Method and device for rapid analysis of carotid artery plaque based on image intelligent analysis

Abstract

The present invention discloses a method and device for rapid analysis of carotid plaques based on image intelligent analysis. The method comprises: extracting sample image features from sample images of each sample of carotid plaque image information, extracting sample basic features of each sample from a sample database and iteratively training an initial analysis model in combination with the sample image features, respectively extracting initial individual image features and initial individual basic features from the input individual information to be analyzed, and inputting the trained target analysis model to obtain a carotid plaque analysis result containing a risk probability. Through the above method, sample image features and sample basic features can be extracted to iteratively train the initial analysis model, and the carotid plaque analysis result containing a risk probability can be accurately obtained based on the trained target analysis model, which greatly improves the accuracy of risk analysis of carotid plaque images.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method and device for rapid analysis of carotid artery plaques based on image intelligent analysis. Background Art

[0002] Ischemic stroke, also known as cerebral infarction, refers to ischemic necrosis or softening of focal brain tissue caused by ischemia and hypoxia due to impaired blood supply to the brain. At present, a large number of clinical and basic medical studies have shown that unstable carotid plaque rupture and secondary thrombosis are the main causes of acute ischemic stroke. In the existing technology, ultrasound imaging of stable carotid plaques can usually obtain relatively accurate judgment results; however, the ultrasound imaging manifestations of unstable carotid plaques are diverse, and it is difficult for existing technical methods to accurately analyze carotid plaque images with the naked eye, resulting in the inability to accurately analyze whether there is a risk in the carotid plaque image. Therefore, there is a problem in the existing technical methods that the risk of carotid plaque images cannot be accurately analyzed. Summary of the invention

[0003] The embodiments of the present invention provide a method and device for rapid analysis of carotid artery plaques based on intelligent image analysis, aiming to solve the problem in the prior art methods that the risk analysis of carotid artery plaque images cannot be accurately performed.

[0004] In a first aspect, an embodiment of the present invention provides a method for rapid analysis of carotid artery plaques based on image intelligent analysis, the method comprising:

[0005] Receiving input carotid artery plaque image information, and extracting sample image features corresponding to each sample from sample images of the carotid artery plaque image information according to a preset image feature extraction model;

[0006] Acquire sample basic features corresponding to each sample from the sample database according to preset basic feature acquisition rules;

[0007] Iteratively training the initial analysis model according to a preset model training strategy, the sample image features and the sample individual features to obtain a target analysis model;

[0008] If the input individual information to be analyzed is received, extracting the corresponding initial individual image features from the carotid artery plaque image to be analyzed of the individual information to be analyzed according to the image feature extraction model;

[0009] Acquire the corresponding initial individual basic features from the individual basic data of the individual information to be analyzed according to the basic feature acquisition rule;

[0010] The initial individual image features and the initial individual basic features are analyzed according to the target analysis model to obtain a carotid artery plaque analysis result including a risk probability.

[0011] In a second aspect, an embodiment of the present invention provides a device for rapid analysis of carotid artery plaques based on image intelligent analysis, wherein the device is used to perform the method for rapid analysis of carotid artery plaques based on image intelligent analysis as described in the first aspect above, and the device comprises:

[0012] A sample image feature acquisition unit, configured to receive input carotid artery plaque image information, and extract sample image features corresponding to each sample from the sample image of the carotid artery plaque image information according to a preset image feature extraction model;

[0013] A sample basic feature acquisition unit, used to acquire the sample basic feature corresponding to each sample from the sample database according to a preset basic feature acquisition rule;

[0014] A model training unit, used for iteratively training the initial analysis model according to a preset model training strategy, the sample image features and the sample individual features, so as to obtain a target analysis model;

[0015] An initial individual image feature acquisition unit is used to extract corresponding initial individual image features from the carotid artery plaque image to be analyzed of the individual information to be analyzed according to the image feature extraction model upon receiving the input individual information to be analyzed;

[0016] An initial individual basic feature acquisition unit, configured to acquire corresponding initial individual basic features from the individual basic data of the individual information to be analyzed according to the basic feature acquisition rule;

[0017] The analysis result acquisition unit is used to analyze the initial individual image features and the initial individual basic features according to the target analysis model to obtain a carotid plaque analysis result including a risk probability.

[0018] In a third aspect, an embodiment of the present invention further provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer device executes the computer program, the method for rapid analysis of carotid plaques based on image intelligent analysis as described in the first aspect above is implemented.

[0019] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the method for rapid analysis of carotid plaque based on image intelligent analysis as described in the first aspect above.

[0020] The embodiment of the present invention provides a method and device for rapid analysis of carotid plaques based on intelligent image analysis. The method includes: extracting sample image features from sample images of each sample of carotid plaque image information, extracting sample basic features of each sample from a sample database and iteratively training an initial analysis model in combination with the sample image features, respectively extracting initial individual image features and initial individual basic features from the input individual information to be analyzed, and inputting the trained target analysis model to obtain a carotid plaque analysis result containing a risk probability. Through the above method, sample image features and sample basic features can be extracted to iteratively train the initial analysis model, and the carotid plaque analysis result containing a risk probability can be accurately obtained based on the trained target analysis model, which greatly improves the accuracy of risk analysis of carotid plaque images. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0022] Figure 1 A schematic diagram of a process for rapid analysis of carotid artery plaques based on intelligent image analysis provided by an embodiment of the present invention;

[0023] Figure 2 A schematic diagram of an application scenario of a method for rapid analysis of carotid artery plaques based on intelligent image analysis provided by an embodiment of the present invention;

[0024] Figure 3 This is an application effect diagram of the method for rapid analysis of carotid artery plaques based on intelligent image analysis provided by an embodiment of the present invention;

[0025] Figure 4 A schematic block diagram of a device for rapid analysis of carotid artery plaques based on intelligent image analysis provided by an embodiment of the present invention;

[0026] Figure 5 A schematic block diagram of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0029] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0030] It should be further understood that the term "and / or" used in the present description and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0031] See also Figure 1 and Figure 2 , Figure 1 A schematic diagram of a process for rapid analysis of carotid artery plaques based on intelligent image analysis provided by an embodiment of the present invention, Figure 2 Schematic diagram of the application scenario of the method for rapid analysis of carotid plaques based on image intelligent analysis provided in an embodiment of the present invention; the method for rapid analysis of carotid plaques based on image intelligent analysis is applied to a terminal device 10, and the terminal device 10 establishes a network connection with an ultrasound device 20 to achieve data information transmission; the method for rapid analysis of carotid plaques based on image intelligent analysis is executed by application software installed in the terminal device 10; the terminal device 10 is a terminal device for executing the method for rapid analysis of carotid plaques based on image intelligent analysis to analyze the risk probability in a static image of a carotid plaque, such as a desktop computer, a laptop computer, a tablet computer or a mobile phone; the ultrasound device 20 is a device for collecting ultrasound images of the human neck, so as to obtain a two-dimensional static image of the longitudinal section of the maximum diameter of the carotid plaque. A wired connection or a wireless connection is established between the terminal device 10 and the ultrasound device 20 to achieve data information transmission, wherein the wireless connection can be a Bluetooth wireless connection or a WiFi wireless connection, etc. Figure 1 As shown, the method includes steps S110 to S160.

[0032] S110, receiving input carotid artery plaque image information, and extracting sample image features corresponding to each sample from sample images of the carotid artery plaque image information according to a preset image feature extraction model.

[0033] The ultrasound device can collect static images of carotid plaques of different individuals. One or more static images of carotid plaques can be collected corresponding to one carotid cross-sectional position of one individual. Then, one individual can obtain static images of carotid plaques corresponding to each carotid cross-sectional position at different carotid cross-sectional positions. The carotid plaque image information includes sample images of multiple samples, each sample corresponds to one individual, and the sample image of each sample can be one or more static images of carotid plaques. Sample image features of each sample can be extracted from the sample image according to the image feature extraction model, and each sample corresponds to a set of sample image features; the image feature extraction model is also an intelligent model for extracting features from the sample image, and the obtained sample image features include feature values ​​corresponding to multiple index items, wherein the index items include plaque echo type, plaque boundary regularity, plaque morphology, sign type, total plaque volume, vascular lumen stenosis rate, etc.

[0034] In a specific embodiment, the above-mentioned step S110 specifically includes the following sub-steps: performing contour extraction on each sample image according to the contour extraction rules in the image feature extraction model to obtain the image contour corresponding to each sample image; identifying the image contour of each sample image according to the contour feature extraction rules in the image feature extraction model to obtain corresponding contour type identification information; integrating the contour type identification information of each sample image of the sample to obtain the corresponding sample image feature.

[0035] Specifically, the contour extraction rules in the image feature extraction model can be used to extract the contour of each sample image, and the contrast information between the pixel point and the adjacent pixel points in each sample image can be calculated. The contrast information can be used to reflect the difference information between the pixel point and the adjacent pixel points. The calculation formula for obtaining the contrast information can be shown as formula (1):

[0036] (1);

[0037] Among them, S is the calculated contrast information corresponding to the central pixel point P0, P0 is the pixel value of the central pixel point (the pixel value includes the pixel values ​​of the three channels of RGB), n is the number of pixels in the first circle outside the pixel point P0, and P i is the pixel value of the i-th pixel in the first circle outside the pixel point P0, e is the base of the natural logarithm, and m is the number of pixels in the second circle outside the pixel point P0.

[0038] According to the contrast information and the point and line information in the sample image, some pixel points with large contrast information and connected in a line are extracted from the sample image as the corresponding image contour.

[0039] Furthermore, the image contour of each sample image is identified according to the contour feature extraction rules in the image feature extraction model to obtain contour type identification information. Then, each sample image can obtain a corresponding set of contour type identification information, and a set of contour type identification information also includes indicator feature information corresponding to multiple indicator items.

[0040] The contour recognition information of each sample image of the sample is integrated, so as to integrate multiple indicator feature information corresponding to the same indicator item of a sample, and obtain the indicator feature integration information uniquely corresponding to each indicator item as the sample image feature of the sample. The multiple indicator feature information corresponding to the same indicator item of the sample (each image corresponds to one indicator feature information, and multiple images of the same indicator item correspond to multiple indicator feature information) can be averaged to obtain the unique corresponding indicator feature integration information.

[0041] In a specific embodiment, the image contour of each of the sample images is identified according to the contour feature extraction rules in the image feature extraction model to obtain corresponding contour type identification information, including: determining the image contour matching each indicator item according to the resolution range corresponding to each indicator item in the contour feature extraction rule and the resolution of the sample image corresponding to the image contour; identifying the image contour corresponding to each indicator item according to the feature extraction strategy corresponding to each indicator item in the contour feature extraction rule, thereby obtaining contour type identification information corresponding to each image contour.

[0042] Specifically, the resolution interval corresponding to each indicator item can be matched with the resolution of the sample image corresponding to each image contour. If the resolution of the sample image is within the resolution interval of the indicator item, it is determined that the image contour of the sample image matches the indicator item. The image contours that match each indicator item can be determined through the above matching judgment. For example, for indicator items such as plaque echo type and total plaque volume, the resolution interval is small, that is, the indicator feature information of the corresponding indicator item can be accurately obtained without a very fine image; for indicator items such as plaque boundary regularity, plaque morphology, and sign type, the resolution interval is large, that is, it is necessary to obtain finer local detail features in the image to accurately obtain the indicator feature information of the corresponding indicator item.

[0043] According to the feature extraction strategy corresponding to each indicator item in the contour feature extraction rule, the image contour corresponding to each indicator item is identified. The pixel information corresponding to each contour area can be extracted from the sample image through the image contour, where the contour area contained in the image contour is as follows: Figure 3As shown, 11 is the carotid contour area, 12 is the plaque contour area, 13 is the fibrous cap area, 14 is the lipid core area, and 15 is the cortical contour. The pixel values ​​of the pixels contained in different contour areas are different. For example, the carotid contour area is usually red, and a crescent-shaped lipid core area is formed outside the carotid contour area. The lipid core contained in the lipid core area is usually light yellow particles. There may be one or more plaque contour areas in the area adjacent to the carotid contour area. There may also be no plaque contour area in the carotid contour area or the lipid core area. The plaque contour area has dark yellow to brown plaques; the carotid contour area and the plaque contour area are separated by a light pink fibrous cap area.

[0044] For the indicator item of plaque echo type, the pixel information of the plaque contour area can be obtained, and the average value of the pixel value of the plaque contour area can be further calculated based on the pixel information of the plaque contour area, and the depth (i.e., brightness) corresponding to the average value of the pixel value of the plaque contour area on the red, yellow, and brown channels configured in the corresponding feature extraction strategy can be determined to obtain the indicator feature information corresponding to the plaque echo type. For the plaque boundary regularity, the gradient direction corresponding to each pixel point of the contour line in the plaque contour area can be obtained according to the corresponding feature extraction strategy, and the angle between the gradient directions of adjacent pixel points can be calculated and the average value of the obtained angle can be calculated as the indicator feature information corresponding to the plaque boundary regularity. For the indicator item of plaque morphology, the pixel information of the plaque contour area can be obtained, and the standard deviation of the pixel value of the plaque contour area can be calculated according to the corresponding feature extraction strategy as the indicator feature information corresponding to the plaque morphology. For the indicator item of sign type, the pixel average value of the plaque contour area can be calculated according to the pixel information of the plaque contour area; the pixel information of the carotid artery contour area is obtained and the pixel average value of the carotid artery contour area is calculated, and the difference between the pixel average value of the plaque contour area and the pixel average value of the carotid artery contour area is calculated according to the corresponding feature extraction strategy, as the indicator feature information corresponding to the sign type. For the indicator item of total plaque volume, the area value of the plaque contour area can be calculated according to the pixel information of the plaque contour area and the cumulative calculation is performed to obtain the indicator feature information corresponding to the total plaque volume; for the indicator item of vascular lumen stenosis rate, the difference between the area value of the carotid artery contour area and the total plaque volume can be calculated according to the corresponding feature extraction strategy and the pixel information of the carotid artery contour area, and the ratio between the difference and the cross-sectional area of ​​the peripheral cortical contour in the image contour is calculated to obtain the indicator feature information corresponding to the vascular lumen stenosis rate.

[0045] S120: Obtain sample basic features corresponding to each sample from a sample database according to preset basic feature acquisition rules.

[0046] Furthermore, the sample basic features of each sample can be obtained from the sample database according to the basic feature acquisition rules. The sample database contains basic data corresponding to each sample, and the basic data of each sample includes age, gender, medical history, blood pressure value, blood sugar value and blood lipid value, etc. The characteristic values ​​corresponding to each item of data in the basic data can be obtained through the basic feature acquisition rules, and the characteristic values ​​of each item of data are combined into the sample basic features corresponding to a sample.

[0047] Specifically, for the numerical data such as blood pressure value and blood sugar value in the basic data of a sample, it can be converted into a feature value accordingly. For example, the specific conversion calculation formula can be expressed by formula (2):

[0048] (2);

[0049] Among them, s is the numerical data, s0 is the corresponding conversion coefficient, and t is the final eigenvalue. For non-numerical data such as gender and medical history in the basic data, they can be converted into eigenvalues ​​according to the mapping relationship of the basic feature acquisition rules. The mapping relationship can map the type information to the eigenvalue.

[0050] S130, iteratively training the initial analysis model according to a preset model training strategy, the sample image features and the sample individual features to obtain a target analysis model.

[0051] In a specific embodiment, the above-mentioned step S130 specifically includes the following sub-steps: combining the sample image features and sample individual features of each sample into corresponding sample comprehensive features; inputting the sample comprehensive features of each sample into the initial analysis model for analysis to obtain the predicted probability value corresponding to each sample; calculating the loss value corresponding to the type labels and predicted probability values ​​of all samples according to a preset loss value calculation formula; calculating the learning rate corresponding to the current number of iterations according to a preset learning rate calculation formula, so as to iteratively optimize the parameter values ​​in the initial analysis model based on the learning rate to obtain an optimized analysis model; judging whether the loss value meets the preset end training condition; if the loss value meets the end training condition, terminating the training process and obtaining the optimized analysis model as the corresponding target analysis model; if the loss value does not meet the end training condition, using the optimized analysis model as the initial analysis model and returning to execute the step of inputting the sample comprehensive features of each sample into the initial analysis model for analysis.

[0052] The sample image features and individual features of the sample are combined to obtain the sample comprehensive features, and the sample comprehensive features can be a 1×M dimensional array. The sample comprehensive features of each sample can be input into the initial analysis model for analysis; the initial analysis model consists of an input layer, an intermediate layer, and an output layer. The number of input nodes contained in the input layer is equal to the number of eigenvalues ​​contained in the sample comprehensive features, and each input node can correspond to an input eigenvalue. The intermediate layer is composed of multiple groups of intermediate nodes, and the intermediate nodes are used to establish the association relationship between the input layer and the output layer. After the association analysis is performed through the intermediate layer, the corresponding predicted probability value can be output through the output node in the output layer. Then, only one output node is configured in the initial analysis model.

[0053] According to the loss value calculation formula, the loss values ​​corresponding to the type labels and predicted probability values ​​of all samples are calculated. Specifically, the loss value calculation formula can be expressed as formula (3):

[0054] (3);

[0055] Among them, n is the total number of samples, y i is the type label of the i-th sample, 0 is a negative sample (no risk), 1 is a positive sample (risky), y i The value of can only be 0 or 1; is the predicted probability value of the i-th sample, with a value range of [0,1], and LS is the calculated loss value corresponding to the current training process.

[0056] In order to accelerate model convergence and make the training results as accurate as possible, the embodiment of the present application adopts a learning rate calculation formula to calculate a variable learning rate for training. The learning rate calculation formula can be expressed as formula (4):

[0057] (4);

[0058] Among them, v is the current iteration number (the number of times training has been performed), lr is the initial learning rate (when v=0), and lr' is the learning rate calculated in the subsequent iterative training process (v≠0). The parameter values ​​in the initial analysis model can be iteratively optimized according to the calculated learning rate corresponding to the current iteration number. The larger the learning rate value, the larger the adjustment range of the parameter values ​​in the initial analysis model for optimization; the smaller the learning rate value, the smaller the adjustment range of the parameter values ​​for optimization.

[0059] Furthermore, it can be determined whether the loss value calculated during the current training process meets the preset end training condition; specifically, the end training condition can be set with the number of historical loss value comparisons, and the loss value obtained from the previous training is obtained according to the number of historical loss value comparisons, and compared with the loss value obtained from the current training. For example, if the number of historical loss value comparisons is set to 3, the loss values ​​calculated during the last three training processes can be obtained and the average value can be calculated as the corresponding historical loss average value; it can be determined whether the loss value calculated during the current training process is not less than the historical loss average value, so as to determine whether the end training condition is met. If the loss value of this training is not less than the historical loss average value, it is determined that the end training condition is met; otherwise, it is determined that the end training condition is not met.

[0060] If the training end conditions are met, the training process is terminated and the optimized analysis model obtained by the current training is obtained as the corresponding target analysis model; if the training end conditions are not met, the optimized analysis model obtained by the current training can be used as the initial analysis model again and fed back to the above steps to realize model iterative training.

[0061] Furthermore, before terminating the training process and acquiring the optimized analysis model as the corresponding target analysis model, it also includes: inputting each group of verification features in the pre-stored verification data set into the optimized analysis model respectively to obtain the predicted probability value corresponding to each verification feature; calculating the corresponding accuracy rate according to the predicted probability value and the verification type label of each verification feature; judging whether the accuracy rate is not less than a preset verification accuracy rate threshold; if the accuracy rate is not less than the verification accuracy rate threshold, executing the step of terminating the training process and acquiring the optimized analysis model as the corresponding target analysis model.

[0062] In order to improve the accuracy of the risk analysis of the final target analysis model, before determining the optimized analysis model as the corresponding target analysis model, each group of verification features contained in the pre-stored verification data set can be input into the optimized analysis model respectively, so as to obtain the predicted probability value obtained by the optimized analysis model for each group of verification features. The corresponding accuracy is further calculated based on the predicted probability value and the verification type label of each verification feature. The specific calculation process can be expressed as formula (5):

[0063] (5);

[0064] Among them, Z is the calculated accuracy, q is the number of verification features contained in the verification data set, and y j is the type label of the validation feature of the jth group; is the predicted probability value of the validation feature of the jth group.

[0065] It is determined whether the obtained accuracy is not less than the verification accuracy threshold. If it is, the verification is determined to be passed and the subsequent steps are continued; if it is less than, the verification is determined to be failed, and a prompt message can be issued to remind the management personnel to iteratively train the analysis model again through a new set of carotid plaque image information.

[0066] S140: If the input individual information to be analyzed is received, extract corresponding initial individual image features from the carotid artery plaque image to be analyzed of the individual information to be analyzed according to the image feature extraction model.

[0067] The individual information to be analyzed can be input into the terminal device, and the individual information to be analyzed includes the carotid plaque image to be analyzed and the individual basic data. The carotid plaque image to be analyzed can be collected by an ultrasound device and input into the terminal device, and the individual basic data can be directly input into the terminal device by the management personnel. The corresponding initial individual image features can be extracted from the carotid plaque image to be analyzed by the image feature extraction model, and the specific feature extraction method is the same as step S110.

[0068] S150: Acquire corresponding initial individual basic features from the individual basic data of the individual information to be analyzed according to the basic feature acquisition rule.

[0069] Furthermore, the corresponding initial individual basic features may be acquired from the individual basic data according to the basic feature acquisition rule, and the specific acquisition method is the same as step S120.

[0070] S160: Analyze the initial individual image features and the initial individual basic features according to the target analysis model to obtain a carotid plaque analysis result including a risk probability.

[0071] The initial individual image features and initial individual basic features are analyzed according to the target analysis model. The initial individual image features and initial individual basic features can be combined and input into the target analysis model, and the corresponding risk probability is obtained through the target analysis model as the carotid plaque analysis result corresponding to the individual information to be analyzed. A high risk probability indicates that the corresponding carotid plaque image has a high risk; a low risk probability indicates that the corresponding carotid plaque image has a low risk.

[0072] Specifically, before using the image feature extraction model, the initially set image feature extraction model can also be tuned through the sample images of the carotid plaque image information. The specific steps include: extracting the initial image features corresponding to each sample from the sample images of the carotid plaque image information according to the pre-stored initial image feature extraction model; grouping the initial image features according to the type labels of each sample to obtain the initial image feature sets corresponding to each type of label; performing difference calculation on each of the initial image feature sets according to the preset difference calculation rules to obtain the difference information corresponding to different types of labels on multiple feature items; filtering the corresponding target feature items from the feature items included in the multiple indicator items according to the preset index item screening rules and the difference information; assigning corresponding weight coefficients to each of the target feature items according to the preset weight allocation strategy and the difference information; configuring and updating the contour feature extraction rules in the initial image feature extraction model according to the weight coefficients of each of the target feature items to obtain the image feature extraction model.

[0073] The feature extraction strategy set by the contour feature extraction rule in the initial image feature extraction model covers more index items than the feature items covered by the image feature extraction model finally used. A corresponding set of initial image features can be extracted from the sample images contained in each sample, and the number of index feature information in the initial image features extracted using the initial image feature extraction model is also greater. For the specific extraction method, refer to step S110, which will not be described in detail here.

[0074] Furthermore, each sample corresponds to a type label, and the type label is risky or non-risky, and all samples can be divided into relevant groups and non-relevant groups according to the type label. Furthermore, the initial image features of each sample are grouped according to the grouping of the samples, so as to obtain an initial image feature set corresponding to the samples of the relevant group and an initial image feature set corresponding to the samples of the non-relevant group.

[0075] According to the difference calculation rules, the difference calculation is performed on each initial image feature set, wherein the difference between two types of labels on multiple feature items can be calculated to obtain the corresponding difference information. Since the index item contains one or more feature items, in order to reflect the difference between different types of labels, it is necessary to perform a difference calculation for each feature item separately, so as to obtain the difference coefficient corresponding to each feature item, and the difference information can be obtained by obtaining the difference coefficient of each feature item. For example, the plaque echo type index item includes the corresponding depths on the red, yellow, and brown channels, that is, the index item contains three feature items, and the corresponding difference coefficients need to be calculated for the three feature items.

[0076] Specifically, the difference calculation can be obtained by formula (6):

[0077] (6);

[0078] Where C is the calculated difference coefficient, f1 is the mean eigenvalue of the feature item corresponding to the first initial image feature set, f2 is the mean eigenvalue of the feature item corresponding to the second initial image feature set, σ1 2 is the variance of the eigenvalues ​​corresponding to the feature item and the first initial image feature set, σ2 2 is the variance of the eigenvalues ​​corresponding to the feature item and the second initial image feature set.

[0079] Furthermore, the characteristic items included in the multiple indicator items can be screened according to the indicator screening rule and the obtained difference information, so as to obtain the target characteristic item. Specifically, the indicator screening rule is configured with a difference coefficient threshold, and it can be determined whether the difference coefficient of each characteristic item in the difference information is greater than the difference coefficient threshold. If it is greater, the characteristic item is determined to be the target characteristic item, and if it is not greater, the characteristic item is determined not to be the target characteristic item.

[0080] Furthermore, a corresponding weight coefficient can be assigned to each target feature item according to the weight allocation strategy and difference information. The weight coefficient is used to quantify the importance of each feature value in the image features extracted later. Specifically, the weight coefficient can be calculated by the following formula (7):

[0081] (7);

[0082] Among them, C k is the difference coefficient of the kth target feature item, G is the total number of target feature items, r k is the weight coefficient corresponding to the kth target feature item.

[0083] The contour feature extraction rules in the initial image feature extraction model can be configured and updated according to the weight coefficient of the target feature item, and the updated image feature extraction model can be obtained after the configuration update.

[0084] In a specific embodiment, the above-mentioned specific steps of performing configuration update include the following sub-steps: updating the feature extraction strategy included in the contour feature extraction rule according to the target feature item to obtain an updated contour feature extraction rule; configuring the extraction weight value of each feature extraction strategy in the updated contour feature extraction rule according to the weight coefficient of the target feature item to obtain an updated image feature extraction model.

[0085] Specifically, the feature extraction strategies included in the contour feature extraction rules may be updated according to the target feature items, that is, only the feature extraction strategies corresponding to the target feature items are retained, and the remaining feature extraction strategies not corresponding to the target feature items are deleted.

[0086] Furthermore, the extraction weight value of each feature extraction strategy in the updated contour feature extraction rule is configured according to the weight coefficient of the target feature item, thereby updating the contour feature extraction rule and obtaining an updated image feature extraction model.

[0087] In the method for rapid analysis of carotid plaques based on intelligent image analysis provided in an embodiment of the present invention, the method includes: extracting sample image features from sample images of each sample of carotid plaque image information, extracting sample basic features of each sample from a sample database and iteratively training an initial analysis model in combination with the sample image features, respectively extracting initial individual image features and initial individual basic features from the input individual information to be analyzed, and inputting the trained target analysis model to obtain a carotid plaque analysis result containing a risk probability. Through the above method, sample image features and sample basic features can be extracted to iteratively train the initial analysis model, and the carotid plaque analysis result containing a risk probability can be accurately obtained based on the trained target analysis model, which greatly improves the accuracy of risk analysis of carotid plaque images.

[0088] The embodiment of the present invention also provides a device for rapid analysis of carotid artery plaques based on image intelligent analysis, which can be configured in the terminal device 10 and is used to execute any embodiment of the aforementioned method for rapid analysis of carotid artery plaques based on image intelligent analysis. Figure 4 , Figure 4 A schematic block diagram of a device for rapid analysis of carotid artery plaque based on intelligent image analysis provided in an embodiment of the present invention.

[0089] like Figure 4 As shown, the rapid analysis device 100 for carotid plaque based on image intelligent analysis includes a sample image feature acquisition unit 110, a sample basic feature acquisition unit 120, a model training unit 130, an initial individual image feature acquisition unit 140, an initial individual basic feature acquisition unit 150 and an analysis result acquisition unit 160.

[0090] The sample image feature acquisition unit 110 is used to receive input carotid artery plaque image information, and extract sample image features corresponding to each sample from the sample images of the carotid artery plaque image information according to a preset image feature extraction model.

[0091] The sample basic feature acquisition unit 120 is used to acquire the sample basic feature corresponding to each sample from the sample database according to a preset basic feature acquisition rule.

[0092] The model training unit 130 is used to iteratively train the initial analysis model according to a preset model training strategy, the sample image features and the sample individual features to obtain a target analysis model.

[0093] The initial individual image feature acquisition unit 140 is used to extract the corresponding initial individual image features from the carotid artery plaque image to be analyzed of the individual information to be analyzed according to the image feature extraction model if the input individual information to be analyzed is received.

[0094] The initial individual basic feature acquisition unit 150 is used to acquire the corresponding initial individual basic feature from the individual basic data of the individual information to be analyzed according to the basic feature acquisition rule.

[0095] The analysis result acquisition unit 160 is used to analyze the initial individual image features and the initial individual basic features according to the target analysis model to obtain a carotid artery plaque analysis result including a risk probability.

[0096] The rapid analysis device for carotid plaques based on image intelligent analysis provided in the embodiment of the present invention applies the rapid analysis method for carotid plaques based on image intelligent analysis, which includes: extracting sample image features from the sample images of each sample of carotid plaque image information, extracting sample basic features of each sample from the sample database and iteratively training the initial analysis model in combination with the sample image features, respectively extracting initial individual image features and initial individual basic features from the input individual information to be analyzed, and inputting the trained target analysis model to obtain the carotid plaque analysis results containing risk probability. Through the above method, the sample image features and sample basic features can be extracted to iteratively train the initial analysis model, and the carotid plaque analysis results containing risk probability can be accurately obtained based on the trained target analysis model, which greatly improves the accuracy of risk analysis of carotid plaque images.

[0097] The above-mentioned method for rapid analysis of carotid artery plaque based on image intelligent analysis can be implemented in the form of a computer program, and the device for rapid analysis of carotid artery plaque based on image intelligent analysis can be implemented as a computer device. The computer program can be used in Figure 5 The computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor; when the computer device executes the computer program, the method for rapid analysis of carotid artery plaques based on image intelligent analysis as described in the above embodiment is implemented.

[0098] See also Figure 5 , Figure 5 is a schematic block diagram of a computer device provided by an embodiment of the present invention. The computer device may be a terminal device for executing a rapid analysis method for carotid plaques based on image intelligent analysis to analyze the risk probability in a static image of carotid plaques.

[0099] See also Figure 5 The computer device 500 includes a processor 502 , a memory and a network interface 505 connected via a system bus 501 , wherein the memory may include a storage medium 503 and an internal memory 504 .

[0100] The storage medium 503 can store an operating system 5031 and a computer program 5032. When the computer program 5032 is executed, the processor 502 can execute a method for rapid analysis of carotid plaques based on image intelligent analysis, wherein the storage medium 503 can be a volatile storage medium or a non-volatile storage medium.

[0101] The processor 502 is used to provide computing and control capabilities to support the operation of the entire computer device 500 .

[0102] The internal memory 504 provides an environment for the operation of the computer program 5032 in the storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute a rapid analysis method for carotid artery plaques based on image intelligent analysis.

[0103] The network interface 505 is used for network communication, such as providing data information transmission, etc. Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device 500 to which the solution of the present invention is applied. The specific computer device 500 may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0104] The processor 502 is used to run the computer program 5032 stored in the memory to implement the corresponding functions of the above-mentioned method for rapid analysis of carotid artery plaque based on image intelligent analysis.

[0105] Those skilled in the art will understand that Figure 5 The embodiments of the computer device shown in the figure do not constitute a limitation on the specific composition of the computer device. In other embodiments, the computer device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently. For example, in some embodiments, the computer device may only include a memory and a processor. In such embodiments, the structure and function of the memory and the processor are the same as those of the embodiment of the present invention. Figure 5 The embodiments shown are consistent and will not be described again here.

[0106] It should be understood that in the embodiment of the present invention, the processor 502 may be a central processing unit (CPU), and the processor 502 may 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, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0107] In another embodiment of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium may be a volatile or non-volatile computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps included in the above-mentioned method for rapid analysis of carotid artery plaques based on image intelligent analysis are implemented.

[0108] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the above-described equipment, devices and units can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. Those of ordinary skill in the art can appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, computer software or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to the function. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0109] In the several embodiments provided by the present invention, it should be understood that the disclosed equipment, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. Units with the same function may also be combined into one unit. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices or units, or may be an electrical, mechanical or other form of connection.

[0110] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present invention.

[0111] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0112] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a computer-readable storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned computer-readable storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a magnetic disk or an optical disk.

[0113] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for rapid analysis of carotid artery plaque based on image intelligent analysis, characterized in that: The method comprises: Receiving input carotid artery plaque image information, and extracting sample image features corresponding to each sample from sample images of the carotid artery plaque image information according to a preset image feature extraction model; Acquire sample basic features corresponding to each sample from the sample database according to preset basic feature acquisition rules; Iteratively training the initial analysis model according to a preset model training strategy, the sample image features and the sample individual features to obtain a target analysis model; If the input individual information to be analyzed is received, extracting the corresponding initial individual image features from the carotid artery plaque image to be analyzed of the individual information to be analyzed according to the image feature extraction model; Acquire the corresponding initial individual basic features from the individual basic data of the individual information to be analyzed according to the basic feature acquisition rule; Analyze the initial individual image features and the initial individual basic features according to the target analysis model to obtain a carotid artery plaque analysis result including a risk probability; Before extracting sample image features corresponding to each sample from the sample image of the carotid artery plaque image information according to the preset image feature extraction model, the method further includes: Extracting initial image features corresponding to each sample from the sample images of the carotid artery plaque image information according to a pre-stored initial image feature extraction model; The initial image features are grouped according to the type labels of each sample to obtain a set of initial image features corresponding to each type label; Performing difference calculation on each of the initial image feature sets according to a preset difference calculation rule to obtain difference information corresponding to different types of labels on multiple feature items; Filtering the feature items included in the multiple indicator items to obtain the corresponding target feature items according to the preset indicator item screening rules and the difference information; According to the preset weight allocation strategy and the difference information, a corresponding weight coefficient is allocated to each of the target feature items; The contour feature extraction rules in the initial image feature extraction model are configured and updated according to the weight coefficients of the target feature items to obtain the image feature extraction model.

2. The method for rapid analysis of carotid artery plaque based on image intelligent analysis according to claim 1, characterized in that: The iterative training of the initial analysis model according to the preset model training strategy, the sample image features and the sample individual features to obtain the target analysis model includes: Combining the sample image features and sample individual features of each sample into corresponding sample comprehensive features; Inputting the comprehensive sample characteristics of each sample into the initial analysis model for analysis to obtain the predicted probability value corresponding to each sample; Calculate the loss value corresponding to the type label and predicted probability value of all samples according to the preset loss value calculation formula; Calculating a learning rate corresponding to the current number of iterations according to a preset learning rate calculation formula, so as to iteratively optimize the parameter values ​​in the initial analysis model based on the learning rate to obtain an optimized analysis model; Determine whether the loss value meets the preset training end condition; If the loss value satisfies the training end condition, the training process is terminated and the optimized analysis model is obtained as the corresponding target analysis model; If the loss value does not meet the training end condition, the optimized analysis model is used as the initial analysis model and the process returns to the step of inputting the comprehensive sample features of each sample into the initial analysis model for analysis.

3. The method for rapid analysis of carotid artery plaque based on image intelligent analysis according to claim 2, characterized in that: Before terminating the training process and obtaining the optimized analysis model as the corresponding target analysis model, the method further includes: Inputting each group of verification features in the pre-stored verification data set into the optimization analysis model respectively to obtain the prediction probability value corresponding to each verification feature; Calculate the corresponding accuracy rate according to the predicted probability value and the verification type label of each verification feature; Determining whether the accuracy rate is not less than a preset verification accuracy rate threshold; If the accuracy is not less than the verification accuracy threshold, the step of terminating the training process and obtaining the optimized analysis model as the corresponding target analysis model is executed.

4. The method for rapid analysis of carotid artery plaque based on image intelligent analysis according to any one of claims 1 to 3, characterized in that: The extracting of sample image features corresponding to each sample from the sample images of the carotid artery plaque image information according to a preset image feature extraction model includes: Extract the contour of each sample image according to the contour extraction rule in the image feature extraction model to obtain the image contour corresponding to each sample image; Identify the image contour of each of the sample images according to the contour feature extraction rule in the image feature extraction model to obtain corresponding contour type identification information; The contour type identification information of each sample image of the sample is integrated to obtain corresponding sample image features.

5. The method for rapid analysis of carotid artery plaque based on image intelligent analysis according to claim 1, characterized in that: The step of identifying the image contour of each sample image according to the contour feature extraction rule in the image feature extraction model to obtain corresponding contour type identification information includes: Determine an image contour that matches each indicator item according to the resolution interval corresponding to each indicator item in the contour feature extraction rule and the resolution of the sample image corresponding to the image contour; The image contours corresponding to the respective index items are identified according to the feature extraction strategies corresponding to the respective index items in the contour feature extraction rules, thereby obtaining contour type identification information corresponding to the respective image contours.

6. The method for rapid analysis of carotid artery plaque based on image intelligent analysis according to claim 1, characterized in that: The configuring and updating the contour feature extraction rules in the image feature extraction model according to the weight coefficients of the target feature items to obtain an updated image feature extraction model includes: updating the feature extraction strategy included in the contour feature extraction rule according to the target feature item to obtain an updated contour feature extraction rule; The extraction weight value of each feature extraction strategy in the updated contour feature extraction rule is configured according to the weight coefficient of the target feature item to obtain an updated image feature extraction model.

7. A rapid analysis device for carotid artery plaque based on image intelligent analysis, characterized in that: The device is used to perform the method for rapid analysis of carotid artery plaque based on image intelligent analysis according to any one of claims 1 to 6, and the device comprises: A sample image feature acquisition unit, configured to receive input carotid artery plaque image information, and extract sample image features corresponding to each sample from the sample image of the carotid artery plaque image information according to a preset image feature extraction model; A sample basic feature acquisition unit, used to acquire the sample basic features corresponding to each sample from the sample database according to a preset basic feature acquisition rule; A model training unit, used for iteratively training the initial analysis model according to a preset model training strategy, the sample image features and the sample individual features, so as to obtain a target analysis model; An initial individual image feature acquisition unit is used to extract corresponding initial individual image features from the carotid artery plaque image to be analyzed of the individual information to be analyzed according to the image feature extraction model upon receiving the input individual information to be analyzed; An initial individual basic feature acquisition unit, configured to acquire corresponding initial individual basic features from the individual basic data of the individual information to be analyzed according to the basic feature acquisition rule; The analysis result acquisition unit is used to analyze the initial individual image features and the initial individual basic features according to the target analysis model to obtain a carotid plaque analysis result including a risk probability.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the computer device executes the computer program, the method for rapid analysis of carotid plaque based on image intelligent analysis as described in any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the method for rapid analysis of carotid artery plaques based on image intelligent analysis as described in any one of claims 1 to 6.

Images