Brain perfusion parameter acquisition method and device based on radiography, equipment and medium

Abstract

The invention provides a brain perfusion parameter acquisition method and device based on angiography, equipment and a medium. The method comprises the following steps: acquiring a cerebral angiography image, segmenting a target blood vessel image from the cerebral angiography image, and extracting a plurality of radiomics data features and cerebral vessel anatomical data features from the target blood vessel image; screening the plurality of image omics data features according to the correlation coefficients of the plurality of image omics data features to obtain a target image omics data feature; inputting the target radiomics data features as a pre-trained classification model, outputting a prediction result of brain perfusion parameters, and constructing a final parameter set according to the anatomical data features and the prediction result; by adopting the method, the brain perfusion parameter acquisition efficiency can be improved, and the acquisition cost is reduced.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to a method, device, equipment and medium for acquiring cerebral perfusion parameters based on angiography. Background Art

[0002] Existing related technologies usually use brain perfusion imaging to collect patients' brain perfusion parameters, including CT brain perfusion imaging and MR brain perfusion imaging. This requires patients to undergo a brain perfusion test and then undergo imaging. The general understanding is that although cerebral angiography can accurately evaluate the stenosis and occlusion of blood vessels, it cannot accurately reflect the blood perfusion of ischemic brain tissue. Therefore, brain perfusion imaging is often used for more accurate brain perfusion parameter evaluation. In addition, it takes a long time and is costly to obtain brain perfusion parameters through brain perfusion imaging. This method can only obtain brain perfusion parameters, but not additional parameters. If other data are required, additional examinations are required, which consumes more time and economic costs. Summary of the invention

[0003] The purpose of this application is to address the deficiencies of the above-mentioned existing related technologies and to propose a method, device, equipment and medium for obtaining cerebral perfusion parameters based on angiography, so as to improve the efficiency of cerebral perfusion parameter acquisition and reduce the acquisition cost.

[0004] In a first aspect, the present application provides a method for acquiring cerebral perfusion parameters based on angiography, comprising:

[0005] Acquiring a cerebral angiography image, segmenting a target blood vessel image from the cerebral angiography image, and extracting a plurality of imaging omics data features and cerebral blood vessel anatomical data features from the target blood vessel image;

[0006] Screening the multiple radiomics data features according to correlation coefficients of the multiple radiomics data features to obtain target radiomics data features;

[0007] The target imaging omics data features are used as input to a pre-trained classification model, and the output of the classification model is used as a prediction result of cerebral perfusion parameters. The anatomical data features and the prediction results are used to construct a final parameter set.

[0008] The present application obtains anatomical data features based on cerebral angiography, so that the anatomical data features can be obtained through a single angiography, and the cerebral angiography images can be further analyzed by using a pre-trained classification model after screening to obtain accurate cerebral perfusion parameters; and obtaining cerebral perfusion parameters based on cerebral angiography can overcome technical bias, achieve accurate assessment of vascular stenosis and occlusion, and at the same time improve the efficiency of parameter acquisition and reduce the cost of parameter acquisition.

[0009] Further, obtaining the classification model includes:

[0010] The cerebral angiography of normal persons and the cerebral angiography of patients are used as training data, and the cerebral perfusion detection data corresponding to the training data are respectively obtained, and the cerebral perfusion detection data are used as labels of the training data. According to the labels and the training data, an initial classification model based on machine learning is trained to obtain the classification model.

[0011] Further, the initial classification model based on machine learning is trained according to the label and the training data to obtain the classification model, including:

[0012] The training data is used as the input of the initial classification model, and the output of the initial classification model is used as the training result of cerebral perfusion parameter prediction, and the classification loss function value is obtained according to the label and the training result, and the parameters of the initial classification model are updated according to the classification loss function value to obtain the classification model.

[0013] Furthermore, the cerebral perfusion parameters include: the average transit time required for the contrast agent to pass from the intracranial arterial side to the venous side or the contrast agent peak time.

[0014] Further, the cerebral angiography image includes: a frontal image or a lateral image; the acquisition of the cerebral angiography image, segmentation of a target blood vessel image from the cerebral angiography image, and extraction of multiple imaging omics data features from the target blood vessel image include:

[0015] The frontal image or the lateral image is collected, a target blood vessel image in the frontal image or the lateral image is obtained through a three-dimensional slicing system, and multiple radiomics data features in the target blood vessel image are extracted through the three-dimensional slicing system.

[0016] Further, the correlation coefficient includes: Spearman correlation coefficient; the multiple imaging omics data features are screened according to the correlation coefficients of the multiple imaging omics data features to obtain target imaging omics data features, including:

[0017] According to the Spearman correlation coefficient or Lasso regression between the multiple radiomics data features, target radiomics data features with significant differences are screened out.

[0018] Furthermore, the target radiomics data features with significant differences are screened out according to the Spearman correlation coefficient or Lasso regression between the multiple radiomics data features, including:

[0019] Lasso regression is performed on the multiple imaging omics data features to screen out multiple candidate imaging omics data features whose correlation with cerebral perfusion parameters is greater than a correlation threshold, and the Spearman correlation coefficients of the multiple candidate imaging omics data features are obtained, and the target imaging omics data features whose Spearman correlation coefficients are greater than the coefficient threshold are selected.

[0020] This application uses Lasso regression to perform primary screening of multiple imaging genomics data features, and uses Spearman correlation coefficient for secondary screening, which can obtain target imaging genomics data features that are more correlated with cerebral perfusion parameters, thereby accurately reflecting the blood perfusion of ischemic brain tissue and obtaining more accurate cerebral perfusion parameters.

[0021] In a second aspect, the present application provides a device for acquiring cerebral perfusion parameters based on angiography, comprising: a data acquisition module, a feature screening module and a parameter prediction module; wherein:

[0022] The data acquisition module is used to acquire a cerebral angiography image, segment a target blood vessel image from the cerebral angiography image, and extract a plurality of imaging omics data features and cerebral blood vessel anatomical data features from the target blood vessel image;

[0023] The feature screening module is used to screen the multiple radiomics data features according to the correlation coefficients of the multiple radiomics data features to obtain target radiomics data features;

[0024] The parameter prediction module is used to use the target imaging genomics data features as input to a pre-trained classification model, and use the output of the classification model as the prediction result of the cerebral perfusion parameter, and construct a final parameter set based on the anatomical data features and the prediction result.

[0025] In a third aspect, the present application provides an electronic device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for acquiring cerebral perfusion parameters based on angiography as described in the first aspect is implemented.

[0026] In a fourth aspect, the present application provides a computer-readable storage medium storing a plurality of instructions suitable for loading by a processor to execute the angiography-based cerebral perfusion parameter acquisition method as described in the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a flowchart of a method for obtaining cerebral perfusion parameters based on angiography provided in the present application;

[0028] Figure 2It is a flowchart of another method for obtaining cerebral perfusion parameters based on angiography provided in the present application;

[0029] Figure 3 It is a structural schematic diagram of a device for acquiring cerebral perfusion parameters based on angiography provided in the present application;

[0030] Figure 4 It is a structural schematic diagram of an electronic device provided by this application. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0032] It is worth noting that cerebral perfusion imaging is commonly used in clinical practice for the examination of cerebrovascular diseases. It is a non-invasive examination method. Cerebral perfusion imaging includes: MR cerebral perfusion imaging and CT cerebral perfusion imaging. CT cerebral perfusion imaging is commonly used in clinical practice, and its results are more accurate than MR cerebral perfusion imaging. This examination method is to examine the blood supply of local brain tissue and the compensation of collateral circulation through intravenous injection of contrast agent. Cerebral perfusion imaging requires brain perfusion first, and then imaging detection. The total time of these two steps is long, and patients need to pay more costs to obtain them; in addition, this method can only obtain cerebral perfusion parameters, and cannot obtain additional parameters. If other data are required, additional examinations are required, which consumes longer time and economic costs.

[0033] Cerebral angiography is a minimally invasive interventional procedure. Under the guidance of a guidewire, a catheter is sent to the target cerebral blood vessel, and a contrast agent is injected into the artery to be examined. The contrast agent is developed under X-rays to check whether the cerebral artery has stenosis, malformation, aneurysm, arteriovenous fistula or smoke-like blood vessels, etc., to guide the next step of treatment. It is the gold standard for evaluating cerebral blood vessels. Existing related technologies generally believe that although cerebral angiography can accurately evaluate the stenosis and occlusion of blood vessels, it cannot accurately reflect the blood perfusion of ischemic brain tissue. Therefore, cerebral perfusion imaging is often used to more accurately evaluate cerebral perfusion parameters.

[0034] The present application adopts a method of obtaining anatomical data features based on cerebral angiography, so that the anatomical data features can be obtained through a single angiography, and the cerebral angiography images can be further analyzed through machine learning to obtain cerebral perfusion parameters; and obtaining cerebral perfusion parameters based on cerebral angiography can overcome technical bias and achieve accurate assessment of vascular stenosis and occlusion, while improving the efficiency of parameter acquisition and reducing the cost of parameter acquisition.

[0035] In order to more clearly illustrate the technical solution of the present application, it will be described in detail from the following embodiments.

[0036] Example 1

[0037] See also Figure 1 , is a flowchart of a method for obtaining cerebral perfusion parameters based on angiography provided in the present application, comprising: steps S11 to S13, specifically:

[0038] Step S11 : acquiring a cerebral angiography image, segmenting a target blood vessel image from the cerebral angiography image, and extracting a plurality of imaging omics data features and cerebral blood vessel anatomical data features from the target blood vessel image.

[0039] In some embodiments, the cerebral angiography image includes: an anterior view or a lateral view of the cerebral angiography.

[0040] In some embodiments, a lateral view of cerebral angiography is preferentially selected to segment the target blood vessel image.

[0041] In some embodiments, a target blood vessel image is segmented from a cerebral angiography image, and multiple imaging omics data features are extracted from the target blood vessel image, including: acquiring an anteroposterior or lateral view, obtaining the target blood vessel image in the anteroposterior or lateral view through a three-dimensional slicing system, and extracting multiple imaging omics data features in the target blood vessel image through the three-dimensional slicing system.

[0042] In some embodiments, the three-dimensional slicing system selects a target image area in the target blood vessel image through graphic processing, and digitizes pixels in the target image area and the relationship between pixels to obtain multiple imaging omics data features.

[0043] In some embodiments, the three-dimensional slicing system may be an image processing system based on 3D slicer software version 4.10 or above.

[0044] It is worth mentioning that the data was extracted from one of the modules of the 3D slicer software, which is open source and built based on the original theory of imaging omics.

[0045] In some embodiments, there may be dozens of imaging omics data features, which may include the following features: Total energy, Long run high gray level emphasis, Zone variance, Imc (Informational Measure of Correlation), Gray level non uniformity, Gray level non uniformity, Low gray level run emphasis, Small area emphasis, Large area emphasis, Low gray level zone emphasis, High gray level zone emphasis.

[0046] Step S12: screening multiple radiomics data features according to correlation coefficients of the multiple radiomics data features to obtain target radiomics data features.

[0047] In some embodiments, the correlation coefficient includes: a Spearman correlation coefficient; based on the correlation coefficients of multiple imaging omics data features, multiple imaging omics data features are screened to obtain target imaging omics data features, including: based on the Spearman correlation coefficient or Lasso regression between multiple imaging omics data features, target imaging omics data features with significant differences are screened.

[0048] In some embodiments, target imaging omics data features with significant differences are screened out based on the Spearman correlation coefficient or LASSO regression between multiple imaging omics data features, including: performing LASSO regression on multiple imaging omics data features to screen out multiple candidate imaging omics data features whose correlation with cerebral perfusion parameters is greater than a correlation threshold, and obtaining the Spearman correlation coefficient of the multiple candidate imaging omics data features, and selecting the target imaging omics data features whose Spearman correlation coefficient is greater than the coefficient threshold.

[0049] This application uses Lasso regression to perform primary screening of multiple imaging genomics data features, and uses Spearman correlation coefficient for secondary screening, which can obtain target imaging genomics data features that are more correlated with cerebral perfusion parameters, thereby accurately reflecting the blood perfusion of ischemic brain tissue and obtaining more accurate cerebral perfusion parameters.

[0050] It is worth mentioning that the full name of Lasso regression is Least Absolute Shrinkage and Selection Operator, and its corresponding full name in Chinese is the least absolute value convergence and selection operator algorithm. Lasso regression overcomes the problem that there may be multicollinearity between features, which leads to unstable or even uncalculated model parameter estimation, by adding a penalty term to adjust the complexity in the regression model, and solves the problem that not all features have a significant impact on the prediction results, so some irrelevant or redundant features may interfere with the prediction performance of the model. Specifically, this penalty term is λ times the sum of the absolute values ​​of the model coefficients, which is the so-called L1 norm, where λ is a non-negative adjustment parameter, so as to maintain the fitting ability on the training set while improving the generalization ability on the test set. However, Lasso regression also has the problem that the weights are easily reduced to 0, making it not arbitrarily differentiable, resulting in the inability to perform reverse differentiation to update the parameters.

[0051] In some existing related technologies, the coordinate descent method is used to improve the problem that Lasso regression is not differentiable at any point. The basic idea of ​​the coordinate descent method is that in each iteration, all variables except one variable are fixed, and the value of the variable is updated by solving sub-problems only about the variable. Then each variable is updated in turn until the stopping criterion is met or the maximum number of iterations is reached. Specifically, at the kth iteration, each updated weight is minimized along a coordinate component direction, specifically, the coordinate component with the minimum sum of the loss function and the penalty term is reselected from multiple coordinate components in the k-1th time as the kth optimal weight update result, but this method always selects the coordinate component with the smallest loss for weight update, resulting in the inability to explore other coordinate components, especially when the loss function is non-convex, it is easier to fall into the local optimum. Based on this, the present application adjusts the iterative update of the weight of Lasso regression.

[0052] In some embodiments, the iterative update of the weights of the Lasso regression is adjusted, including: based on a greedy algorithm, randomly generating a coordinate component selection probability each time; if the coordinate component selection probability is greater than a preset threshold, then based on a random walk model, updating the weight update result of the previous moment in any direction to obtain the weight update result of this moment; if the coordinate component selection probability is not greater than the preset threshold, then selecting the coordinate component with the minimum sum of the loss function and the penalty term from the coordinate components of the previous moment as the weight update result of this moment.

[0053] This application uses a greedy algorithm and a random walk model to explore weight updates in multiple candidate coordinate components, so that when the loss function is non-convex, it can jump out of the local optimum, thereby accurately reflecting the blood perfusion of ischemic brain tissue and obtaining more accurate brain perfusion parameters.

[0054] Step S13: using the target imaging omics data features as input to a pre-trained classification model, and using the output of the classification model as the prediction result of the cerebral perfusion parameters, and constructing a final parameter set based on the anatomical data features and the prediction result.

[0055] In some embodiments, obtaining a classification model includes: using cerebral angiography of normal persons and cerebral angiography of patients as training data, and respectively obtaining cerebral perfusion detection data corresponding to the training data, using the cerebral perfusion detection data as a label of the training data, and training an initial classification model based on machine learning according to the label and the training data to obtain a classification model.

[0056] In some embodiments, the training of the initial classification model based on machine learning is supervised training or semi-supervised training.

[0057] It is worth noting that normal people are experimental samples that are not sick or are sick but do not need to test cerebral perfusion parameters, and patients are experimental samples that are sick and need to test cerebral perfusion parameters.

[0058] In some embodiments, the classification model can be a machine learning algorithm such as a random forest or a support vector machine.

[0059] In some embodiments, an initial classification model based on machine learning is trained according to labels and training data to obtain a classification model, including: taking the training data as the input of the initial classification model, taking the output of the initial classification model as the training result of cerebral perfusion parameter prediction, and obtaining the classification loss function value according to the labels and training results, and updating the parameters of the initial classification model according to the classification loss function value to obtain the classification model.

[0060] In some embodiments, the cerebral perfusion parameters include: the average transit time required for contrast agent to travel from the intracranial arterial side to the intracranial venous side or the contrast agent peak time.

[0061] In some embodiments, see Figure 2 , is a flowchart of another method for obtaining cerebral perfusion parameters based on contrast provided in this application. When training or testing the imaging genomics data features, data analysis such as Spearman correlation coefficient and Lasso regression can be combined, and then machine learning methods such as random forest or support vector machine can be used to obtain the average transit time or contrast agent peak time required for the contrast agent to pass from the intracranial arterial side to the venous side in the cerebral perfusion results.

[0062] The present application adopts a method of obtaining anatomical data features based on cerebral angiography, so that the anatomical data features can be obtained through a single angiography, and the cerebral angiography images can be further analyzed through machine learning to obtain cerebral perfusion parameters; and obtaining cerebral perfusion parameters based on cerebral angiography can overcome technical bias and achieve accurate assessment of vascular stenosis and occlusion, while improving the efficiency of parameter acquisition and reducing the cost of parameter acquisition.

[0063] Example 2

[0064] See also Figure 3 , is a structural schematic diagram of a cerebral perfusion parameter acquisition device based on angiography provided in the present application, comprising: a data acquisition module 21, a feature screening module 22 and a parameter prediction module 23.

[0065] In some embodiments, the data acquisition module 21 outputs multiple imaging genomics data features and anatomical data features of cerebral blood vessels based on the acquired cerebral angiography images, and transmits the multiple imaging genomics data features to the feature screening module 22, and transmits the anatomical data features to the parameter prediction module 23; after receiving the multiple imaging genomics data features, the feature screening module 22 outputs the target imaging genomics data features to the parameter prediction module 23; after receiving the anatomical data features and the target imaging genomics data features, the parameter prediction module 23 outputs the final parameter set.

[0066] The data acquisition module 21 is used to acquire cerebral angiography images, segment target blood vessel images from the cerebral angiography images, and extract multiple imaging genomics data features and cerebral blood vessel anatomical data features from the target blood vessel images.

[0067] In some embodiments, a target blood vessel image is segmented from a cerebral angiography image, and multiple imaging omics data features are extracted from the target blood vessel image, including: acquiring an anteroposterior or lateral view, obtaining the target blood vessel image in the anteroposterior or lateral view through a three-dimensional slicing system, and extracting multiple imaging omics data features in the target blood vessel image through the three-dimensional slicing system.

[0068] The feature screening module 22 is used to screen multiple radiomics data features according to the correlation coefficients of the multiple radiomics data features to obtain target radiomics data features.

[0069] In some embodiments, the correlation coefficient includes: a Spearman correlation coefficient; based on the correlation coefficients of multiple imaging omics data features, multiple imaging omics data features are screened to obtain target imaging omics data features, including: based on the Spearman correlation coefficient or Lasso regression between multiple imaging omics data features, target imaging omics data features with significant differences are screened.

[0070] In some embodiments, target imaging omics data features with significant differences are screened out based on the Spearman correlation coefficient or Lasso regression between multiple imaging omics data features, including: performing Lasso regression on multiple imaging omics data features to screen out multiple candidate imaging omics data features whose correlation with cerebral perfusion parameters is greater than a correlation threshold, and obtaining the Spearman correlation coefficient of the multiple candidate imaging omics data features, and selecting the target imaging omics data features whose Spearman correlation coefficient is greater than the coefficient threshold.

[0071] The parameter prediction module 23 is used to take the target imaging genomics data features as the input of a pre-trained classification model, and use the output of the classification model as the prediction result of the cerebral perfusion parameters, and construct the final parameter set based on the anatomical data features and the prediction results.

[0072] In some embodiments, obtaining a classification model includes: using cerebral angiography of normal persons and cerebral angiography of patients as training data, and respectively obtaining cerebral perfusion detection data corresponding to the training data, using the cerebral perfusion detection data as a label of the training data, and training an initial classification model based on machine learning according to the label and the training data to obtain a classification model.

[0073] In some embodiments, an initial classification model based on machine learning is trained according to labels and training data to obtain a classification model, including: taking the training data as the input of the initial classification model, taking the output of the initial classification model as the training result of cerebral perfusion parameter prediction, and obtaining the classification loss function value according to the labels and training results, and updating the parameters of the initial classification model according to the classification loss function value to obtain the classification model.

[0074] In some embodiments, the cerebral perfusion parameters include: the average transit time required for contrast agent to travel from the intracranial arterial side to the intracranial venous side or the contrast agent peak time.

[0075] The present application acquires anatomical data features through cerebral angiography by the data acquisition module 21, and acquires anatomical data features through one angiography. After screening multiple imaging genomics data features by the feature screening module 22, the classification model of the parameter prediction module 23 continues to analyze the cerebral angiography image to acquire cerebral perfusion parameters. Furthermore, acquiring cerebral perfusion parameters based on cerebral angiography can overcome technical bias, achieve accurate assessment of vascular stenosis and occlusion, and at the same time improve parameter acquisition efficiency and reduce parameter acquisition costs.

[0076] Example 3

[0077] See also Figure 4, is a structural diagram of an electronic device provided in the present application, comprising: a memory 31, a processor 32, and a computer program stored in the memory 31 and executable on the processor, wherein the processor 32 implements the method for acquiring cerebral perfusion parameters based on angiography as described in Example 1 when executing the computer program.

[0078] In some embodiments of the present application, the electronic device further includes: a communication interface 33 and a communication bus 34 ; wherein the processor 32 , the communication interface 33 and the memory 31 communicate with each other via the communication bus 34 .

[0079] The present application integrates angiography-based cerebral perfusion parameter acquisition method into an electronic device, and can perform operations such as on-site cerebral perfusion parameter acquisition through various electronic devices. It has a wider range of applicable scenarios and stronger scalability.

[0080] Example 4

[0081] The present application provides a computer-readable storage medium, which stores a plurality of instructions, and the plurality of instructions are suitable for loading by a processor to execute the angiography-based cerebral perfusion parameter acquisition method as described in Example 1.

[0082] When the present application stores the angiography-based cerebral perfusion parameter acquisition method in a storage medium as a program, the angiography-based cerebral perfusion parameter acquisition method can be visualized on different operating systems or operating applications by running or reading the file processing program in the storage medium. It has a wider range of applicable scenarios and stronger scalability.

[0083] Those skilled in the art will appreciate that the embodiments of the present application may also provide computer program products. Therefore, the present application may adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application may adopt the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program codes.

[0084] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0085] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0086] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0087] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for obtaining cerebral perfusion parameters based on angiography, characterized in that: include: Acquiring a cerebral angiography image, segmenting a target blood vessel image from the cerebral angiography image, and extracting a plurality of imaging omics data features and cerebral blood vessel anatomical data features from the target blood vessel image; Screening the multiple radiomics data features according to correlation coefficients of the multiple radiomics data features to obtain target radiomics data features; The target imaging omics data features are used as input to a pre-trained classification model, and the output of the classification model is used as a prediction result of cerebral perfusion parameters. The anatomical data features and the prediction results are used to construct a final parameter set.

2. The method for obtaining cerebral perfusion parameters according to claim 1, characterized in that: Acquiring the classification model includes: The cerebral angiography of normal persons and the cerebral angiography of patients are used as training data, and the cerebral perfusion detection data corresponding to the training data are respectively obtained, and the cerebral perfusion detection data are used as labels of the training data. According to the labels and the training data, an initial classification model based on machine learning is trained to obtain the classification model.

3. The method for obtaining cerebral perfusion parameters according to claim 2, characterized in that: The step of training an initial classification model based on machine learning according to the label and the training data to obtain the classification model includes: The training data is used as the input of the initial classification model, and the output of the initial classification model is used as the training result of cerebral perfusion parameter prediction, and the classification loss function value is obtained according to the label and the training result, and the parameters of the initial classification model are updated according to the classification loss function value to obtain the classification model.

4. The method for obtaining cerebral perfusion parameters according to claim 3, characterized in that: The brain perfusion parameters include: the average transit time required for the contrast agent to travel from the intracranial arterial side to the intracranial venous side or the contrast agent peak time.

5. The method for obtaining cerebral perfusion parameters according to claim 1, characterized in that: The cerebral angiography image includes: a frontal image or a lateral image; the acquisition of the cerebral angiography image, segmentation of a target blood vessel image from the cerebral angiography image, and extraction of multiple imaging omics data features from the target blood vessel image include: The frontal image or the lateral image is collected, a target blood vessel image in the frontal image or the lateral image is obtained through a three-dimensional slicing system, and multiple radiomics data features in the target blood vessel image are extracted through the three-dimensional slicing system.

6. The method for obtaining cerebral perfusion parameters according to claim 1, characterized in that: The correlation coefficient includes: Spearman correlation coefficient; the multiple imaging omics data features are screened according to the correlation coefficients of the multiple imaging omics data features to obtain target imaging omics data features, including: According to the Spearman correlation coefficient or Lasso regression between the multiple radiomics data features, target radiomics data features with significant differences are screened out.

7. The method for obtaining cerebral perfusion parameters according to claim 6, characterized in that: The method of screening out target radiomics data features with significant differences based on the Spearman correlation coefficient or Lasso regression between the multiple radiomics data features includes: Lasso regression is performed on the multiple imaging omics data features to screen out multiple candidate imaging omics data features whose correlation with cerebral perfusion parameters is greater than a correlation threshold, and the Spearman correlation coefficients of the multiple candidate imaging omics data features are obtained, and the target imaging omics data features whose Spearman correlation coefficients are greater than the coefficient threshold are selected.

8. A device for acquiring cerebral perfusion parameters based on angiography, characterized in that: include: Data acquisition module, feature screening module and parameter prediction module; among them, The data acquisition module is used to acquire a cerebral angiography image, segment a target blood vessel image from the cerebral angiography image, and extract a plurality of imaging omics data features and cerebral blood vessel anatomical data features from the target blood vessel image; The feature screening module is used to screen the multiple radiomics data features according to the correlation coefficients of the multiple radiomics data features to obtain target radiomics data features; The parameter prediction module is used to use the target imaging genomics data features as input to a pre-trained classification model, and use the output of the classification model as the prediction result of the cerebral perfusion parameter, and construct a final parameter set based on the anatomical data features and the prediction result.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for acquiring cerebral perfusion parameters based on angiography as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor to execute the method for acquiring cerebral perfusion parameters based on contrast agent as described in any one of claims 1-7.

Images