executed by a processor to perform the steps of the method provided in any of the above embodiments.

[0011] The technical solution provided by the embodiments of the present disclosure has the following beneficial effects: By pre-training the positioning network, the images collected by the image acquisition device operated by the user are analyzed and the output image status is identified in real time. The image status is used to guide the user to adjust the position of the image acquisition device. When it is determined that the completeness of the image status indication meets the preset conditions, the last frame of the image in the obtained image sequence is used as the target image for detecting the arterial blood vessels. This solves the problem that users without a medical background find it difficult to accurately locate the position of the arterial blood vessels, lowers the threshold for using the image acquisition device, and after positioning, the arterial blood vessels in the acquired target image can be detected to determine the detection results for indicating the health risks of the arterial blood vessels, effectively reducing the complexity of user operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the drawings required for describing the embodiments of the present disclosure are briefly introduced below.

[0013] Figure 1 A schematic diagram of a flow chart of an artery positioning detection method provided in an embodiment of the present disclosure; Figure 2 A schematic diagram of the steps of an artery positioning detection method provided by an embodiment of the present disclosure; Figure 3 A schematic diagram of a carotid artery ultrasound image provided by an embodiment of the present disclosure; Figure 4 A schematic diagram of the structure of an artery positioning detection device provided in an embodiment of the present disclosure; Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0014] The embodiments of the present disclosure are described below in conjunction with the drawings in the present disclosure. It should be understood that the implementation methods described below in conjunction with the drawings are exemplary descriptions for explaining the technical solutions of the embodiments of the present disclosure and do not constitute a limitation on the technical solutions of the embodiments of the present disclosure.

[0015] It will be understood by those skilled in the art that, unless specifically stated, the singular forms "one", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the terms "including" and "comprising" used in the embodiments of the present disclosure refer to that the corresponding features can be implemented as the presented features, information, data, steps, operations, elements and / or components, but do not exclude the implementation as other features, information, data, steps, operations, elements, components and / or combinations thereof supported by the technical field. It should be understood that when we say that an element is "connected" or "coupled" to another element, the one element may be directly connected or coupled to the other element, or it may refer to that the one element and the other element establish a connection relationship through an intermediate element. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein indicates at least one of the items defined by the term, for example, "A and / or B" may be implemented as "A", or as "B", or as "A and B".

[0016] In order to make the objectives, technical solutions and advantages of the present disclosure more clear, the embodiments of the present disclosure will be further described in detail below with reference to the accompanying drawings.

[0017] The following describes several exemplary embodiments to illustrate the technical solutions of the embodiments of the present disclosure and the technical effects produced by the technical solutions of the present disclosure. It should be noted that the following embodiments can refer to, draw on or combine with each other, and the same terms, similar features and similar implementation steps in different embodiments will not be described repeatedly.

[0018] It can be understood that in the artery positioning detection method provided by the embodiment of the present disclosure, any method step can be executed by the terminal and / or the server, and all the steps in the method can be executed independently by the terminal or the server, or jointly by the terminal and the server.

[0019] The server may be an independent physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides cloud computing services. The terminal may be a smart phone, tablet computer, laptop computer, desktop computer, intelligent voice interaction device (such as smart speaker), wearable electronic device (such as smart watch), vehicle-mounted terminal, smart home appliance (such as smart TV), AR / VR device, etc., but is not limited thereto.

[0020] The following describes the embodiments of the present disclosure using a server as an execution subject; however, this does not constitute a limitation to the embodiments of the present disclosure.

[0021] Figure 1 A flow chart of an artery location detection method provided by an embodiment of the present disclosure is shown in FIG. Figure 1 As shown, the technical solution provided by the embodiment of the present disclosure includes the following steps: Step S101, in the process of continuously moving the image acquisition device at the target part, iteratively execute at least one round of the following steps until the completeness meets the preset conditions: The image sequence collected by the image acquisition device in this round is obtained, the image sequence is stacked into a stacked image, the stacked image is input into the positioning network, and the image state output by the positioning network is obtained.

[0022] The image status indicates the degree of integrity of the vessel wall of the arterial vessel identified according to the corresponding stacked image.

[0023] Specifically, the artery location detection method provided in the embodiment of the present disclosure is used to help users without medical background to quickly locate arteries and determine detection results. Specifically, the method can be divided into two parts: location identification and artery detection.

[0024] Among them, positioning and identification means that the user can continuously collect images at the target site through the image acquisition device, locate and identify the artery based on the collected images, determine whether the identified images meet the detection requirements, and use the images that meet the detection requirements as the correctly positioned images.

[0025] It is understood that the target part refers to the relevant part of the artery to be detected, and the target part can be determined according to the actual needs of the artery to be detected. The image refers to a medical image that can clearly show the blood vessels, such as an ultrasound image (Ultrasound Imaging), etc. The image acquisition device is a device required to acquire the corresponding type of image (for example, an ultrasound probe), and its specific type can be determined according to actual needs.

[0026] Taking the case where the artery to be detected is the carotid artery and the type of image is an ultrasonic image as an example, the artery positioning detection method provided by the embodiment of the present disclosure is described in detail. Since the artery to be detected is the carotid artery, it can be determined that the target part is the neck area. Since the user may not have a professional medical background, it is difficult to accurately and directly locate the position of the carotid artery. By moving the image acquisition device, according to the image state output by the positioning network, the positioning accuracy of the current image can be fed back in real time to determine whether the accurate positioning of the carotid artery is completed. If not, the image acquisition device is continuously moved to help the user quickly find the position of the carotid artery. It can be understood that if the user can accurately locate the carotid artery, only one round of the following steps needs to be completed.

[0027] In step S101, while the image acquisition device is continuously moved at the target site, at least one round of the following steps is iteratively performed until the degree of completeness can be determined to meet preset conditions based on the image state output by the positioning network, wherein meeting the preset conditions means that the acquired image can meet the relevant requirements for arterial blood vessel detection, and the arterial blood vessel wall can be clearly and completely displayed in the image.

[0028] The iterative execution step includes: obtaining an image sequence collected by the image acquisition device in this round, stacking each image frame in the image sequence, and obtaining a stacked image. The stacked image contains dynamic information due to the continuity of the acquisition, which can reflect the arterial pulsation, assist in arterial positioning, and improve the accuracy of positioning. The image sequence includes multiple consecutive frames (for example, 3 consecutive frames) of images, and the specific number can be determined according to actual needs.

[0029] The stacked image is input into a pre-trained positioning network to obtain the image state output by the positioning network. It can be understood that the positioning network is an image recognition and classification network, which can determine and output the image state that matches the current image from multiple image states through image recognition. The image state represents the integrity of the vascular wall of the arterial blood vessel identified according to the corresponding stacked image, and can be used to reflect whether the artery is accurately positioned in the acquired image.

[0030] Among them, the positioning network uses sample stacking images as training samples, and the sample image states corresponding to the sample stacking images are obtained by label training. The specific network model structure and training method can be determined according to actual needs.

[0031] For example, since the carotid artery is on the side of the neck, the user can use the Adam's apple as the initial position of the image acquisition device and move the image acquisition device toward the side of the neck. During the movement, the image acquisition device will continue to acquire images and output image status. When the degree of completeness indicated by the image status does not meet the preset conditions, the user is prompted to continue moving the image acquisition device until it is determined that the degree of completeness indicated by the image status meets the preset conditions, and the user is instructed to stop moving.

[0032] It is understandable that the specific mode of moving the acquisition device can be continuous movement or intermittent movement, etc., and the image acquisition device can acquire images in a continuous manner or by acquiring multiple frames of images at intervals, etc., and the embodiments of the present disclosure do not limit this.

[0033] For example, if the user moves the image acquisition device in a manner of continuously moving it against the skin, the image acquisition device may continuously acquire multiple frames of images at a preset time interval; if the user moves the image acquisition device in a manner of moving it against the skin and then away from the skin, and moves to the next position and again against the skin, the image acquisition device may continuously acquire multiple frames of images at a preset time interval only after it is against the skin, and remain on standby without acquiring images when it leaves the skin.

[0034] Step S102, taking the last frame of the image sequence obtained when the completeness meets the preset condition as the target image, detecting the arterial blood vessels in the target image, and determining the detection result.

[0035] Among them, the test results are used to indicate the health risks of arterial blood vessels.

[0036] Specifically, in step S102, the last frame of the image sequence obtained when the completeness determined in step S101 meets the preset conditions is used as the target image, and the arterial blood vessels in the target image are detected to determine the detection result, which is used to indicate the health risk of the arterial blood vessels. After obtaining the detection result, the detection result can be fed back to the user to give a health risk indication.

[0037] It is understandable that the content of the arterial blood vessel detection may include detecting the thickness of the arterial blood vessel wall, detecting whether there are plaques and calcification foci, etc. The corresponding detection method can be determined according to actual needs.

[0038] The technical solution provided by the embodiment of the present disclosure analyzes the images collected by the image acquisition device operated by the user through pre-training of the positioning network, identifies the output image status in real time, and guides the user to adjust the position of the image acquisition device based on the image status. When it is determined that the completeness of the image status indication meets the preset conditions, the last frame of the image in the obtained image sequence is used as the target image for detecting the arterial blood vessels, thereby solving the problem that users without a medical background find it difficult to accurately locate the position of the arterial blood vessels, lowering the threshold for using the image acquisition device, and after positioning, the arterial blood vessels in the acquired target image can be detected to determine the detection results for indicating the health risks of the arterial blood vessels, thereby effectively reducing the complexity of user operations.

[0039] The method provided by the embodiment of the present disclosure can effectively assist users in operation and automatically realize the diagnosis of arterial blood vessels, which can reduce the complexity of user operation and can be widely used by users with or without medical background, effectively reducing operation time and improving diagnosis efficiency. At the same time, it has low hardware requirements, fast response speed, high accuracy, can be deployed in the cloud, mobile terminal and computer terminal, and can be applied to the field of image-assisted diagnosis of blood vessels in the carotid artery, heart, abdomen and other parts.

[0040] In a possible implementation, the image status includes correct positioning and incorrect positioning; the type of incorrect positioning includes any one of close to correct positioning and wrong positioning; Correct positioning means that the complete vessel wall of the arterial blood vessel is identified according to the corresponding stacked images, nearly correct positioning means that part of the vessel wall of the arterial blood vessel is identified according to the corresponding stacked images, and incorrect positioning means that the vessel wall of the arterial blood vessel cannot be identified according to the corresponding stacked images; The localization network is generated in the following way: Acquire a sample stacked image set, where the sample stacked image set includes at least one sample stacked image; Constructing multiple batches of sample subsets according to the sample stacked image set, each sample subset including multiple sample stacked images and a sample image state corresponding to each sample stacked image; According to the sample subsets of each batch, the initial positioning network is trained in batches with the minimization of the loss function as the optimization goal until the preset training stop condition is met to obtain the positioning network; Among them, the loss function is a multivariate cross entropy loss function that describes the difference between the predicted image state of the positioning network corresponding to the sample stacked image and the sample image state.

[0041] Specifically, in order to better assist users in locating arterial blood vessels, the image status in the embodiment of the present disclosure includes correct positioning and incorrect positioning, wherein the type of incorrect positioning includes any one of correct positioning and wrong positioning, and incorrect positioning is divided into a finer granularity.

[0042] Correspondingly, when pre-training the positioning network, the labels of the training samples include three types of image states: correct positioning, nearly correct positioning, and incorrect positioning.

[0043] Among them, correct positioning means that the complete vessel wall of the arterial blood vessel is identified according to the corresponding stacked image, that is, the proximal end and the distal end of the vessel wall of the arterial blood vessel can be completely and clearly seen in the image. Nearly correct positioning means that part of the vessel wall of the arterial blood vessel is identified according to the corresponding stacked image, that is, the proximal end or the distal end of the vessel wall of part of the arterial blood vessel is not visible or not clearly presented in the image. Incorrect positioning means that the vessel wall of the arterial blood vessel cannot be identified according to the corresponding stacked image, that is, the arterial blood vessel is not included in the image.

[0044] When training the positioning network, it is necessary to construct a sample stacking image set, which includes at least one sample stacking image. It can be understood that the sample stacking image is obtained by stacking each sample image in the sample image sequence, and each sample image refers to a continuous image obtained by collecting videos of target parts of people of different age groups and different body mass indexes through an image acquisition device.

[0045] In the embodiment of the present disclosure, the positioning network adopts a batch training method, so multiple batches of sample subsets are constructed based on the sample stacked images, each sample subset includes multiple sample stacked images and the sample image state corresponding to each sample stacked image. For each sample stacked image, its annotated label (sample image state) is manually annotated by the observer based on the sample image.

[0046] After obtaining multiple sample subsets, the initial positioning network is trained in batches according to the sample subsets of each batch. In each training, the multivariate cross entropy loss function that describes the difference between the predicted image state of the positioning network corresponding to the sample stacked image and the sample image state is used as the loss function during training. With minimizing the loss function as the optimization goal, the parameters of the positioning network are updated until the preset training stop condition is met to obtain the positioning network.

[0047] It is understandable that the training stop condition may be that the number of training times reaches a preset number or the positioning network converges, etc., which can be determined according to actual needs.

[0048] Figure 2 A schematic diagram of the steps of an artery positioning detection method provided by an embodiment of the present disclosure is shown in FIG. Figure 2As shown, the image acquisition device is an ultrasonic probe. When the trained positioning network is applied, when the user moves the probe, the probe acquires images and obtains a stacked image based on the acquired images to input the positioning network. The positioning network feeds back to the user whether the probe is in the correct position based on the output image status (correct positioning, close to correct positioning or incorrect positioning).

[0049] If it is not in the correct position, the user is fed back whether the probe is close to the correct position, and based on whether it is close to the correct position, the user is instructed whether the movement needs to be slight, until it is determined that the probe is in the correct position (the image status is correctly positioned). The last frame of the image sequence acquired at this time is used as the target image, and the detection result is determined based on the detection of the target image.

[0050] The technical solution provided by the embodiment of the present disclosure defines image states including correct positioning, close to correct positioning and misaligned positioning. It can provide feedback to the user whether the position of the image acquisition device is correct and whether it is close to the artery based on the real-time acquired image. Compared with the solution of direct feedback of yes or no, setting close to correct positioning can enable the user to move the image acquisition device more slowly to avoid the problem of positioning error caused by moving too fast or moving too far, which can significantly reduce the user's operation time and assist users with non-medical backgrounds to accurately find the position of arteries.

[0051] In a possible implementation, multiple batches of sample subsets are constructed according to the sample stacked image set, including: For multiple observers, obtaining a labeled sample set corresponding to each observer, each labeled sample set including each sample stack image in the sample stack image set and a sample image state labeled by the observer for each sample stack image, wherein the sample image state is determined by the observer according to the last frame image in the sample image sequence corresponding to the sample stack image; According to the similarity between the labeled sample sets corresponding to the respective observers, a plurality of target observers are selected from the respective observers, and the similarity between the labeled sample sets corresponding to the respective target observers is greater than a preset threshold; According to the labeled sample sets corresponding to each target observer, multiple batches of sample subsets are constructed, and the number of sample stack images from each labeled sample set in each sample subset is consistent.

[0052] Specifically, in order to improve the quality of sample subsets during training, the embodiments of the present disclosure introduce diversity by using annotations of multiple observers. For multiple observers, the sample image state that each observer independently annotates for the last frame of an image in a sample image sequence corresponding to the sample stacked image is obtained, and the sample image state is used as the image state of the corresponding sample stacked image.

[0053] It can be understood that in practical applications, the image state annotated by the observer for each image frame in the video can be obtained, and multiple image sequences can be obtained by segmenting the video frames (including direct segmentation and overlapping segmentation, etc.), and then the images in the image sequence are stacked to obtain stacked images, thereby constructing a sample stacked image set.

[0054] The image state annotated by each observer for each sample stacked image in the sample stacked image set is obtained, and an annotated sample set corresponding to each observer is constructed, where each annotated sample set includes each sample stacked image in the sample stacked image set and the sample image state annotated by the sample stacked image.

[0055] The similarity between the labeled sample sets corresponding to each observer is calculated, and multiple target observers are selected from each observer based on the similarity.

[0056] It is understandable that the purpose of screening observers is to eliminate observers who are too different from other observers to reduce the possibility of sample labeling errors. Therefore, it is necessary to ensure that the similarity between the labeled sample sets corresponding to each target observer is greater than a preset threshold.

[0057] It should be noted that the method for determining that the similarity is greater than the preset threshold can be to determine the specific value of the preset threshold for screening according to actual needs, or to combine the clustering algorithm to eliminate observers with lower similarity with most observers based on the similarity between the labeled sample sets corresponding to each observer.

[0058] Since the training of the positioning network is based on batch training (such as the mini-batch learning algorithm), after determining the target observer, multiple batches of sample subsets are constructed according to the labeled sample sets corresponding to each target observer. Each sample subset needs to evenly contain labeled data from multiple target observers, that is, the number of sample stacked images from each labeled sample set is consistent. It can be understood that the consistency of quantity is not completely equivalent to the same quantity, and a small amount of difference can be allowed.

[0059] The technical solution provided by the embodiment of the present disclosure uses labeled data from multiple observers to construct training samples for the positioning network, which can reduce the deviation of the model to a single annotation and ensure that the number of sample stacked images from each labeled sample set in each sample subset is consistent, which helps to maintain the balance of the data set, avoid the model being biased towards certain specific observer annotations, and improve the performance and reliability of the model.

[0060] In a possible implementation, the stacked image is input into the localization network to obtain the image state output by the localization network, including: Perform feature extraction on the stacked image and output a first feature map; Convert the first feature map into a one-dimensional vector and output a first feature vector; The first eigenvector is converted into the prediction probability corresponding to each image state, and the image state with the highest probability is obtained and output.

[0061] Specifically, the stacked image is input into the localization network, and the localization network processes the stacked image, including three steps: feature extraction, feature integration and vector prediction.

[0062] Among them, feature extraction means that through a series of convolutional layers and pooling layers, the positioning network will automatically extract features from the stacked image and output the first feature map, which contains the abstract feature information of the stacked image.

[0063] Feature integration refers to converting the first feature map from a multidimensional array into a one-dimensional array, converting each pixel value in the first feature map into an element in a one-dimensional vector, and outputting a first feature vector, which contains all feature information extracted from the stacked images.

[0064] Vector prediction refers to inputting the first feature vector into one or more fully connected layers (Dense Layer), which processes the first feature vector and connects the output of the fully connected layer to the classification layer. The classification layer outputs the predicted probability corresponding to each image state, and obtains the image state with the highest probability as the image state predicted by the positioning network and outputs it.

[0065] It can be understood that the various steps implemented by the above-mentioned positioning network are implemented by specific modules included in the positioning network. The embodiments of the present disclosure only limit the method steps that need to be implemented by the positioning network. The network model with the same function can have a variety of different implementation methods. The specific architecture of the positioning network and the type of each module can be determined according to actual needs, and the embodiments of the present disclosure do not limit this.

[0066] For example, consider the huge market demand and potential in the field of medical equipment. With the aging of the population and the improvement of health awareness, the demand for medical imaging technology continues to increase. Portable ultrasound, as a low-cost, radiation-free imaging device, has a large market. In addition, artificial intelligence-assisted imaging equipment can effectively lower the operating threshold and has great market potential in homes, communities and nursing homes.

[0067] The number of deaths and serious cases caused by cardiovascular and cerebrovascular diseases is increasing rapidly around the world, and the demand for portable medical devices is growing. This device can be used for early screening of stroke risk and has application prospects and opportunities in the global market.

[0068] The disclosed embodiment designs the network structure of the positioning network based on serving portable devices, and designs a positioning network based on a lightweight deep convolutional neural network. The positioning network consists of an image input layer, a convolution layer, a convolution module, a pooling layer, and a fully connected layer. Each convolution module includes a 3×3 depth-separable convolution and a 1×1 convolution layer. The two convolution layers are respectively followed by a normalization layer and an activation layer. The normalization layer adopts batch normalization, and the activation layer adopts a linear rectification function. Finally, softmax (normalized exponential function) is used to output the predicted image state value.

[0069] Taking the cross entropy loss function as the target optimization function, the positioning network is trained by the stochastic gradient descent method (Adam) with adaptive first-order and second-order moment estimation. It is possible to consider reducing the number of convolution kernels in the convolution layer to reduce the weight size of the network model without reducing the receptive field of the network model, thereby reducing the calculation requirements and improving the calculation speed.

[0070] It is understandable that the above positioning network architecture and module design are only used as a specific example to assist in illustrating the method provided by the embodiment of the present disclosure. In addition, other architectures and module designs may be adopted according to actual needs, and the embodiment of the present disclosure does not limit this.

[0071] The technical solution provided by the disclosed embodiment can extract and integrate features from the input stacked images and ultimately predict the image status by constructing and pre-training a positioning network. Compared with the related solutions in the prior art, during the positioning process, the technical solution does not need to obtain the segmentation information of the inner membrane and the middle membrane, but only needs to use the original image information of the image acquisition device to assist the user in accurately positioning the arterial blood vessels with stronger real-time performance.

[0072] In a possible implementation, detecting an artery in a target image and determining a detection result includes: Input the target image into the pre-trained segmentation network to obtain the binary image output by the segmentation network. The binary image is used to segment the intima-media region and the non-intima-media region of the vascular wall of the arterial blood vessel. The segmentation network uses the sample image as a training sample, and the sample binary image corresponding to the sample image is obtained by label training; Extracting coordinates of a plurality of intima edge points and a plurality of adventitia edge points from the intima-media region of the binary image; Determine at least one matching point pair according to the coordinates of the plurality of inner membrane edge points and the plurality of outer membrane edge points, the matching point pair comprising an inner membrane matching point and an outer membrane matching point, and satisfying that the inner membrane matching point is the inner membrane edge point with the closest pixel distance to the outer membrane matching point, and the outer membrane matching point is the outer membrane edge point with the closest pixel distance to the inner membrane matching point; Convert the pixel distance between the inner membrane matching point and the outer membrane matching point in at least one matching point pair into an actual distance as the inner membrane thickness value; The detection result is determined based on the intima-media thickness value and the preset intima-media thickness threshold.

[0073] Specifically, Figure 3 A schematic diagram of a carotid artery ultrasound image provided by an embodiment of the present disclosure, such as Figure 3 As shown, the vascular wall of the artery can be divided into the intima, the middle membrane and the adventitia, wherein the proximal and distal ends of the vascular wall refer to the positional relationship relative to the image acquisition device, the intima is adjacent to the vascular cavity, the intima-media thickness refers to the thickness of the intima and the middle membrane, and the intima-media thickness value of the vascular wall of the artery is an important indicator for measuring the health risks of the artery.

[0074] It is understandable that when the image type is an ultrasound image, due to the acoustic characteristics of the vascular adventitia, it has a strong reflection of ultrasound, and artifacts are likely to appear on its back side, which will affect the presentation of the proximal intima-media membrane, thereby causing measurement errors. It is necessary to measure the vascular wall at the distal end to ensure the accuracy of the measurement. In actual applications, the intima-media area of ​​the vascular wall marked by the binary image and the sample binary image refers to the intima-media area of ​​the distal vascular wall. If the image is of other types, the marking position can also be determined according to actual needs.

[0075] The target image determined based on the positioning network is input into the pre-trained segmentation network to obtain a binary image output by the segmentation network. The binary image is used to segment the intima-media region and the non-intima-media region of the vascular wall of the arterial blood vessel.

[0076] Among them, the segmentation network can convert the input image into a binary image that can indicate the inner membrane area and the non-inner membrane area. The segmentation network uses the sample image as the training sample, and the sample binary image corresponding to the sample image is obtained by label training. The specific network model structure and training method can be determined according to actual needs.

[0077] The sample binary image is a binary image with the edge points of the inner membrane and the edge points of the outer membrane marked relative to the sample image. When marking, a number of edge points can be manually marked on the edge of the inner membrane and the edge of the outer membrane of the sample image, and a number of corresponding edge points can be generated by interpolation between adjacent edge points of the inner membrane and adjacent edge points of the outer membrane, and the image marked with the corresponding edge points is converted into a binary image to obtain the sample binary image.

[0078] For example, when training the segmentation network model, the Dice similarity coefficient between the segmentation results of the inner membrane and the outer membrane in the binary image output by the segmentation network and the segmentation results of the inner membrane and the outer membrane in the sample binary image can be calculated to determine whether the segmentation network has converged. If it is determined that it has not converged, the training is continued until the segmentation network is determined to converge, thereby obtaining a trained segmentation network.

[0079] The coordinates of multiple inner membrane edge points and multiple outer membrane edge points are extracted from the inner membrane region of the binary image output by the positioning network.

[0080] For example, edge detection and contour extraction can be performed on the intima-media region in the binary image, and all pixel points on the intima edge and the adventitia edge can be determined as intima edge points and adventitia edge points, respectively, and the corresponding coordinates can be determined.

[0081] The disclosed embodiment calculates the thickness of the inner membrane and the middle membrane based on a symmetrical point correspondence algorithm. For multiple inner membrane edge points and multiple outer membrane edge points obtained, the distance between each edge point can be calculated through the coordinates of each edge point, and at least one matching point pair can be determined based on the nearest neighbor principle of pixel distance.

[0082] The matching point pair includes an inner membrane matching point and an outer membrane matching point, and the inner membrane matching point is an inner membrane edge point with the shortest pixel distance to the outer membrane matching point, and the outer membrane matching point is an outer membrane edge point with the shortest pixel distance to the inner membrane matching point.

[0083] It is understandable that for the inner membrane edge points A1-An and the outer membrane edge points B1-Bn, it may happen that for a given inner membrane edge point A1, the outer membrane edge point closest to A1 pixel is B2, but for a given outer membrane edge point B2, the inner membrane edge point closest to B2 pixel is A2. Therefore, it is necessary to consider both the inner membrane edge points and the outer membrane edge points, and the matching points can be determined only when the pixel distance is the shortest.

[0084] For example, the coordinates of the edge points of the inner membrane and the outer membrane of the inner membrane can be extracted from left to right on the two-dimensional segmented image. For each inner membrane edge point Ai, find the outer membrane edge point Bj closest to the point. Then find the inner membrane edge point closest to the outer membrane edge point Bj. If the inner membrane edge point is equal to Ai, the match is successful. The first inner membrane edge point obtained according to the above algorithm is the initial point. According to the above algorithm, all the inner membrane edge points and the outer membrane edge points are traversed to obtain a series of successfully matched point pairs.

[0085] After determining at least one matching point pair, the pixel distance between the intima matching point and the adventitia matching point in the at least one matching point pair is converted into an actual distance (ie, the pixel distance is converted into an actual thickness value of the intima-media of the blood vessel wall) as the intima-media thickness value.

[0086] It is understandable that when there are multiple matching point pairs, the maximum thickness value, the minimum thickness value or the average thickness value may be obtained based on the intima-media thickness values ​​calculated from different matching point pairs.

[0087] It should be noted that in clinical medicine, the intima-media thickness can be used to assess the risk of cardiovascular disease. For example, in clinical practice, the normal value of the intima-media thickness of the carotid artery is less than 0.9 mm. If the thickness is between 1 mm and 1.5 mm, carotid artery sclerosis or carotid intima-media thickening can be considered. If the thickness is greater than 1.5 mm, plaques may be present.

[0088] In the disclosed embodiment, different intima-media thickness thresholds may be preset for arterial vessels at different locations, and when setting the thresholds, the specific values ​​may be adjusted according to actual needs to meet the user's needs for health risk monitoring.

[0089] After determining the intima-media thickness value, the intima-media thickness value and the preset intima-media thickness threshold are compared, and the test result is determined based on the size relationship. The specific display method of the test result can be determined according to actual needs.

[0090] For example, for the carotid artery, the intima-media thickness thresholds are set to 0.9mm, 1.2mm and 1.5mm. If the intima-media thickness value is determined to be between 0.9mm-1.2mm based on the size relationship, the user is reminded that there is a mild health risk. If the intima-media thickness value is determined to be between 1.2mm-1.5mm, the user is reminded that there is a moderate health risk. If the intima-media thickness value is determined to be greater than 1.5mm, the user is reminded that there is a serious health risk and is advised to seek medical examination.

[0091] The technical solution provided by the embodiment of the present disclosure can automatically segment the acquired target image to obtain a binary image that can indicate the segmentation of the intima-media of the arterial blood vessels, and calculate the intima-media thickness value by constructing and pre-training a segmentation network. This can reduce the error in thickness calculation caused by subjective judgment in segmenting the intima-media, ensure the standardization of the intima-media segmentation and thickness calculation methods, and accurately realize thickness calculation regardless of whether the user has a medical background.

[0092] Compared with the related solutions in the prior art, the present technical solution does not need to use binary images of blood vessels. It can directly generate a binary image indicating the intima-media segmentation based on the original target image, and can produce continuous and dense intima-media segmentation results without the need for interpolation. The calculation of the intima-media thickness based on the acquired binary image is more accurate.

[0093] In a possible implementation, the pixel distance between the inner membrane matching point and the outer membrane matching point in at least one matching point pair is converted into an actual distance, which also includes: For two adjacent matching point pairs, determine the inner membrane edge curve segment and the outer membrane edge curve segment between the two matching point pairs; Determine a target intima edge point from each intima edge point on the intima edge curve segment, determine a target adventitia edge point from each adventitia edge point on the adventitia edge curve segment, and the distance ratio between the target intima edge point and the target adventitia edge point on their respective corresponding edge curve segments is consistent; The target inner membrane edge point and the target outer membrane edge point are used as the newly added matching point pairs.

[0094] Specifically, in order to further improve the accuracy of the intima-media thickness value, the disclosed embodiment considers that after the matching point pairs are extracted for the first time, the distribution of the obtained matching point pairs may be uneven, and there may be a large distance between the matching point pairs, which cannot accurately reflect the intima-media thickness.

[0095] Therefore, the embodiment of the present disclosure generates a newly added matching point pair by interpolating between two adjacent matching point pairs based on each matching point pair obtained for the first time.

[0096] For example, for any two adjacent matching point pairs selected, on the inner membrane edge, the inner membrane edge between the two inner membrane matching points is recorded as an inner membrane edge curve segment, and the outer membrane edge between the two outer membrane matching points is recorded as an outer membrane edge curve segment. There may be multiple inner membrane edge points on the inner membrane edge curve segment, and there may be multiple outer membrane edge points on the outer membrane edge curve segment.

[0097] From each intima edge point on the intima edge curve segment and each adventitia edge point on the adventitia edge curve segment, based on the consistency of corresponding distance ratios, the target intima edge point and the target adventitia edge point are determined, and the distance ratios of the target intima edge point and the target adventitia edge point on their respective corresponding edge curves are consistent.

[0098] It can be understood that the inner membrane edge curve segment and the outer membrane edge curve segment are actually two curve segments, wherein the target inner membrane edge point and the target outer membrane edge point divide their respective corresponding edge curve segments into two parts, and the consistent distance ratio means that the ratio of the corresponding curve arc lengths of the two parts obtained by dividing each edge curve segment is consistent.

[0099] For example, taking the fixed outer membrane edge point and the interpolation determination of the corresponding inner membrane edge point as an example, the method of determining the target outer membrane edge point and the target inner membrane edge point is described in detail: For any two adjacent matching point pairs, the arc length of the curve of the inner membrane edge curve segment between the two inner membrane matching points and the arc length of the curve of the outer membrane edge curve segment between the two outer membrane matching points are calculated.

[0100] From the adventitia edge curve segment, a target adventitia edge point (non-adventitia matching point) is selected, and the ratio of the arc length of the curve from the adventitia edge point to the adjacent adventitia matching point on the left to the arc length of the curve of the adventitia edge curve segment (i.e., the distance ratio) is calculated. It can be understood that the distance ratio can also be a corresponding ratio calculated by the above method with the right side as a reference, or a corresponding ratio of the arc length of the curve formed by the target adventitia edge point and the adventitia matching point on the left to the arc length of the curve formed by the target adventitia edge point and the adventitia matching point on the right, etc.

[0101] The calculated ratio is multiplied by the arc length of the intima edge curve segment to estimate the arc length of the curve between the target intima edge point corresponding to the current target outer membrane edge point and the left intima matching point.

[0102] The target endocardial edge point is interpolated in the endocardial edge curve segment based on the calculated arc length of the curve between the target endocardial edge point and the left endocardial matching point. It can be understood that the target endocardial edge point can be a new endocardial edge point generated on the endocardial edge curve segment, or an endocardial edge point with the smallest error selected from the original endocardial edge points.

[0103] The obtained target inner membrane edge point and the corresponding outer membrane edge point are used as a newly added matching point pair.

[0104] It can be understood that when calculating the intima-media thickness value, for the initial matching point pair (i.e., the matching point pair obtained for the first time) and the newly added matching point pair, an intima-media thickness value can be determined based on the pixel distance between the intima edge point and the adventitia edge point corresponding to each matching point pair.

[0105] Through the above method, based on the matching point pairs obtained for the first time, multiple new matching point pairs can be determined between adjacent matching point pairs by interpolation. When the intima-media thickness is further calculated based on the matching point pairs, the calculation error of the intima-media thickness caused by the uneven distribution of the matching point pairs extracted for the first time can be reduced. A high accuracy can also be maintained in areas with large changes in the curvature of the vascular wall, effectively improving the accuracy of the calculation of the intima-media thickness value.

[0106] In a possible implementation, inputting a target image into a pre-trained segmentation network and obtaining a binary image output by the segmentation network includes: Extract features of the target current image based on multiple different scales, and output multiple second feature maps of different scales; For each second feature map, adjust the second feature map according to the channel and spatial attention weights, and output the adjusted second feature map corresponding to each second feature map; All adjusted second feature maps are fused to obtain a fused feature map, and the fused feature map is mapped into a binary image and output, and the binary image has the same size as the target current image.

[0107] Specifically, the target image is input into the segmentation network, and the segmentation network processes the stacked image, including three steps: feature encoding, attention weighting, and feature decoding.

[0108] Among them, feature encoding refers to scaling the target image at different scales to obtain multiple images of different scales, and performing feature extraction on the image of each scale to determine a corresponding feature map and output it, that is, obtaining multiple second feature maps of different scales.

[0109] The positioning network applies channel attention mechanism and spatial attention mechanism to adjust each second feature map. The channel attention mechanism emphasizes the channel that contributes most to the task and suppresses irrelevant or redundant channels by assigning different weights to each channel. The spatial attention mechanism weights the specific spatial positions in the feature map to highlight the areas that contribute most to the task and suppress irrelevant or redundant areas.

[0110] The channel and space attention weights are determined by training the positioning network, that is, the attention weights include channel attention weights (weight vectors) and spatial attention weights (spatial heat maps). Attention weighting means that for each second feature map, according to the channel attention weights, the second feature map is weighted with the channel weight corresponding to each channel to obtain the channel feature map adjusted by the channel attention. According to the spatial attention weights, each pixel in the channel feature map is weighted to obtain the spatial feature map adjusted by the spatial attention. The spatial feature map is used as the adjusted second feature map, and the adjusted second feature map corresponding to each second feature map is output.

[0111] Feature decoding refers to finding the second feature map of the corresponding size for each adjusted second feature map through deconvolution and upsampling operations and stacking them, restoring the resolution of the adjusted second feature map, restoring the resolution to the same as the target image, obtaining each restored feature map, and fusing each restored feature map to obtain a fused feature map, mapping the fused feature map to a binary image and outputting it, and the binary image has the same size as the target current image.

[0112] It can be understood that each step implemented by the above-mentioned segmentation network is implemented by a specific module included in the segmentation network. The embodiments of the present disclosure only limit the method steps required to be implemented by the segmentation network. The network model with the same function can have a variety of different implementation methods. The specific architecture of the segmentation network and the type of each module can be determined according to actual needs, and the embodiments of the present disclosure do not limit this.

[0113] For example, we can design a segmentation network of a lightweight U-type (U-Net network structure) attention convolutional network based on the same starting point as the network structure of the positioning network. The segmentation network consists of an encoder, a decoder, and an attention module.

[0114] Based on the U-Net network structure, a segmentation network including an encoder, an attention module and a decoder was constructed to handle the segmentation task of the intima-media membrane in the image, automatically identify and accurately mark the boundaries of the intima-media membrane, reduce the errors that may occur during manual image segmentation, and improve the accuracy of intima-media membrane thickness calculation.

[0115] Among them, the encoder consists of multiple convolution modules and pooling layers. Each convolution module contains 3×3 depth-separable convolutions. The separable convolution operation can significantly reduce the weight size of the model compared to conventional convolutions. The batch normalization layer and linear rectification activation function enable the encoder to generate multi-scale image features.

[0116] The attention mechanism is between the encoder and the decoder, and consists of a spatial attention module and a channel attention module. It can generate spatial heat maps and channel weights to extract effective spatial and channel information.

[0117] The decoder contains multiple convolution modules and upsampling operations that are the same as the encoder, fuses multi-scale features, and maps the features into a binary image of the same size as the input image through a sigmoid function.

[0118] When training the segmentation network, the goal can be to minimize the Dice similarity coefficient and train the segmentation network through the stochastic gradient descent method (Adam) with adaptive first-order and second-order moment estimation.

[0119] Based on the design of positioning network and segmentation network with serving portable devices as the starting point, the hardware requirements of network model can be reduced. A portable ultrasound device based on artificial intelligence assistance is designed. As a low-cost, radiation-free and easy-to-operate imaging device, portable ultrasound device has the advantages of easy operation, short response time, high accuracy and low hardware requirements, and has broad application prospects.

[0120] It can provide clinicians with more accurate and comprehensive diagnostic information in the field of medical imaging, which can help with early disease detection and treatment monitoring. It can also be used by users without a medical background to assess health risks.

[0121] It is understandable that the above segmentation network architecture and module design are only used as a specific example to assist in illustrating the method provided in the embodiment of the present disclosure. In addition, other architectures and module designs may be adopted according to actual needs, and the embodiment of the present disclosure does not limit this.

[0122] It should be understood that the embodiments of the present disclosure are only used as a specific example to explain the present invention in detail, and the described embodiments are only intended to facilitate the understanding of the present invention and do not have any limiting effect on the present invention.

[0123] Figure 4 A schematic diagram of the structure of an artery positioning detection device provided in an embodiment of the present disclosure is shown in FIG. Figure 4 As shown, the artery positioning detection device 400 includes: The positioning identification module 401 is used to iteratively perform at least one round of the following steps in the process of continuously moving the image acquisition device at the target part until the completeness meets the preset conditions: obtaining an image sequence acquired by the image acquisition device in this round, stacking the image sequence into a stacked image, inputting the stacked image into the positioning network, and obtaining an image state output by the positioning network, wherein the image state represents the completeness of the vascular wall of the arterial blood vessel identified according to the corresponding stacked image; The artery detection module 402 is used to use the last frame image in the image sequence obtained when the completeness meets the preset conditions as the target image, detect the arterial blood vessels in the target image, and determine the detection result, which is used to indicate the health risk of the arterial blood vessels.

[0124] In a possible implementation, the image status includes correct positioning and incorrect positioning; the type of incorrect positioning includes any one of close to correct positioning and wrong positioning; Correct positioning means that the complete vessel wall of the arterial blood vessel is identified according to the corresponding stacked images, nearly correct positioning means that part of the vessel wall of the arterial blood vessel is identified according to the corresponding stacked images, and incorrect positioning means that the vessel wall of the arterial blood vessel cannot be identified according to the corresponding stacked images; The localization network is generated in the following way: Acquire a sample stacked image set, where the sample stacked image set includes at least one sample stacked image; Constructing multiple batches of sample subsets according to the sample stacked image set, each sample subset including multiple sample stacked images and a sample image state corresponding to each sample stacked image; According to the sample subsets of each batch, the initial positioning network is trained in batches with the minimization of the loss function as the optimization goal until the preset training stop condition is met to obtain the positioning network; Among them, the loss function is a multivariate cross entropy loss function that describes the difference between the predicted image state of the positioning network corresponding to the sample stacked image and the sample image state.

[0125] In a possible implementation, the artery location detection device further includes a sample construction module; Sample building blocks for: For multiple observers, obtaining a labeled sample set corresponding to each observer, each labeled sample set including each sample stack image in the sample stack image set and a sample image state labeled by the observer for each sample stack image, wherein the sample image state is determined by the observer according to the last frame image in the sample image sequence corresponding to the sample stack image; According to the similarity between the labeled sample sets corresponding to the respective observers, a plurality of target observers are selected from the respective observers, and the similarity between the labeled sample sets corresponding to the respective target observers is greater than a preset threshold; According to the labeled sample sets corresponding to each target observer, multiple batches of sample subsets are constructed, and the number of sample stack images from each labeled sample set in each sample subset is consistent.

[0126] In a possible implementation, the positioning and identification module is used to: Perform feature extraction on the stacked image and output a first feature map; Convert the first feature map into a one-dimensional vector and output a first feature vector; The first eigenvector is converted into the prediction probability corresponding to each image state, and the image state with the highest probability is obtained and output.

[0127] In a possible implementation, the artery detection module is used to: Input the target image into the pre-trained segmentation network to obtain the binary image output by the segmentation network. The binary image is used to segment the intima-media region and the non-intima-media region of the vascular wall of the arterial blood vessel. The segmentation network uses the sample image as a training sample, and the sample binary image corresponding to the sample image is obtained by label training; Extracting coordinates of a plurality of intima edge points and a plurality of adventitia edge points from the intima-media region of the binary image; Determine at least one initial matching point pair according to the coordinates of the plurality of inner membrane edge points and the plurality of outer membrane edge points, the matching point pair comprising an inner membrane matching point and an outer membrane matching point, and satisfying that the inner membrane matching point is the inner membrane edge point with the closest pixel distance to the outer membrane matching point, and the outer membrane matching point is the outer membrane edge point with the closest pixel distance to the inner membrane matching point; Convert the pixel distance between the inner membrane matching point and the outer membrane matching point in at least one matching point pair into an actual distance as the inner membrane thickness value; The detection result is determined based on the intima-media thickness value and the preset intima-media thickness threshold.

[0128] In a possible implementation, the artery detection module is used to: For two adjacent matching point pairs, determine the inner membrane edge curve segment and the outer membrane edge curve segment between the two matching point pairs; Determine a target intima edge point from each intima edge point on the intima edge curve segment, determine a target adventitia edge point from each adventitia edge point on the adventitia edge curve segment, and the distance ratio between the target intima edge point and the target adventitia edge point on their respective corresponding edge curve segments is consistent; The target inner membrane edge point and the target outer membrane edge point are used as the newly added matching point pairs.

[0129] In a possible implementation, the artery detection module is used to: Extract features of the target current image based on multiple different scales, and output multiple second feature maps of different scales; For each second feature map, adjust the second feature map according to the channel and spatial attention weights, and output the adjusted second feature map corresponding to each second feature map; All adjusted second feature maps are fused to obtain a fused feature map, and the fused feature map is mapped into a binary image and output, and the binary image has the same size as the target current image.

[0130] The device of the embodiments of the present disclosure can execute the method provided by the embodiments of the present disclosure, and the implementation principles are similar. The actions performed by each module in the device of each embodiment of the present disclosure correspond to the steps in the method of each embodiment of the present disclosure. For the detailed functional description of each module of the device, please refer to the description in the corresponding method shown in the previous text, which will not be repeated here.

[0131] In addition, in the embodiments of the present disclosure, the term "module" or "unit" refers to a computer program or a part of a computer program with a predetermined function, and works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as processing circuits or memories), or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be part of an overall module or unit that includes the function of the module or unit.

[0132] An embodiment of the present disclosure provides an electronic device (computer device / equipment / system), including a memory, a processor, and a computer program stored in the memory, and the processor executes the above-mentioned computer program to implement the steps of the method provided in any optional embodiment of the present disclosure and achieve corresponding technical effects.

[0133] In an optional embodiment, an electronic device is provided. Figure 5 A schematic diagram of the structure of an electronic device provided by an embodiment of the present disclosure is shown in FIG. Figure 5As shown, the electronic device 500 includes: a processor 501 and a memory 503. The processor 501 and the memory 503 are connected, such as through a bus 502. Optionally, the electronic device 500 may also include a transceiver 504, which may be used for data interaction between the electronic device and other electronic devices, such as data transmission and / or data reception. It should be noted that in actual applications, the transceiver 504 is not limited to one, and the structure of the electronic device 500 does not constitute a limitation on the embodiments of the present disclosure.

[0134] Processor 501 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of the present invention. Processor 501 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0135] The bus 502 may include a path to transmit information between the above components. The bus 502 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 502 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0136] The memory 503 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), magnetic disk storage medium, other magnetic storage devices, or any other medium that can be used to carry or store computer programs and can be read by a computer, without limitation herein.

[0137] The memory 503 is used to store the computer program for executing the embodiment of the present disclosure, and the execution is controlled by the processor 501. The processor 501 is used to execute the computer program stored in the memory 503 to implement the steps shown in the above method embodiment.

[0138] The electronic devices in the embodiments of the present disclosure may include, but are not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), wearable devices, etc., and fixed terminals such as digital TVs, desktop computers, etc.

[0139] An embodiment of the present disclosure provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps and corresponding contents of the aforementioned method embodiment can be implemented.

[0140] The embodiments of the present disclosure also provide a computer program product, including a computer program, which can implement the steps and corresponding contents of the aforementioned method embodiments when executed by a processor.

[0141] It should be noted that the computer-readable storage medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0142] In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus, or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0143] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0144] The terms "first", "second", "third", "fourth", "1", "2", etc. (if any) in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than that shown or described in the text.

[0145] It should be understood that, although the flowchart of the embodiment of the present disclosure indicates each operation step by arrows, the implementation order of these steps is not limited to the order indicated by the arrows. Unless clearly stated herein, in some implementation scenarios of the embodiment of the present disclosure, the implementation steps in each flowchart can be executed in other orders according to demand. In addition, some or all of the steps in each flowchart may include multiple sub-steps or multiple stages based on the actual implementation scenario. Some or all of these sub-steps or stages can be executed at the same time, and each sub-step or stage in these sub-steps or stages can also be executed at different times. In scenarios with different execution times, the execution order of these sub-steps or stages can be flexibly configured according to demand, and the embodiment of the present disclosure does not limit this.

[0146] The above is only an optional implementation method for some implementation scenarios of the present disclosure. It should be pointed out that for ordinary technicians in this technical field, without departing from the technical concept of the scheme of the present disclosure, other similar implementation methods based on the technical ideas of the present disclosure are also within the protection scope of the embodiments of the present disclosure.

Claims

1. A method for detecting arterial location, characterized in that: include: In the process of continuously moving the image acquisition device at the target part, iteratively execute at least one round of the following steps until the completeness meets the preset conditions: Obtaining an image sequence acquired by the image acquisition device in this round, stacking the image sequence into a stacked image, inputting the stacked image into a positioning network, and obtaining an image state output by the positioning network, wherein the image state represents the integrity of the vessel wall of the arterial vessel identified according to the corresponding stacked image; The last frame of the image in the image sequence obtained when the completeness meets the preset conditions is used as the target image, the arterial blood vessels in the target image are detected, and the detection result is determined, and the detection result is used to indicate the health risk of the arterial blood vessels.

2. The artery location detection method according to claim 1, characterized in that: The image status includes correct positioning and incorrect positioning; the type of incorrect positioning includes any one of close to correct positioning and wrong positioning; The correct positioning means that the complete vessel wall of the arterial blood vessel is identified according to the corresponding stacked images, the nearly correct positioning means that part of the vessel wall of the arterial blood vessel is identified according to the corresponding stacked images, and the incorrect positioning means that the vessel wall of the arterial blood vessel cannot be identified according to the corresponding stacked images; The positioning network is generated by: Acquire a sample stacked image set, wherein the sample stacked image set includes at least one sample stacked image; According to the sample stacked image set, constructing a plurality of batches of sample subsets, each sample subset including a plurality of sample stacked images and a sample image state corresponding to each sample stacked image; According to the sample subsets of each batch, the initial positioning network is trained in batches with minimization of the loss function as the optimization goal until a preset training stop condition is met, thereby obtaining the positioning network; The loss function is a multivariate cross entropy loss function that describes the difference between the predicted image state of the positioning network corresponding to the sample stacked image and the sample image state.

3. The artery location detection method according to claim 2, characterized in that: The step of constructing a plurality of batches of sample subsets according to the sample stacked image set comprises: For multiple observers, obtaining a labeled sample set corresponding to each observer, each labeled sample set including each sample stack image in the sample stack image set and a sample image state labeled by the observer for each sample stack image, wherein the sample image state is determined by the observer according to the last frame image in the sample image sequence corresponding to the sample stack image; According to the similarity between the labeled sample sets corresponding to the respective observers, a plurality of target observers are selected from the respective observers, and the similarity between the labeled sample sets corresponding to the respective target observers is greater than a preset threshold; According to the labeled sample sets corresponding to each target observer, multiple batches of sample subsets are constructed, and the number of sample stack images from each labeled sample set in each sample subset is consistent.

4. The artery location detection method according to claim 1, characterized in that: The step of inputting the stacked image into a positioning network to obtain an image state output by the positioning network comprises: Performing feature extraction on the stacked image and outputting a first feature map; Convert the first feature map into a one-dimensional vector and output a first feature vector; The first feature vector is converted into prediction probabilities corresponding to various image states, and the image state with the highest probability is obtained and output.

5. The artery positioning detection method according to any one of claims 1 to 4, characterized in that: The detecting of the arterial blood vessels in the target image and determining the detection result includes: Input the target image into a pre-trained segmentation network to obtain a binary image output by the segmentation network, wherein the binary image is used to segment the intima-media region and the non-intima-media region of the vascular wall of the arterial blood vessel, wherein the segmentation network uses the sample image as a training sample, and the sample binary image corresponding to the sample image is obtained through label training; Extracting coordinates of a plurality of intimal edge points and a plurality of adventitia edge points from the intimal-medial region of the binary image; Determine at least one matching point pair according to the coordinates of the multiple inner membrane edge points and the multiple outer membrane edge points, the matching point pair includes an inner membrane matching point and an outer membrane matching point, and the inner membrane matching point is the inner membrane edge point with the closest pixel distance to the outer membrane matching point, and the outer membrane matching point is the outer membrane edge point with the closest distance to the inner membrane matching point; Convert the pixel distance between the inner membrane matching point and the outer membrane matching point in the at least one matching point pair into an actual distance as an inner membrane thickness value; The detection result is determined according to the intima-media thickness value and a preset intima-media thickness threshold.

6. The artery location detection method according to claim 5, characterized in that: The step of converting the pixel distance between the inner membrane matching point and the outer membrane matching point in the at least one matching point pair into an actual distance also includes: For two adjacent matching point pairs, determine the inner membrane edge curve segment and the outer membrane edge curve segment between the two matching point pairs; Determine a target intima edge point from each intima edge point on the intima edge curve segment, and determine a target adventitia edge point from each adventitia edge point on the adventitia edge curve segment, wherein the distance ratio between the target intima edge point and the target adventitia edge point on their respective corresponding edge curve segments is consistent; The target inner membrane edge point and the target outer membrane edge point are used as a newly added matching point pair.

7. The artery location detection method according to claim 5, characterized in that: The step of inputting the target image into a pre-trained segmentation network to obtain a binary image output by the segmentation network comprises: Extracting features from the target current image based on multiple different scales, and outputting multiple second feature maps at different scales; For each second feature map, adjusting the second feature map according to the channel and spatial attention weights, and outputting the adjusted second feature map corresponding to each second feature map; All adjusted second feature maps are fused to obtain a fused feature map, and the fused feature map is mapped into a binary image and output, and the binary image has the same size as the target current image.

8. An artery location detection device, characterized in that: include: The positioning and identification module is used to iteratively perform at least one round of the following steps in the process of continuously moving the image acquisition device at the target part until the degree of integrity meets the preset conditions: obtaining an image sequence acquired by the image acquisition device in this round, stacking the image sequence into a stacked image, inputting the stacked image into a positioning network, and obtaining an image state output by the positioning network, wherein the image state represents the degree of integrity of the vascular wall of the arterial blood vessel identified according to the corresponding stacked image; The artery detection module is used to use the last frame image in the image sequence obtained when the completeness meets the preset conditions as the target image, detect the arterial blood vessels in the target image, and determine the detection result, which is used to indicate the health risk of the arterial blood vessels.

9. An electronic device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

11. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images