A target type identification method, device and ultrasonic equipment

Abstract

The application discloses a target type identification method and device and ultrasonic equipment. The method comprises the following steps: acquiring an ultrasonic contrast image; performing target feature analysis on continuous multiple target contrast images in the ultrasonic contrast image respectively to determine corresponding regions of interest; acquiring position information and color information of edge points in each region of interest respectively to obtain edge point information of each target contrast image; fusing the edge point information of the continuous multiple target contrast images to obtain fusion features, and performing classification processing on the ultrasonic contrast image based on the fusion features to acquire a target type corresponding to the ultrasonic contrast image. The method solves the problem that, when the ultrasonic contrast technology is used to identify the target type, a doctor needs to identify and analyze the features of the ultrasonic contrast image manually, a large amount of time and labor is consumed, and the identification result is extremely easy to be wrong when the doctor is tired. The method reduces the consumption of labor and improves the accuracy of the identification result.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical imaging, and in particular, to a target type identification method and device and an ultrasound device. BACKGROUND

[0002] Ultrasound contrast technology is based on conventional ultrasound examination, and by injecting contrast agent into the human body through the vein, the entering and exiting mode of the contrast agent in the human body, such as speed, mode and intensity, is dynamically and clearly displayed, thereby assisting doctors in making specific diagnosis. Ultrasound contrast technology has the advantages of dynamic, real-time and continuous display of vascular structure and tissue perfusion, and has very good clinical value in identifying the characteristics of benign and malignant space-occupying lesions.

[0003] Taking the identification of liver characteristics based on ultrasound contrast technology as an example, in actual operation, the ultrasound contrast is divided into three contrast periods, namely, the arterial phase, the portal phase and the delay phase, according to the time sequence of the contrast agent reaching different tissues of the liver. Doctors need to identify and analyze the characteristics of the ultrasound contrast image based on the ultrasound contrast image according to the perfusion of the contrast agent in the liver in different contrast periods, and then assist subsequent work.

[0004] However, when using ultrasound contrast technology to identify the target type, doctors need to identify and analyze the characteristics of the ultrasound contrast image, which consumes a lot of time and manpower, and the recognition result is extremely prone to error when the doctors are working tiredly. SUMMARY

[0005] Therefore, the embodiments of the present application provide a target type identification method, device and ultrasound device to solve the problem that doctors need to identify and analyze the characteristics of the ultrasound contrast image when using ultrasound contrast technology to identify the target type, which consumes a lot of time and manpower, and the recognition result is extremely prone to error when the doctors are working tiredly.

[0006] According to a first aspect, the embodiments of the present application provide a target type identification method, comprising:

[0007] obtaining an ultrasound contrast image;

[0008] analyzing the target characteristics of a plurality of continuous target contrast images in the ultrasound contrast image respectively, and determining the region of interest corresponding to each target contrast image;

[0009] obtaining the position information of the edge points of each region of interest respectively;

[0010] Based on the pixel value of each pixel point of the region of interest, color information of the edge point is acquired to obtain edge point information of each frame of the target contrast image, and the edge point information includes the position information and the color information.

[0011] The edge point information of the continuous multiple frames of target contrast images is fused to obtain fusion features, and the ultrasound contrast image is classified based on the fusion features to obtain a target type corresponding to the ultrasound contrast image.

[0012] The target type recognition method provided by the embodiment of the application can determine the region of interest by analyzing the target features of the continuous multiple frames of target contrast images in the ultrasound contrast image, save the calculation amount, reduce the recognition range, and accurately obtain the region to be recognized; the edge point information of each region of interest is acquired to accurately obtain the feature information in each frame of target contrast image, improve the accuracy of target type recognition; the edge point information is fused to obtain fusion features, which further saves the calculation amount; the ultrasound contrast image is classified based on the fusion features to accurately recognize the target type corresponding to the ultrasound contrast image.

[0013] In combination with the first aspect, in a first implementation manner of the first aspect, the position information of the edge point of each region of interest is acquired respectively, including:

[0014] The edge of each region of interest is extracted from the target ultrasound contrast image.

[0015] A predetermined number of edge points are extracted from the extracted edge to obtain the position information of the edge point of each region of interest.

[0016] The target type recognition method provided by the embodiment of the application can obtain a predetermined number of edge points and position information from the edge of the region of interest, and can perform overall position analysis on the features of each region of interest.

[0017] In combination with the first aspect, in a first implementation manner of the first aspect, the position information of the edge point of each region of interest is acquired respectively, including:

[0018] The centroid of each region of interest is extracted from the target ultrasound contrast image.

[0019] The region of interest is divided at equal angles based on the centroid to obtain the position information of the predetermined number of edge points.

[0020] The target type recognition method provided by the embodiment of the present application can improve the rationality and accuracy of position analysis of each region of interest by using the centroid of each region of interest to perform equi-angle division on each region of interest to obtain position information of a predetermined number of edge points.

[0021] In the third implementation form of the first aspect, in combination with the second implementation form of the first aspect, the color information of the edge points is obtained based on the pixel values of the pixel points of each region of interest, and the color information of the edge points comprises:

[0022] The pixel mean values of the pixel values of the pixel points of each region of interest sub-region obtained after equi-angle division of the region of interest are respectively obtained;

[0023] The pixel mean values of the corresponding region of interest sub-regions are determined as the color information of the corresponding edge points.

[0024] The target type recognition method provided by the embodiment of the present application can improve the rationality and accuracy of color analysis of each region of interest by determining the pixel mean values of the pixel values of the pixel points in each region of interest sub-region as the color information of the corresponding edge points.

[0025] In the fourth implementation form of the first aspect, in combination with the first aspect, the position information of the edge points comprises horizontal coordinates and vertical coordinates of the edge points, and the fusion feature is obtained by fusing the edge point information of the continuous multiple frames of target contrast images, and the fusion feature comprises:

[0026] The feature length of each frame of the target contrast images is obtained.

[0027] For each region of interest of each frame of the target contrast images, the horizontal coordinates, the vertical coordinates and the color values of the corresponding edge points are sequentially spliced based on the feature length, to obtain the horizontal coordinate feature vector, the vertical coordinate feature vector and the color feature vector corresponding to each frame of the target contrast images.

[0028] The horizontal coordinate feature vectors, the vertical coordinate feature vectors and the color feature vectors of all the target contrast images are fused respectively to obtain a three-channel matrix corresponding to the ultrasonic contrast image, so as to determine the fusion feature.

[0029] The target type recognition method provided by the embodiment of the present application can save the calculation amount and improve the recognition efficiency of the target type by fusing the horizontal coordinate feature vectors, the vertical coordinate feature vectors and the color feature vectors of all the target contrast images into a three-channel matrix.

[0030] With reference to any one of the first to fourth implementation manners of the first aspect, in a fifth implementation manner of the first aspect, the analysis of the target feature of each of the plurality of continuous target contrast images in the contrast-enhanced ultrasound image, and the determination of the region of interest corresponding to each of the plurality of continuous target contrast images, include:

[0031] The plurality of continuous target contrast images and the plurality of continuous target B-mode images are respectively extracted from the contrast-enhanced ultrasound image, and the plurality of continuous target B-mode images correspond one-to-one to the plurality of continuous target contrast images;

[0032] A preset image resolution is obtained;

[0033] The plurality of continuous target contrast images and the plurality of continuous target B-mode images are respectively scaled based on the preset image resolution, to obtain scaled target contrast images and scaled target B-mode images;

[0034] The scaled target contrast images are segmented based on the scaled target B-mode images, to determine the region of interest included in each of the plurality of continuous target contrast images.

[0035] The target type recognition method provided in the embodiments of the present application can effectively save the calculation amount by respectively scaling the plurality of continuous target contrast images and the plurality of continuous target B-mode images, and can further reduce the calculation amount of target type recognition and improve the recognition accuracy by segmenting the scaled target contrast images to determine the region of interest included in each of the target contrast images.

[0036] With reference to the fifth implementation manner of the first aspect, in a sixth implementation manner of the first aspect, the segmentation of the scaled target contrast images based on the scaled target B-mode images, to determine the region of interest included in each of the target contrast images, includes:

[0037] An RGB channel of the scaled target contrast images and a grayscale channel of the scaled target B-mode images are obtained;

[0038] The scaled target contrast images are instance segmented based on the RGB channel and the grayscale channel, to obtain the region of interest included in each of the target contrast images.

[0039] The target type recognition method provided in the embodiments of the present application can accurately obtain the region of interest by instance segmenting the target contrast images based on the RGB channel and the grayscale channel.

[0040] According to a second aspect, an embodiment of the present application provides a target type identification device, applied to the target type identification method in the first aspect of the present application, comprising:

[0041] an image acquisition module, configured to acquire an ultrasound contrast image;

[0042] an interested region acquisition module, configured to analyze target features of continuous target contrast images in the ultrasound contrast image respectively, and determine an interested region corresponding to each of the target contrast images;

[0043] an edge point position information acquisition module, configured to acquire position information of an edge point of each of the interested regions respectively;

[0044] an edge point information acquisition module, configured to acquire color information of the edge point based on pixel values of pixel points of each of the interested regions, so as to obtain edge point information of each of the target contrast images, the edge point information comprising the position information and the color information;

[0045] a target type acquisition module, configured to fuse the edge point information of the continuous target contrast images to obtain fusion features, and perform classification processing on the ultrasound contrast image based on the fusion features, so as to acquire a target type corresponding to the ultrasound contrast image.

[0046] The target type identification device provided by the embodiment of the present application saves calculation amount, reduces the identification range, and accurately acquires the region to be identified by analyzing target features of continuous target contrast images in the ultrasound contrast image respectively through the interested region acquisition module; the edge point information acquisition module accurately obtains feature information in each of the target contrast images by acquiring edge point information of each of the interested regions, thereby improving the accuracy of target type identification; the target type acquisition module fuses the edge point information to obtain fusion features, thereby further saving calculation amount, and performs classification processing on the ultrasound contrast image based on the fusion features, thereby accurately identifying the target type corresponding to the ultrasound contrast image.

[0047] According to a third aspect, an embodiment of the present application provides an ultrasound device, comprising at least one processor, and a memory connected with the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform steps of the target type identification method in the first aspect of the present application.

[0048] According to a fourth aspect, an embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the target type identification method according to the first aspect of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the present application, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, for those skilled in the art, other drawings can also be obtained based on these drawings without any creative effort.

[0050] Figure 1 is a method flow chart of the target type identification method disclosed by the embodiment of the present application.

[0051] Figure 2 is a method flow chart of the target type identification method disclosed by the embodiment of the present application.

[0052] Figure 3 is an ultrasound contrast image related to the embodiment of the present application.

[0053] Figure 4 is an instance segmentation algorithm block diagram related to the embodiment of the present application.

[0054] Figure 5 is a portal vein feature map after liver contrast segmentation related to the embodiment of the present application.

[0055] Figure 6 is a segmentation schematic diagram of a region of interest related to the embodiment of the present application.

[0056] Figure 7 is a feature length map of a single-frame target contrast image related to the embodiment of the present application.

[0057] Figure 8 is a three-channel matrix related to the embodiment of the present application.

[0058] Figure 9 is a classification network schematic diagram related to the embodiment of the present application.

[0059] Figure 10 is a structure block diagram of the target type identification device disclosed by the embodiment of the present application.

[0060] Figure 11 is an ultrasound device schematic diagram disclosed by the embodiment of the present application. DETAILED DESCRIPTION

[0061] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. The present invention has been described in detail above with reference to specific embodiments and exemplary examples; however, these descriptions should not be construed as limiting the present invention. Those skilled in the art will understand that various equivalent substitutions, modifications, or improvements can be made to the technical solutions and implementation methods of the present invention without departing from the spirit and scope of the present invention, and all of these fall within the scope of the present invention. The scope of protection of the present invention is determined by the appended claims.

[0062] It should be noted that the target type identification method, apparatus, and ultrasound equipment provided in this embodiment of the invention can be applied to the field of ultrasound contrast imaging technology, such as the field of liver ultrasound contrast imaging technology. No specific application area is limited herein, and it can be used according to actual circumstances. In the embodiments described below, an application to the field of liver ultrasound contrast imaging technology is used as an example for detailed description.

[0063] When using ultrasound contrast imaging technology to identify target types, doctors need to manually identify and analyze the features of the ultrasound contrast imaging images, which consumes a lot of time and manpower. Moreover, when doctors are working while fatigued, the identification results are extremely prone to errors.

[0064] Based on this, embodiments of the present invention propose a target type identification method, such as... Figure 1 As shown, the method includes:

[0065] Step S101: Obtain ultrasound contrast imaging.

[0066] Specifically, the ultrasound contrast imaging is obtained by saving and retrieving the real-time contrast imaging image and B-mode image simultaneously provided by the ultrasound machine during ultrasound contrast imaging. For example... Figure 3 The ultrasound contrast imaging image shown is divided into left and right halves, representing the contrast mode and ultrasound B-mode images, respectively. Figure 3 In the image, the left side shows the contrast-enhanced image, and the right side shows the ultrasound B-mode image. This means that the contrast-enhanced image and the B-mode image can display the blood perfusion and echogenicity at a specific moment. During ultrasound contrast-enhanced imaging, contrast agents are injected intravenously to dynamically and clearly display the entry and exit patterns of the contrast agent, thus acquiring the characteristics of the ultrasound contrast-enhanced image.

[0067] In step S102, the target feature analysis is performed on each of the continuous target contrast images in the contrast ultrasound image to determine the region of interest corresponding to each of the target contrast images.

[0068] Specifically, after the contrast ultrasound image is acquired, the contrast ultrasound image is analyzed frame by frame. The continuous target contrast images and the continuous target B-mode images corresponding to the continuous target contrast images are extracted from the contrast ultrasound image. The continuous target contrast images and the continuous target B-mode images are scaled based on a preset image resolution. For example, in general, in order to save the calculation amount, a relatively low image resolution can be preset, and the continuous target contrast images and the continuous target B-mode images are scaled down according to the preset relatively low image resolution. When a more delicate and clear image is needed, a relatively high image resolution can be preset, and the continuous target contrast images and the continuous target B-mode images are scaled up according to the preset relatively high image resolution.

[0069] The scaled target B-mode image is converted into a gray-scale image, and the gray-scale channel is obtained. The RGB channel of the scaled target contrast image is obtained. The RGB channel and the gray-scale channel are taken as inputs of an instance segmentation algorithm, and the instance segmentation algorithm is performed on the scaled target contrast image to obtain the region of interest contained in each of the target contrast images. That is, a single-stage detector can be used. The RGB channel and the gray-scale channel are taken as inputs of the instance segmentation algorithm, and the extracted features of the detected part are further calculated in the convolution head to generate a corresponding segmentation output. The segmented region is the region of interest contained in each of the target contrast images. Figure 4The example segmentation algorithm block diagram shows a schematic diagram of a possible implementation of a single-stage detector for instance segmentation. The RGB channel and the grayscale channel are put into the backbone network as the input of the instance segmentation algorithm and are convolved to obtain the detected region of interest (roi). The bounding box (bbox) of the detected region of interest (roi) is obtained, and the classification and binary mask processing are performed to generate the final output. When using the instance segmentation algorithm, a single-stage detector can be used, and the detected region of interest (roi) is extracted for feature extraction into the convolution head for further operation to generate the corresponding segmentation output. A two-stage detection algorithm can also be used to output the detected region of interest (roi) in the second stage. The method of splicing the detection network and the semantic segmentation network can also be used, i.e., the image is first input into the detection network to generate the detection result of the corresponding bounding box (bounding box, bbox), and then the image content in the bounding box (bounding box, bbox) is extracted and input into the semantic segmentation network to generate the final output. For example Figure 5 As shown, Figure 5 The ultrasound contrast imaging of liver tissue is taken as an example, in which the region of interest is a liver tissue feature such as liver parenchyma, arterial blood vessels, intrahepatic bile ducts, and portal veins. The number of regions of interest is increased or decreased according to the specific target type to be classified, and Figure 5 The region of interest segmented in the image is the portal vein in the liver tissue feature.

[0070] The continuous multiple frames of target contrast images and the continuous multiple frames of target B-mode images are respectively scaled based on the preset image resolution, which can save the calculation amount, reduce the target type recognition range, and perform instance segmentation processing on the scaled target contrast images through the RGB channel and the grayscale channel to obtain the region of interest contained in each frame of target contrast image, which can accurately obtain the region to be recognized and improve the accuracy of target type recognition.

[0071] Optionally, the target type can be used to indicate different states of the liver, gastrointestinal tract or heart. For example, the target type can include liver absence, liver shadow, liver without obvious abnormalities, etc. After identifying the target type, the state of the liver can be further determined by the computer device or manually according to the identified target type. For example, based on the target type, whether the liver has shadow, absence, abnormality, etc. is identified, and then the liver pathological state is further analyzed by the computer device or manually according to the identified target type. The liver pathological state can include liver hemangioma, inflammatory pseudotumor, focal nodular lesion, primary cancer and metastatic cancer, etc.

[0072] In step S103, the position information of the edge point of each of the regions of interest is obtained.

[0073] Specifically, after obtaining the regions of interest, the edge of each of the regions of interest and the centroid of each of the regions of interest are extracted from the target ultrasound contrast image. The regions of interest are equally divided based on the image coordinates of the centroid of each of the regions of interest, and the position information of a predetermined number of edge points is obtained. As shown in the left image of FIG. 12, the 12 edge points and the centroid are extracted, and the regions of interest are equally divided into 12 sub-regions, which form an umbrella-shaped region. Figure 6 Figure 6 The position information of the edge point can be the image coordinates of the edge point, including the horizontal coordinate and the vertical coordinate. By obtaining the position information of the edge point of each of the regions of interest, the position information of the region to be identified can be accurately obtained.

[0074] In step S104, the color information of the edge point is obtained based on the pixel value of the pixel point of each of the regions of interest, so as to obtain the edge point information of each frame of the target contrast image, which includes the position information and the color information.

[0075] ​Specifically, after the interest region is equally divided, the pixel mean value of the pixel value of each pixel point in each interest sub-region is obtained; and the pixel mean value corresponding to each interest sub-region is determined as the color information of the corresponding edge point. The pixel mean value of each interest sub-region can be an average RGB value. After the edge of each interest region and the centroid of each interest region are extracted, the image coordinates of the centroid of each interest region are used to equally divide each interest region into a plurality of interest sub-regions, and the position information of a predetermined number of edge points is obtained. Therefore, in each interest region, the number of interest sub-regions is the same as the predetermined number of edge points, and each interest sub-region corresponds to two edge points. The pixel mean value corresponding to each interest sub-region is determined as the color information of any one of the two edge points corresponding to the interest sub-region, so as to ensure that each edge point in the interest region has corresponding color information. As shown in Figure 6 the right graph of FIG. 1, Figure 6 the pixel mean value (i.e., the average RGB value) of the actually extracted interest region. By obtaining the edge point information of each interest region, the characteristics of each interest region can be accurately obtained.

[0076] In step S105, the edge point information of the continuous multiple frames of target contrast images is fused to obtain a fusion feature, and the ultrasound contrast image is classified based on the fusion feature to obtain the target type corresponding to the ultrasound contrast image.

[0077] Specifically, the feature length of each frame of the target contrast image is obtained, and the feature length is determined by the number of edge points in each frame of the target contrast image. The horizontal coordinates, vertical coordinates and color values of all edge points of each frame of the target contrast image are arranged in order according to different interest regions and placed in the corresponding positions to obtain the horizontal coordinate feature vector, vertical coordinate feature vector and color feature vector corresponding to each frame of the target contrast image. If the frame of the target contrast image does not have a corresponding interest sub-region, the corresponding position is set to zero. As shown in Figure 7 the right graph of FIG. 1, Figure 7 Taking a single frame of target contrast image as an example, it is composed of three rows of feature vectors. The first row is the horizontal coordinate feature vector, which is used to store the horizontal coordinates (x coordinates) of the edge points of each interest region in the target contrast image. The second row is the vertical coordinate feature vector, which is used to store the vertical coordinates (y coordinates) of the edge points of each interest region in the target contrast image. The third row is the color feature vector, which is used to store the color values (average color values of the interest sub-regions corresponding to each edge point) of the edge points of each interest region in the target contrast image, Figure 7Each small block in the three rows of feature vectors stores single edge point information, and if the target contrast image does not have a corresponding region of interest, the corresponding position is set to zero.

[0078] Specifically, as shown in Figure 8 The horizontal coordinate feature vector, the vertical coordinate feature vector and the color feature vector extracted from each frame of the target contrast image are fused respectively to obtain a three-channel matrix corresponding to the contrast ultrasound image. The three-channel matrix can be classified using a convolutional neural network (CNN) to obtain the target type corresponding to the contrast ultrasound image.

[0079] Specifically, as shown in Figure 9 The convolution-normalization-activation function block is stacked to form a classification network, and the size of the feature map is reduced to half of the original size after each block using maximum pooling. An FC layer is connected at the end of the network, and the output is subjected to softmax logistic regression as the final classification output.

[0080] Figure 2 Another flowchart of a target type recognition method according to an embodiment of the present application is given, which can include the following steps:

[0081] Step S201, obtaining a contrast ultrasound image.

[0082] Step S202, performing target feature analysis on continuous multiple frames of target contrast images in the contrast ultrasound image to determine the region of interest corresponding to each frame of the target contrast image.

[0083] In a possible implementation, first, continuous multiple frames of target contrast images and continuous multiple frames of target B-mode images are respectively extracted from the contrast ultrasound image, and the continuous multiple frames of target B-mode images correspond one-to-one to the continuous multiple frames of target contrast images. Then, a preset image resolution is obtained, and the continuous multiple frames of target contrast images and the continuous multiple frames of target B-mode images are respectively scaled based on the preset image resolution to obtain scaled target contrast images and scaled target B-mode images. The RGB channel of the scaled target contrast images and the grayscale channel of the scaled target B-mode images are obtained. Finally, the scaled target contrast images are subjected to instance segmentation processing based on the RGB channel and the grayscale channel to obtain the region of interest contained in each frame of the target contrast image.

[0084] Step S203, edges and centroids of each of the regions of interest are extracted from the target ultrasound contrast image, and the regions of interest are equally-angledly divided based on the centroids to obtain position information of a predetermined number of edge points on the edges.

[0085] Step S204, color information of the edge points is obtained based on pixel values of pixel points of each of the regions of interest to obtain edge point information of each of the target contrast images, the edge point information including the position information and the color information.

[0086] In a possible implementation, a pixel mean value of pixel values of pixel points of each of the regions of interest obtained after the regions of interest are equally-angledly divided is obtained respectively;

[0087] The pixel mean value of each of the regions of interest is determined as color information of a corresponding edge point. The pixel mean value of each of the regions of interest can be an average RGB value. After the edges of each of the regions of interest and the centroids of each of the regions of interest are extracted, each of the regions of interest is equally-angledly divided into a plurality of regions of interest based on image coordinates of the centroids of each of the regions of interest to obtain the position information of the predetermined number of edge points. Therefore, in each of the regions of interest, the number of regions of interest is the same as the predetermined number of edge points, and each of the regions of interest corresponds to two edge points. The pixel mean value of each of the regions of interest is determined as color information of any one of the two edge points corresponding to the region of interest, so as to ensure that each of the edge points in the region of interest has corresponding color information.

[0088] Step S205, fusion features are obtained by fusing the edge point information of the continuous multiple frames of target contrast images, and the ultrasound contrast image is classified based on the fusion features to obtain a target type corresponding to the ultrasound contrast image.

[0089] In a possible implementation, the position information of the edge points includes horizontal coordinates and vertical coordinates of the edge points, and a feature length of each of the target contrast images is obtained.

[0090] For each of the regions of interest of each of the target contrast images, the horizontal coordinates, the vertical coordinates, and the color values of the corresponding edge points are sequentially spliced based on the feature length to obtain a horizontal coordinate feature vector, a vertical coordinate feature vector, and a color feature vector corresponding to each of the target contrast images.

[0091] The horizontal coordinate feature vector, the vertical coordinate feature vector and the color feature vector of all the target contrast images are fused respectively to obtain a three-channel matrix corresponding to the ultrasound contrast image, so as to determine the fusion feature, and the ultrasound contrast image is classified based on the fusion feature, so as to obtain the target type corresponding to the ultrasound contrast image.

[0092] The embodiment of the present application provides a target type recognition device, as shown in the figure, Figure 10 The target type recognition device is applied to the target type recognition method provided by the embodiment of the present application, and includes the following steps:

[0093] The image acquisition module 101 is configured to acquire the ultrasound contrast image.

[0094] The region of interest acquisition module 102 is configured to analyze target features of continuous multiple frames of target contrast images in the ultrasound contrast image respectively, and determine a region of interest corresponding to each of the target contrast images.

[0095] The edge point position information acquisition module 103 is configured to acquire position information of an edge point of each of the regions of interest.

[0096] The edge point information acquisition module 104 is configured to acquire color information of the edge point based on a pixel value of a pixel point of each of the regions of interest, so as to obtain edge point information of each of the target contrast images, and the edge point information includes the position information and the color information.

[0097] The target type acquisition module 105 is configured to fuse the edge point information of the continuous multiple frames of target contrast images to obtain fusion feature, and classify the ultrasound contrast image based on the fusion feature, so as to obtain a target type corresponding to the ultrasound contrast image.

[0098] The edge point position information acquisition module 103 includes:

[0099] The edge acquisition unit is configured to extract an edge of each of the regions of interest from the target ultrasound contrast image.

[0100] The edge point position information acquisition unit is configured to extract a predetermined number of edge points from the extracted edge, so as to obtain position information of the edge point of each of the regions of interest.

[0101] The edge point position information acquisition unit includes:

[0102] The center of mass acquisition subunit is configured to extract a center of mass of each of the regions of interest from the target ultrasound contrast image.

[0103] An edge point position information obtaining subunit is configured to perform equi-angle division on the region of interest based on the centroid to obtain position information of a predetermined number of edge points.

[0104] The edge point information obtaining module 104 comprises:

[0105] A statistical analysis unit is configured to obtain pixel mean values of pixel values of pixel points of each sub-region of interest obtained after equi-angle division of the region of interest.

[0106] A color information obtaining unit is configured to determine the pixel mean values of the respective sub-regions of interest as color information of corresponding edge points.

[0107] The target type obtaining module 105 comprises:

[0108] A feature length obtaining unit is configured to obtain a feature length of each frame of the target contrast image.

[0109] A feature vector obtaining unit is configured to, for each region of interest of each frame of the target contrast image, sequentially splice the horizontal coordinate, the vertical coordinate and the color value of the corresponding edge points based on the feature length to obtain a horizontal coordinate feature vector, a vertical coordinate feature vector and a color feature vector corresponding to each frame of the target contrast image.

[0110] A fusion feature obtaining unit is configured to fuse the horizontal coordinate feature vectors, the vertical coordinate feature vectors and the color feature vectors of all the target contrast images respectively to obtain a three-channel matrix corresponding to the ultrasound contrast image to determine the fusion feature.

[0111] The region of interest obtaining module 102 comprises:

[0112] A target contrast image and target B-mode image obtaining unit is configured to extract a plurality of continuous frames of target contrast images and a plurality of continuous frames of target B-mode images from the ultrasound contrast image, the plurality of continuous frames of target B-mode images corresponding one-to-one to the plurality of continuous frames of target contrast images.

[0113] An image resolution obtaining unit is configured to preset an image resolution.

[0114] A scaling processing unit is configured to perform scaling processing on the plurality of continuous frames of target contrast images and the plurality of continuous frames of target B-mode images based on the preset image resolution to obtain scaled target contrast images and scaled target B-mode images.

[0115] A region of interest obtaining unit is configured to perform segmentation processing on the scaled target contrast images based on the scaled target B-mode images to determine regions of interest contained in each frame of the target contrast images.

[0116] The region of interest acquisition unit comprises:

[0117] A color channel acquisition subunit is configured to acquire an RGB channel of the scaled target contrast image and a grayscale channel of the scaled target B-mode image.

[0118] A region of interest acquisition subunit is configured to perform instance segmentation processing on the scaled target contrast image based on the RGB channel and the grayscale channel, and acquire a region of interest contained in each frame of the target contrast image.

[0119] The embodiment of the present application provides an ultrasonic device, such as Figure 11 As shown in the figure, the ultrasonic device comprises at least one processor and a memory connected with the at least one processor; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the steps of the target type identification method provided by the embodiment of the present application.

[0120] The processor can be a central processing unit (CPU). The processor can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or combinations thereof.

[0121] The memory is a non-transitory computer readable storage medium, which can be used to store non-transitory software programs, non-transitory computer executable programs and modules, such as program instructions / modules corresponding to the method in the embodiment of the present application. The processor performs various functional applications and data processing of the processor by running the non-transitory software programs, instructions and modules stored in the memory, that is, implements the method in the method embodiment.

[0122] The memory can include a program storage area and a data storage area, where the program storage area can store an operating system, application programs required by at least one function, and the like; and the data storage area can store data created by the processor and the like. In addition, the memory can include a high-speed random access memory, and can also include a non-transitory memory, such as at least one disk memory device, a flash memory device, or other non-transitory solid-state memory device. In some embodiments, the memory can optionally include a memory disposed remotely from the processor, which can be connected to the processor through a network. Examples of the above network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0123] The computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the steps of the target type identification method provided by the embodiments of the present application. The computer readable storage medium stores at least one computer program, and the at least one computer program is loaded and executed by the processor to implement all or part of the steps of the above method. For example, the computer readable storage medium can be a read-only memory (Read-Only Memory, ROM), a random access memory (Random Access Memory, RAM), a compact disc read-only memory (Compact Disc Read-Only Memory, CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, and the like.

[0124] The embodiment of the application discloses a target type identification method and device and ultrasonic equipment. The method comprises the following steps: acquiring an ultrasonic contrast image; analyzing the target feature of a plurality of continuous target contrast images in the ultrasonic contrast image respectively, and determining the region of interest corresponding to each target contrast image; acquiring the position information and color information of the edge point in each region of interest respectively to obtain the edge point information of each target contrast image; fusing the edge point information of the plurality of continuous target contrast images to obtain fusion features, and performing classification processing on the ultrasonic contrast image based on the fusion features to acquire the target type corresponding to the ultrasonic contrast image. The method solves the problem that when the ultrasonic contrast technology is used to identify the target type, the doctor needs to identify and analyze the features of the ultrasonic contrast image manually, which consumes a large amount of time and manpower, and the identification result is extremely easy to be wrong when the doctor is tired. The target feature of the target contrast image in the ultrasonic contrast image is analyzed to determine the region of interest, the calculation amount is saved, the identification range is reduced, and the region to be identified is accurately acquired; the edge point information of each region of interest is acquired to accurately obtain the feature information of each target contrast image, and the accuracy of target type identification is improved; the edge point information is fused to obtain fusion features, and the calculation amount is further saved; the ultrasonic contrast image is classified and processed based on the fusion features, the target type corresponding to the ultrasonic contrast image is accurately identified, and the manpower is saved.

[0125] The application is described in detail above in combination with the specific embodiments and exemplary examples, but these descriptions cannot be understood as limitations of the application. Those skilled in the art understand that the technical scheme of the application and the embodiments thereof can be variously replaced, modified or improved without departing from the spirit and scope of the application, and these all fall within the scope of the application. The protection scope of the application is subject to the appended claims.

Claims

1. A target type identification method characterized by, The method comprises: acquiring an ultrasound contrast image; performing target feature analysis on each of a plurality of continuous target contrast images in the ultrasound contrast image to determine a region of interest corresponding to each of the target contrast images; acquiring position information of edge points of each of the regions of interest; based on pixel values of pixel points of each of the regions of interest, acquiring color information of the edge points to obtain edge point information of each of the target contrast images, the edge point information comprising the position information and the color information; fusing the edge point information of the plurality of continuous target contrast images to obtain fusion features, and performing classification processing on the ultrasound contrast image based on the fusion features to acquire a target type corresponding to the ultrasound contrast image.

2. The method of claim 1, wherein, The acquiring of the position information of the edge points of each of the regions of interest comprises: extracting edges of each of the regions of interest from the target ultrasound contrast image; extracting a predetermined number of edge points from the extracted edges to obtain the position information of the edge points of each of the regions of interest.

3. The method of claim 2, wherein, The extracting of the predetermined number of edge points from the extracted edges to obtain the position information of the edge points of each of the regions of interest comprises: extracting a centroid of each of the regions of interest from the target ultrasound contrast image; performing equi-angle division on the region of interest based on the centroid to obtain the position information of the predetermined number of edge points.

4. The method of claim 3, wherein, The acquiring of the color information of the edge points based on the pixel values of the pixel points of each of the regions of interest comprises: acquiring pixel mean values of pixel values of pixel points of each of the regions of interest obtained after equi-angle division of the regions of interest; determining the pixel mean values corresponding to each of the regions of interest as color information of the corresponding edge points.

5. The method of claim 1, wherein, The position information of the edge points comprises horizontal coordinates and vertical coordinates of the edge points, and the fusing of the edge point information of the plurality of continuous target contrast images to obtain the fusion features comprises: acquiring a feature length of each of the target contrast images; for each of the regions of interest of each of the target contrast images, sequentially concatenating horizontal coordinates, vertical coordinates and color values of the corresponding edge points based on the feature length to obtain a horizontal coordinate feature vector, a vertical coordinate feature vector and a color feature vector corresponding to each of the target contrast images; fusing the horizontal coordinate feature vectors, the vertical coordinate feature vectors and the color feature vectors of all of the target contrast images to obtain a three-channel matrix corresponding to the ultrasound contrast image to determine the fusion features.

6. The method according to any one of claims 1-5, characterized in that, The performing of target feature analysis on each of the plurality of continuous target contrast images in the ultrasound contrast image to determine a region of interest corresponding to each of the target contrast images comprises: separately extracting a plurality of continuous target contrast images and a plurality of continuous target B-mode images from the ultrasound contrast image, the plurality of continuous target B-mode images corresponding one-to-one to the plurality of continuous target contrast images; acquiring a preset image resolution; scaling processing is performed on the continuous multiple target contrast images and the continuous multiple target B mode images based on the preset image resolution, to obtain scaled target contrast images and scaled target B mode images; segmentation processing is performed on the scaled target contrast images based on the scaled target B mode images, to determine a region of interest contained in each of the target contrast images.

7. The method of claim 6, wherein, The segmentation processing performed on the scaled target contrast images based on the scaled target B mode images, to determine a region of interest contained in each of the target contrast images, includes: An RGB channel of the scaled target contrast images and a grayscale channel of the scaled target B mode images are obtained. Instance segmentation processing is performed on the scaled target contrast images based on the RGB channel and the grayscale channel, to obtain a region of interest contained in each of the target contrast images.

8. A target type identification apparatus characterized by comprising: The method is applied to any one of claims 1-7, and includes: An image acquisition module is configured to acquire an ultrasound contrast image; A region of interest acquisition module is configured to analyze target features of continuous multiple target contrast images in the ultrasound contrast image, to determine a region of interest corresponding to each of the target contrast images; An edge point position information acquisition module is configured to acquire position information of an edge point of each of the regions of interest; An edge point information acquisition module is configured to acquire color information of the edge point based on pixel values of pixel points of each of the regions of interest, to obtain edge point information of each of the target contrast images, the edge point information including the position information and the color information; A target type acquisition module is configured to fuse the edge point information of the continuous multiple target contrast images to obtain fusion features, and perform classification processing on the ultrasound contrast image based on the fusion features, to obtain a target type corresponding to the ultrasound contrast image.

9. An ultrasound apparatus, characterized by includes: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to cause the at least one processor to perform the steps of the target type identification method of any one of claims 1-7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the target type identification method of any one of claims 1-7.

Images