Aortic dissection acute and chronic stage classification method and device based on deep learning, medium and program product

Abstract

The invention provides an aortic dissection acute and chronic stage classification method and device based on deep learning, a medium and a program product, and relates to the field of intelligent medical treatment. The method comprises the following steps: acquiring two adjacent aorta cross-section interface ultrasonic images of a subject at the same position of at least two different time points; extracting aorta wall features and inner diaphragm features in the ultrasonic image; constructing a coordinate system comprising an x axis and a y axis by taking an intersection point of the aorta wall and the inner diaphragm features as an original point, and indexing areas where the inner diaphragm features in the two ultrasonic images are located by taking the x axis and the y axis as references to obtain two groups of coordinate point sets; the difference value of the inner diaphragm feature points at the same position on the x axis and / or the y axis is calculated; and outputting an auxiliary prediction result that the subject is acute or chronic according to the difference value. According to the method, by introducing the deep learning technology, automatic classification of the acute stage and the chronic stage of the aortic dissection is achieved, rapid and objective diagnosis and evaluation are expected to be achieved, and clinical decision making is assisted.

Description

Technical Field

[0001] The present invention relates to the field of intelligent medicine, and more particularly, to a method, device, medium and program product for classifying acute and chronic stages of aortic dissection based on deep learning. Background Art

[0002] Aortic dissection (also known as aortic dissecting aneurysm) is a tear in the middle intima of the aortic wall, and there is flowing or coagulated blood at the tear.

[0003] Aortic dissection (AD) is a serious cardiovascular emergency in which blood enters the aortic wall through a proximal intimal tear of the aorta, causing arterial wall separation, forming true and false lumens, and secondary intimal tears at the distal end, with a relatively high mortality rate. This disease is divided into two types: acute and chronic. Accurate staging diagnosis is crucial for formulating an effective treatment plan. Currently, the staging diagnosis mainly relies on the subjective judgment and rich experience of clinicians, with certain subjectivity and inevitable human errors. Summary of the Invention

[0004] The present invention aims to at least solve one of the technical problems existing in the prior art. To this end, the present invention provides a method, device, medium and program product for classifying acute and chronic stages of aortic dissection based on deep learning; the method of the present invention realizes the automated classification of acute and chronic stages of aortic dissection by introducing deep learning technology, in order to achieve faster and more objective diagnosis and evaluation, and assist clinical decision-making.

[0005] The first aspect of the present application discloses a method for classifying acute and chronic stages of aortic dissection based on deep learning, the method comprising:

[0006] S101, obtaining two adjacent aortic cross-sectional ultrasound images at the same position of a subject at at least two different time points;

[0007] S102, extracting the aortic wall features and intimal flap features in the ultrasound images;

[0008] S103, taking the intersection point of the aortic wall and intimal flap features as the origin, starting from the origin, the straight line extending along the tangent direction of the circle as the x-axis; starting from the origin, the straight line perpendicular to the x-axis and pointing to the inside or outside of the circle as the y-axis;

[0009] S104, indexing the regions where the intimal flap features are located in the two ultrasound images with the x-axis and y-axis as references to obtain two sets of coordinate point sets; calculating the differences of the intimal flap feature points at the same position on the x-axis and / or y-axis;

[0010] S105, outputting an auxiliary prediction result of whether the subject is acute or chronic according to the differences.

[0011] In some embodiments, when calculating the differences of the intimal flap feature points at the same position on the x-axis and / or y-axis in S104, the calculation method includes: there is at least 1 feature point. When there is 1 feature point, calculate the first x-axis difference of the x-axis coordinates of the coordinate points of the first feature point in two ultrasonic images, and calculate the first y-axis difference of the y-axis coordinates of the coordinate points of the first feature point in two ultrasonic images.

[0012] Optionally, the two different time points include adjacent first and second time points.

[0013] In some embodiments, in S105, outputting an auxiliary prediction result of whether the subject is acute or chronic based on the differences includes: if the first x-axis difference is greater than the first threshold but the first y-axis difference is less than the first threshold, output an auxiliary prediction result with a high probability that the subject is acute.

[0014] Optionally, if the first x-axis difference is less than the first threshold but the first y-axis difference is greater than the first threshold, output an auxiliary prediction result with a high probability that the subject is acute.

[0015] Optionally, if both the first x-axis difference and the first y-axis difference are greater than the first threshold, output an auxiliary prediction result with a high probability that the subject is acute.

[0016] Optionally, if both the first x-axis difference and the first y-axis difference are less than the first threshold, output an auxiliary prediction result with a high probability that the subject is chronic.

[0017] In some embodiments, when the at least two different time points are three time points, obtain the first ultrasonic image, the second ultrasonic image, and the third ultrasonic image at the same position at the three time points; the method further includes: there is at least 1 feature point. When there is 1 feature point, calculate the second x-axis difference of the x-axis coordinates of the first feature point in the first ultrasonic image and the second ultrasonic image, calculate the third x-axis difference of the x-axis coordinates of the first feature point in the second ultrasonic image and the third ultrasonic image, calculate the x-axis change difference between the second x-axis difference and the third x-axis difference. If the x-axis change difference is greater than the second threshold, output an auxiliary prediction result with a high probability that the subject is acute; if the x-axis change difference is less than the second threshold, output an auxiliary prediction result with a high probability that the subject is chronic.

[0018] Optionally, the method further includes: calculate the second y-axis difference of the y-axis coordinates of the first feature point in the first ultrasonic image and the second ultrasonic image; calculate the third y-axis difference of the y-axis coordinates of the first feature point in the second ultrasonic image and the third ultrasonic image; calculate the y-axis change difference between the second y-axis difference and the third y-axis difference. If the y-axis change difference is greater than the second threshold, output an auxiliary prediction result with a high probability that the subject is acute; if the y-axis change difference is less than the second threshold, output an auxiliary prediction result with a high probability that the subject is chronic.

[0019] In some embodiments, the method further includes:

[0020] Obtain one of the two adjacent cross-sectional ultrasound images of the aorta; extract the aortic wall features and intimal flap features from the cross-sectional ultrasound image;

[0021] Select a first connection point on the circumference where the aortic wall is located. The line segment between the first connection point and the origin fits the corresponding circumference. Connect the origin and the first connection point to form a first line segment; select a second connection point on the line where the intimal flap feature is located. The line segment between the second connection point and the origin fits the corresponding line of the intimal flap feature. Connect the origin and the second connection point to form a second line segment;

[0022] Calculate the angle between the first line segment and the second line segment, and output an auxiliary prediction result of acute or chronic for the subject according to whether the angle is obtuse or acute; if the angle is acute, output an auxiliary prediction result with a high probability of the subject being acute; if the angle is obtuse, output an auxiliary prediction result with a high probability of the subject being chronic.

[0023] In some embodiments, between S102 and S103, the method further includes: S1021, determine whether the ultrasound image meets the requirements of a standard ultrasound image. If it meets the requirements, execute S103; if it does not meet the requirements, interrupt the execution; the requirements of the standard ultrasound image include being able to identify the aortic wall features and intimal flap features of the aorta;

[0024] Optionally, if it does not meet the requirements, execute S1022, compare the ultrasound image of the subject with the standard ultrasound image, and obtain and output the position in the ultrasound image of the subject that does not meet the standard ultrasound image.

[0025] In some embodiments, the method further includes performing normalization processing on the ultrasound image to ensure standardized display of the blood flow area and intimal flap features in the image.

[0026] The second aspect of the present application discloses a computer device, which includes: a memory and a processor; the memory is used to store a computer program; the processor executes the computer program to implement the steps of the above method.

[0027] The third aspect of the present application discloses a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the above method.

[0028] The fourth aspect of the present application discloses a computer program product, including a computer program, and when the computer program is executed by a processor, it implements the steps of the above method.

[0029] The present application has the following beneficial effects:

[0030] 1. The present application innovatively discloses a method for classifying the acute and chronic phases of aortic dissection based on deep learning. According to the different characteristics of aortic dissection in the acute and chronic phases, a coordinate system is established with the intersection point between the intimal flap and the aortic wall as the origin. By quantifying the movement trajectory of the intimal flap at adjacent time points, or by quantifying the difference in the change of two movement trajectories composed of at least three adjacent time points, the effect of quantifying the movement trajectory of the intimal flap in the acute phase at different times is achieved, and further the result of whether there is an artifact in the intimal flap is output, so as to realize the objective diagnosis and evaluation of aortic dissection.

[0031] 2. The present application innovatively first draws lines that respectively fit the lines where the intimal flap and the aortic wall are located, and then calculates the angle between the two line segments. Based on whether the angle is obtuse or acute, the movement trajectory of the intimal flap is fused to judge the stage of aortic dissection. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those skilled in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0033] Figure 1 is a schematic flowchart of the method provided in the first aspect of the embodiment of the present invention;

[0034] Figure 2 is a schematic diagram of a classification system for the acute and chronic phases of aortic dissection based on deep learning provided by the embodiment of the present invention;

[0035] Figure 3 is a schematic diagram of a computer device provided by the embodiment of the present invention;

[0036] Figure 4 is a schematic diagram of the architecture of an exemplary computing device provided by the embodiment of the present invention;

[0037] Figure 5 is a schematic diagram of a storage medium provided by the embodiment of the present invention;

[0038] Figure 6 is a flowchart of a classification model for the acute and chronic phases of aortic dissection provided by the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] In order to enable those skilled in the art to better understand the solution of the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention.

[0040] In some of the processes described in the specification, claims, and the above-mentioned drawings of the present invention, a plurality of operations appear in a specific order. However, it should be clearly understood that these operations can be performed not in the order in which they appear herein or in parallel. The operation numbers, such as 101, 102, etc., are only used to distinguish different operations, and the numbers themselves do not represent any execution order. Additionally, these processes can include more or fewer operations, and these operations can be performed sequentially or in parallel. It should be noted that the descriptions such as "first" and "second" in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, nor do they limit that "first" and "second" are of different types.

[0041] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present invention.

[0042] Figure 1 It is a schematic flowchart of a method for classifying acute and chronic aortic dissection based on deep learning provided by an embodiment of the present invention. Specifically, the method includes the following steps:

[0043] S101: Obtain two adjacent cross-sectional ultrasound images of the aorta at the same position of the subject at at least two different time points; the cross-sectional ultrasound images are obtained through an aortic three-dimensional model.

[0044] In some embodiments, the terms "subject" or "test subject" or "test sample" used herein refer to any animal (e.g., mammal), including but not limited to humans, non-human primates, rodents, etc., that will be the recipient of a specific treatment. Generally, the terms "subject" and "patient" can be used interchangeably herein when referring to human subjects. Preferably, the subject is a human. In some embodiments, the test sample is a patient clinically used for prognostic evaluation.

[0045] S102: Extract the aortic wall features and intimal flap features from the ultrasound images;

[0046] S103: Take one of the intersection points of the aortic wall and the intimal flap features as the origin. Starting from the origin, the straight line extending along the tangent direction of the circumference is the x-axis; starting from the origin, the straight line perpendicular to the x-axis and pointing to the inside or outside of the circle is the y-axis; the intersection point of the intimal flap and the aortic wall is positioned as the origin. Although due to the thinness of the intimal flap, there may be position changes in the intimal flap, its intersection point with the aortic wall will not change.

[0047] S104: With the x-axis and y-axis as references, obtain two sets of coordinate point sets for the regions where the intimal flap features are located in two ultrasound images; calculate the differences in the x-axis and / or y-axis of the intimal flap feature points at the same position.

[0048] In some embodiments, when the images obtained in step S101 are images at two adjacent time points, taking one of the intersection points of the aortic wall and the intimal flap feature as the origin, obtain the first set of coordinate point sets by indexing the region where the intimal flap feature is located in the first image, and obtain the second set of coordinate point sets by indexing the region where the intimal flap feature is located in the second image; select two sets of coordinate points of the first feature point at the same position from the first set of coordinate point sets and the second set of coordinate point sets respectively, calculate the x-axis difference of the x-axis coordinates of the two sets of coordinate points of the first feature point at the same position, calculate the y-axis difference of the y-axis coordinates of the two sets of coordinate points of the first feature point at the same position, and output an auxiliary prediction result of whether the subject is acute or chronic according to the changes in the x-axis difference and the y-axis difference. Specifically,

[0049] When calculating the differences in the x-axis and / or y-axis of the intimal flap feature points at the same position in S104, the calculation method includes: the feature points include at least one. When there is one feature point, calculate the first x-axis difference of the x-axis coordinates of the coordinate points of the first feature point in the two ultrasound images, and calculate the first y-axis difference of the y-axis coordinates of the coordinate points of the first feature point in the two ultrasound images;

[0050] Optionally, the two different time points include adjacent first and second time points.

[0051] In some embodiments, in S105, outputting an auxiliary prediction result of whether the subject is acute or chronic according to the difference includes: if the first x-axis difference is greater than the first threshold but the first y-axis difference is less than the first threshold, output an auxiliary prediction result with a high probability that the subject is acute;

[0052] Optionally, if the first x-axis difference is less than the first threshold but the first y-axis difference is greater than the first threshold, output an auxiliary prediction result with a high probability that the subject is acute;

[0053] Optionally, if both the first x-axis difference and the first y-axis difference are greater than the first threshold, output an auxiliary prediction result with a high probability that the subject is acute;

[0054] Optionally, if both the first x-axis difference and the first y-axis difference are less than the first threshold, output an auxiliary prediction result with a high probability that the subject is chronic.

[0055] In some embodiments, when the images obtained in step S101 are images at three adjacent time points, and when the at least two different time points are three time points, the first ultrasound image, the second ultrasound image, and the third ultrasound image at the same position at the three time points are obtained; the method further includes: the feature points include at least one. When there is one feature point, calculate the second x-axis difference between the x-axis coordinates of the first feature point in the first ultrasound image and the second ultrasound image, calculate the third x-axis difference between the x-axis coordinates of the first feature point in the second ultrasound image and the third ultrasound image, calculate the x-axis change difference between the second x-axis difference and the third x-axis difference. If the x-axis change difference is greater than the second threshold, output an auxiliary prediction result indicating a high probability that the subject is acute; if the x-axis change difference is less than the second threshold, output an auxiliary prediction result indicating a high probability that the subject is chronic.

[0056] Optionally, the method further includes: calculating the second y-axis difference between the y-axis coordinates of the first feature point in the first ultrasound image and the second ultrasound image; calculating the third y-axis difference between the y-axis coordinates of the first feature point in the second ultrasound image and the third ultrasound image; calculating the y-axis change difference between the second y-axis difference and the third y-axis difference. If the y-axis change difference is greater than the second threshold, output an auxiliary prediction result indicating a high probability that the subject is acute; if the y-axis change difference is less than the second threshold, output an auxiliary prediction result indicating a high probability that the subject is chronic.

[0057] S105, output an auxiliary prediction result indicating that the subject is acute or chronic according to the difference.

[0058] In some embodiments, the outputting an auxiliary prediction result indicating that the subject is acute or chronic according to the difference in S105 includes: if the first x-axis difference is greater than the first threshold but the first y-axis difference is less than the first threshold, output an auxiliary prediction result indicating a high probability that the subject is acute; optionally, if the first x-axis difference is less than the first threshold but the first y-axis difference is greater than the first threshold, output an auxiliary prediction result indicating a high probability that the subject is acute; optionally, if both the first x-axis difference and the first y-axis difference are greater than the first threshold, output an auxiliary prediction result indicating a high probability that the subject is acute; optionally, if both the first x-axis difference and the first y-axis difference are less than the first threshold, output an auxiliary prediction result indicating a high probability that the subject is chronic.

[0059] In some embodiments, the auxiliary prediction result includes, but is not limited to, the form of a paper or electronic report. This result is only obtained by the intelligent machine based on the relevant data of the subject and is only used as a reference for medical staff and does not serve as the final diagnosis result of the subject.

[0060] In some embodiments, the method further includes:

[0061] Obtain one of the two adjacent cross-sectional ultrasound images of the aorta; extract the aortic wall features and intimal flap features in the cross-sectional ultrasound image;

[0062] Select a first connection point on the circumference where the aortic wall is located. The line segment between the first connection point and the origin fits the corresponding circumference, and connect the origin and the first connection point as the first line segment; select a second connection point on the line where the intimal flap feature is located. The line segment between the second connection point and the origin fits the corresponding line of the intimal flap feature, and connect the origin and the second connection point as the second line segment;

[0063] Calculate the angle between the first line segment and the second line segment, and output an auxiliary prediction result of acute or chronic for the subject according to whether the angle is obtuse or acute; if the angle is acute, output an auxiliary prediction result that the subject has a high probability of being acute; if the angle is obtuse, output an auxiliary prediction result that the subject has a high probability of being chronic.

[0064] In some embodiments, between S102 and S103, the method further includes: S1021, determine whether the ultrasound image meets the requirements of a standard ultrasound image. If it meets the requirements, execute S103; if it does not meet the requirements, interrupt the execution; the requirements of the standard ultrasound image include being able to identify the aortic wall features and intimal flap features of the aorta;

[0065] Optionally, if it does not meet the requirements, execute S1022, compare the ultrasound image of the subject with the standard ultrasound image, and obtain and output the position in the ultrasound image of the subject that does not meet the standard ultrasound image.

[0066] In some embodiments, the method further includes performing normalization processing on the ultrasound image to ensure the standardized display of the blood flow area and intimal flap features in the image.

[0067] In some embodiments, the first threshold and the second threshold are obtained through training with training set samples, and include any one of the following: a numerical value, a numerical range, and the specific form is not specifically limited in this embodiment.

[0068] The method when using training set samples for training includes:

[0069] Obtain training set samples including ultrasound images of the aorta and classification labels, where the classification labels include the acute phase and the chronic phase; the key features of the acute phase and the chronic phase have been manually marked in the ultrasound images; specifically, the key features are marked by introducing the method of expert consensus to mark the ultrasound images in the training set samples.

[0070] Input the ultrasound image into a machine learning model to obtain a predicted classification result, compare it with the classification label, and optimize the model according to the comparison result to obtain a constructed prediction model.

[0071] The specific method is asFigure 6 As shown in Figure 6 , there are four modules for the ultrasound images of the training set samples: data preprocessing, model construction, parameter optimization and training, and model evaluation.

[0072] 1. Data preprocessing: Data preprocessing is crucial for ensuring the consistency of model input and the effectiveness of model training. In this study, the following steps were taken for data processing:

[0073] Image adjustment and normalization: All images were uniformly adjusted to a size of 512x512 pixels, and the pixel values were normalized to ensure the standardized display of features, making the white blood flow area and the black intimal flap features more prominent.

[0074] Data augmentation: To improve the generalization ability of the model and reduce the risk of overfitting, we performed data augmentation through rotation and scaling operations. Using the ImageDataGenerator function in the Keras library, the rotation angle was set to ±15°, and the scaling range was set to 0.8 to 1.2 to increase data diversity.

[0075] Data annotation: The acute and chronic stage labels of the images were manually annotated to ensure the accuracy of the annotation for each image. These annotations will serve as the classification targets of the model for the supervised learning of the model.

[0076] 2. Model construction: In terms of the model structure, we adopted ResNet50 as the basic architecture and further trained it through transfer learning to obtain better performance in medical image classification tasks. ResNet50 is a deep residual network constructed by 50 convolutional structures. The key lies in the introduction of the residual module, which alleviates the problem of gradient disappearance through cross-layer connections. The overall process of the model is as Figure 6 shown. The specific layer structure is as follows:

[0077] Input layer: The input is a grayscale image with a size of 512x512x1. First, a zero-padding operation is performed to ensure the integrity of the image boundary and prevent the reduction of the image size caused by subsequent convolution operations.

[0078] Initial convolutional layer: A 7x7 large convolutional kernel is used for convolution (convolutional layer), combined with batch normalization and ReLU activation function to ensure the stability of feature extraction. Then, a max-pooling layer is added to reduce the size of the feature map and provide a suitable input for the subsequent residual modules.

[0079] Residual Module: oConv Block: It contains multiple 3x3 convolutional layers, and a shortcut connection is added within each convolutional block to directly pass the input to the output, which is used to learn the residual between the input and the output, thereby alleviating the vanishing gradient problem.

[0080] oIdentity Block: Similar to the Conv Block, but it does not change the size of the input. The shortcut connection directly makes a cross-layer connection, further improving the deep feature extraction ability of the model.

[0081] Convolutional Kernel and Pooling Layer: Use small 3x3 convolutional kernels, and gradually increase the number of convolutional kernels as the deep structure progresses to capture more complex features. A max pooling layer is used after each residual module to further reduce the size of the feature map while retaining key features.

[0082] Global Average Pooling Layer: After all convolutional and residual modules, a global average pooling layer is used to reduce the feature map to a size of 1x1, integrating all features into a feature vector in a small area, providing input for the fully connected layer.

[0083] Flatten Layer: Flatten the features after global pooling into a one-dimensional vector, which is convenient for input into the fully connected layer.

[0084] Fully Connected (FC) Layer: It contains two fully connected layers, which are used to further integrate features and improve the classification effect.

[0085] Output Layer: Use the Softmax activation function to convert the output into classification probabilities for the acute and chronic phases.

[0086] 3. Parameter Optimization and Training: In the model training stage, we adopt the following optimization strategies to improve the training effect and classification accuracy of the model:

[0087] Learning Rate Adjustment: The initial learning rate is set to 0.001, and a learning rate decay strategy is adopted to gradually reduce the learning rate to enhance the convergence of the model.

[0088] Batch Size: Select 32 as the batch size and observe its impact on the model performance through experiments.

[0089] Regularization and Dropout: Add L2 regularization and a Dropout layer (0.3) to the fully connected layer to prevent overfitting and enhance the generalization ability of the model.

[0090] Optimizer Selection: Adopt the Adam optimizer, which has a fast convergence speed, is suitable for complex network structures, and can effectively accelerate the training process.

[0091] 4. Model Evaluation: In the model evaluation stage, multiple metrics are used to comprehensively evaluate the classification performance of the model to ensure its reliability in clinical applications:

[0092] Accuracy: Measures the overall correct rate of the model in classifying acute and chronic phases.

[0093] Sensitivity and Specificity: Evaluate the model's recognition ability for acute and chronic phases respectively to detect the classification effect of the model on different categories.

[0094] ROC Curve: Evaluate the performance of the model at different thresholds through the ROC curve and the area under the curve (AUC) to further understand the discrimination ability of the model on various types of samples.

[0095] Expected Results and Application Value: Through the deep learning model based on ResNet50 constructed in this study, it is expected to achieve accurate classification of acute and chronic aortic dissections. On the basis of optimizing data processing and model parameters, we hope to achieve a high classification accuracy, with both sensitivity and specificity being better than traditional clinical observation methods. The successful application of this model will provide an automated auxiliary diagnostic tool for clinicians, improving the diagnostic speed and objectivity, reducing human errors, and thus supporting more efficient treatment decisions.

[0096] Figure 3 It is a schematic diagram of a computer device provided by an embodiment of the present invention. As Figure 3 shown, the device 2000 may include: one or more processors 2010, and one or more memories 2020; wherein, computer-readable code is stored in the memory, and when the computer-readable code is run by the one or more processors, the method described above can be executed.

[0097] The processor in this embodiment may be an integrated circuit chip with signal processing capabilities. The above-mentioned processor may be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute the various methods, operations and logic block diagrams disclosed in the embodiments of the present disclosure. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc., and may be of the X86 architecture or the ARM architecture.

[0098] In general, the various example embodiments of the present disclosure may be implemented in hardware or in special-purpose circuits, software, firmware, logic, or any combination thereof. Some aspects may be implemented in hardware, while other aspects may be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device. When aspects of the embodiments of the present disclosure are illustrated or described as block diagrams, flowcharts, or using some other graphical representation, it will be understood that the blocks, devices, systems, techniques, or methods described herein may be implemented as non-limiting examples in hardware, software, firmware, special-purpose circuits or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0099] For example, the method or apparatus according to an embodiment of the present disclosure may also be implemented by means of Figure 4 the architecture of the computing device 3000 shown. As Figure 4 shown, the computing device 3000 may include a bus 3010, one or more CPUs 3020, a read-only memory (ROM) 3030, a random access memory (RAM) 3040, a communication port 3050 connected to a network, an input / output component 3060, a hard disk 3070, etc. The storage device in the computing device 3000, such as the ROM 3030 or the hard disk 3070, may store various data or files used for the processing and / or communication of the method provided by the present disclosure and the program instructions executed by the CPU. The computing device 3000 may also include a user interface 3080. Of course, Figure 4 the architecture shown is only exemplary, and when implementing different devices, one or more components shown in the Figure 4 computing device may be omitted according to actual needs.

[0100] An embodiment of the present invention also provides a computer-readable storage medium, such as Figure 5As shown, it is a schematic diagram of a storage medium 4000 provided by an embodiment of the present invention. Computer-readable instructions 4010 are stored on the computer storage medium 4020. When the computer-readable instructions 4010 are run by a processor, the methods according to the embodiments of the present disclosure described with reference to the above figures can be executed. The computer-readable storage medium in the embodiments of the present disclosure may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. The non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus random access memory (DR RAM). It should be noted that the memories of the methods described herein are intended to include but not be limited to these and any other suitable types of memories. It should be noted that the memories of the methods described herein are intended to include but not be limited to these and any other suitable types of memories.

[0101] The embodiments of the present disclosure also provide a computer program product or system, including a computer program, which implements the steps of the above method when executed by a processor.

[0102] In some embodiments, this embodiment also discloses a deep learning-based acute and chronic aortic dissection classification system, as Figure 2 shown, the system includes:

[0103] An image acquisition module 201, configured to or for acquiring two adjacent aortic cross-sectional ultrasound images of the subject at the same position at at least two different time points;

[0104] A feature extraction module 202, configured to or for extracting aortic wall features and intimal flap features from the ultrasound images;

[0105] A coordinate system construction module 203, configured to or for taking the intersection point of the aortic wall and intimal flap features as the origin, starting from the origin, the straight line extending along the tangent direction of the circumference as the x-axis; starting from the origin, the straight line perpendicular to the x-axis and pointing to the inside or outside of the circle as the y-axis;

[0106] The difference calculation module 204 is used or configured to, with the x-axis and y-axis as references, obtain two sets of coordinate point sets for the regions where the intimal flap features are located in two ultrasonic images; calculate the differences of the intimal flap feature points at the same position on the x-axis and / or y-axis.

[0107] The result output module 205 is used or configured to output an auxiliary prediction result of whether the subject is acute or chronic according to the difference.

[0108] In some embodiments, the system further includes:

[0109] The line segment connection module is used or configured to select a first connection point on the circumference where the aortic wall is located. The line segment between the first connection point and the origin fits the corresponding circumference, and the line segment connecting the origin and the first connection point is the first line segment; select a second connection point on the line where the intimal flap feature is located. The line segment between the second connection point and the origin fits the corresponding line of the intimal flap feature, and the line segment connecting the origin and the second connection point is the second line segment.

[0110] The included angle judgment module is used or configured to calculate the angle of the included angle between the first line segment and the second line segment, and output an auxiliary prediction result of whether the subject is acute or chronic according to whether the included angle is obtuse or acute; if the included angle is acute, output an auxiliary prediction result with a high probability that the subject is acute; if the included angle is obtuse, output an auxiliary prediction result with a high probability that the subject is chronic.

[0111] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code, and the module, program segment, or part of code includes one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0112] In general, the various example embodiments of the present disclosure can be implemented in hardware or dedicated circuits, software, firmware, logic, or any combination thereof. Certain aspects can be implemented in hardware, while other aspects can be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device. When aspects of the embodiments of the present disclosure are illustrated or described as block diagrams, flowcharts, or using some other graphical representation, it will be understood that the blocks, devices, systems, techniques, or methods described herein can be implemented as non-limiting examples in hardware, software, firmware, dedicated circuits or logic, general hardware or a controller or other computing device, or some combination thereof.

[0113] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0114] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0115] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0116] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

[0117] The example embodiments of the present disclosure described in detail above are merely illustrative and not restrictive. Those skilled in the art should understand that various modifications and combinations can be made to these embodiments or their features without departing from the principles and spirit of the present disclosure, and such modifications should fall within the scope of the present disclosure.

Claims

1. A method for classifying the acute and chronic phases of aortic dissection based on deep learning, characterized in that, The method includes: S101. Obtain two cross-sectional ultrasound images of the aorta at the same position of the subject at at least two different time points; S102. Extract the aorta wall features and intimal flap features in the ultrasound images; S103. Take the intersection point of the aorta wall and intimal flap features as the origin. Starting from the origin, the straight line extending along the tangent direction of the circumference is the x-axis; starting from the origin, the straight line perpendicular to the x-axis and pointing to the inside or outside of the circle is the y-axis; S104. With the x-axis and y-axis as references, index the regions where the intimal flap features are located in the two ultrasound images to obtain two sets of coordinate point sets; calculate the differences of the intimal flap feature points at the same position on the x-axis and / or y-axis; S105. Output an auxiliary prediction result of whether the subject is acute or chronic according to the differences.

2. The method for classifying acute and chronic stages of aortic dissection based on deep learning according to claim 1, wherein When calculating the differences of the intimal flap feature points at the same position on the x-axis and / or y-axis in S104, the calculation method includes: the feature points include at least one. When there is one feature point, calculate the first x-axis difference of the x-axis coordinates of the coordinate points of the first feature point in the two ultrasound images, and calculate the first y-axis difference of the y-axis coordinates of the coordinate points of the first feature point in the two ultrasound images; Optionally, the at least two different time points include an adjacent first time point and a second time point.

3. The method for classifying the acute and chronic phases of aortic dissection based on deep learning according to claim 2, characterized in that, The output of the auxiliary prediction result of whether the subject is acute or chronic in S105 includes: if the first x-axis difference is greater than the first threshold but the first y-axis difference is less than the first threshold, output an auxiliary prediction result with a high probability that the subject is acute; Optionally, if the first x-axis difference is less than the first threshold but the first y-axis difference is greater than the first threshold, output an auxiliary prediction result with a high probability that the subject is acute; Optionally, if both the first x-axis difference and the first y-axis difference are greater than the first threshold, output an auxiliary prediction result with a high probability that the subject is acute; Optionally, if both the first x-axis difference and the first y-axis difference are less than the first threshold, output an auxiliary prediction result with a high probability that the subject is chronic.

4. The method for classifying acute and chronic stages of aortic dissection based on deep learning according to claim 1, wherein When the at least two different time points are three time points, obtain the first ultrasound image, the second ultrasound image, and the third ultrasound image at the same position at the three time points; the method further includes: the feature points include at least one. When there is one feature point, calculate the second x-axis difference of the x-axis coordinates of the first feature point in the first ultrasound image and the second ultrasound image, calculate the third x-axis difference of the x-axis coordinates of the first feature point in the second ultrasound image and the third ultrasound image, calculate the x-axis change difference between the second x-axis difference and the third x-axis difference. If the x-axis change difference is greater than the second threshold, output an auxiliary prediction result with a high probability that the subject is acute; if the x-axis change difference is less than the second threshold, output an auxiliary prediction result with a high probability that the subject is chronic; Optionally, the method further includes: calculating a second y-axis difference between the y-axis coordinates of the first feature point in the first ultrasonic image and the second ultrasonic image; calculating a third y-axis difference between the y-axis coordinates of the first feature point in the second ultrasonic image and the third ultrasonic image; calculating a y-axis change difference between the second y-axis difference and the third y-axis difference, and if the y-axis change difference is greater than a second threshold, outputting an auxiliary prediction result indicating a high probability that the subject is acute; if the y-axis change difference is less than the second threshold, outputting an auxiliary prediction result indicating a high probability that the subject is chronic.

5. The method for classifying the acute and chronic phases of aortic dissection based on deep learning according to claim 1, wherein The method further includes: obtaining one of the two adjacent cross-sectional ultrasonic images of the aorta; extracting the aortic wall feature and the intimal flap feature in the cross-sectional ultrasonic image; selecting a first connection point on the circumference where the aortic wall is located, the line segment between the first connection point and the origin fitting the corresponding circumference, and connecting the origin and the first connection point as the first line segment; selecting a second connection point on the line where the intimal flap feature is located, the line segment between the second connection point and the origin fitting the corresponding line of the intimal flap feature, and connecting the origin and the second connection point as the second line segment; calculating the angle between the first line segment and the second line segment, and outputting an auxiliary prediction result indicating that the subject is acute or chronic according to whether the angle is obtuse or acute; if the angle is acute, outputting an auxiliary prediction result indicating a high probability that the subject is acute; if the angle is obtuse, outputting an auxiliary prediction result indicating a high probability that the subject is chronic.

6. The method for classifying the acute and chronic phases of aortic dissection based on deep learning according to any one of claims 1-5, characterized in that, Between S102 and S103, the method further includes: S1021, determining whether the ultrasonic image meets the requirements of a standard ultrasonic image, and if it meets, executing S103; if it does not meet, interrupting the execution; the requirements of the standard ultrasonic image include being able to identify the aortic wall feature and the intimal flap feature of the aorta; Optionally, if it does not meet, execute S1022, comparing the ultrasonic image of the subject with the standard ultrasonic image, and obtaining and outputting the position in the ultrasonic image of the subject that does not meet the standard ultrasonic image.

7. The method for classifying acute and chronic stages of aortic dissection based on deep learning according to claim 1, wherein The method further includes performing a normalization process on the ultrasonic image to ensure the standardized display of the blood flow area and the intimal flap feature in the image.

8. A computer device, characterized in that, The device includes: a memory and a processor; the memory is used for storing a computer program; the processor executes the computer program to implement the steps of the method according to any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, A computer program is stored thereon, and when the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-7.

Images