Intracranial aneurysm rupture state evaluation method based on hemodynamic mode characteristics of injection flow and impact domain

Abstract

The invention discloses an intracranial aneurysm rupture state evaluation method based on haemodynamic mode characteristics of an incident flow and an impact domain, and the method comprises the steps: firstly constructing an intracranial aneurysm three-dimensional geometric model through the image segmentation and three-dimensional reconstruction method based on clinical 3DRA image data; and a computational fluid mechanics method is applied to obtain a hemodynamic cloud picture and corresponding hemodynamic parameters. The method comprises the following steps: determining an impact domain and an incident flow by applying hemodynamic flow field mode feature distribution, further exporting hemodynamic cloud charts in the two areas, constructing an 8-dimensional hemodynamic cloud depth information feature calculation model based on a Pointnet model, and calculating to obtain 8-dimensional depth information features of hemodynamic clouds of the impact domain and the incident flow. Morphological and hemodynamic parameters of the intracranial aneurysm and 8-dimensional depth information characteristics of hemodynamic clouds of an impact domain and an incident flow are integrated, and the rupture state of the intracranial aneurysm is classified and evaluated by adopting a relevance vector machine.

Description

Technical Field

[0001] The present invention relates to the field of clinical intracranial aneurysm rupture risk assessment methods, and provides a new computational analysis and assessment technology for intracranial aneurysm rupture risk assessment, which is applied to clinical aneurysm rupture status classification assessment and treatment intervention decision-making. Background Art

[0002] Intracranial aneurysms are a common cerebrovascular disease caused by local pathological dilatation of intracranial arteries. According to a study published by Etminan in 2016, the incidence of intracranial aneurysms in the population is about 3% to 5%, and there is a trend of increasing year by year. The rupture rate of intracranial aneurysms is about 0.95% to 0.2%. The rupture of aneurysms is the primary cause of subarachnoid hemorrhage, and there is a high probability that the patient will be disabled or even die. Secondly, the current treatment methods for aneurysms cannot guarantee that the aneurysm will not recur or bleed after surgery, and the patient has to bear greater risks during the operation. For some patients, the risk of treatment is even higher than the risk of natural rupture of the aneurysm. Therefore, the rupture risk assessment of intracranial aneurysms is of great clinical significance and can provide an effective reference for the treatment and intervention of intracranial aneurysms.

[0003] At present, the rupture risk assessment of intracranial aneurysms is mainly based on morphological characteristics and hemodynamic characteristics. Morphological characteristics, including size, growth location and shape, will affect the rupture of aneurysms. In addition, the height-to-diameter ratio, size ratio, length-to-width ratio and incident angle of intracranial aneurysms are significantly correlated with the rupture of intracranial aneurysms. The formation, growth and rupture of intracranial aneurysms are processes of hemodynamic action, so hemodynamic characteristics are particularly important for the assessment of the rupture risk of aneurysms. The size and distribution of wall shear stress and shear stress oscillation factor are different between ruptured and unruptured aneurysms. In addition, the blood flow pattern and complexity inside the aneurysm sac can also be used as significant factors in assessing the risk of aneurysm rupture.

[0004] The computational analysis of morphological features usually uses aneurysm images acquired by imaging equipment, and the geometric parameters are measured by software after the aneurysm model is obtained through three-dimensional reconstruction. The calculation of hemodynamic characteristics requires the use of computational fluid dynamics methods for numerical simulation, and the numerical calculation of hemodynamics is performed through meshing, setting boundary conditions, inlet blood flow velocity, etc. Finally, the hemodynamic parameters such as pressure, velocity, wall shear stress, and shear stress oscillation factor are calculated. Satoru's research in 2020 showed that the effectiveness of combining the morphological characteristics and hemodynamic parameters of intracranial aneurysms to assess their rupture risk is significantly better than using morphological characteristics or hemodynamic characteristics alone.

[0005] The impact domain and the injection flow have always been considered to be two key indicators for distinguishing the rupture state of aneurysms. The impact domain refers to the area on the surface of the aneurysm where the blood flows into the aneurysm sac and produces an impact. The impact domain will affect the distribution of wall shear stress on the surface of the aneurysm sac, and the size of the impact domain will also affect the rupture state of the aneurysm. The injection flow refers to the streamline with a higher flow velocity when blood flows into the aneurysm sac area. After the injection flow collides with the blood vessel wall, its direction and speed will change. Cebral's research shows that a more concentrated injection flow and a smaller impact domain significantly increase the probability of aneurysm rupture. However, current research has certain limitations. There is no quantitative standard for the analysis of the injection flow and the impact domain. It is impossible to classify, evaluate and predict the rupture state of intracranial aneurysms through simple qualitative analysis alone. Summary of the invention

[0006] In view of the deficiencies of the prior art, the present invention provides a method for evaluating the rupture status of an intracranial aneurysm based on the hemodynamic pattern characteristics of the injection flow and the impact domain.

[0007] The present invention is achieved through the following technical solutions:

[0008] A method for assessing the rupture status of an intracranial aneurysm based on the hemodynamic pattern characteristics of an injection flow and an impact domain comprises the following steps:

[0009] S1: Computational fluid dynamics numerical simulation is used to perform hemodynamic analysis on intracranial aneurysms and obtain the hemodynamic characteristic distribution of aneurysms;

[0010] S2: Determine the impact domain based on the distribution of high wall shear stress on the tumor surface and the area where blood flow impacts the blood vessel wall, and determine the outflow and inflow based on the direction of blood flow in the tumor area and the area with high flow velocity;

[0011] S3: According to the spatial position of the impact domain and the injection flow, the hemodynamic characteristic distribution cloud map in these two areas is extracted and the data is exported;

[0012] S4: Construct an 8-dimensional deep information feature calculation model of the hemodynamic cloud based on the Pointnet model.

[0013] S5: input the data derived from step S3 into the model constructed in step S4 to calculate and obtain the 8-dimensional depth information features of the hemodynamic cloud of the impact domain and the injection flow;

[0014] S6: The morphological and hemodynamic parameters of intracranial aneurysms as well as the 8-dimensional depth information features of the impact domain and the injection flow hemodynamic cloud are integrated, and a classification model is established using the relevance vector machine algorithm to classify and evaluate the rupture status of intracranial aneurysms.

[0015] Furthermore, in step S2, the area with higher wall shear stress is found based on the distribution cloud map of the wall shear stress on the surface of the tumor capsule, and then the distribution of blood streamlines is observed to find the spatial position where the blood flow impacts the blood vessel wall, and the area of ​​the impact domain is determined based on these two conditions.

[0016] Furthermore, in step S2, according to the direction and speed of the streamline, the flow with a larger speed when the blood flows into the tumor capsule area is determined as the injection flow.

[0017] Furthermore, the data exported in step S3 includes extracting four hemodynamic characteristic cloud maps of wall shear stress, shear stress oscillation factor, pressure and velocity in the impact domain and the corresponding position of the injection flow.

[0018] Furthermore, in step S4, the specific structure of the 8-dimensional deep information feature calculation model of the hemodynamic cloud based on the Pointnet model is as follows: it includes an input conversion module, two multi-layer perceptron MLP (8, 8) modules and a maximum pooling module, and a feature conversion module is set between the two multi-layer perceptron MLP (8, 8) modules.

[0019] Furthermore, the morphological parameters include: aneurysm sac height, aneurysm sac width, aneurysm neck width, aneurysm volume, aneurysm body area, height-to-width ratio, aspect ratio, and size ratio.

[0020] Compared with the prior art, the present invention has the following beneficial effects:

[0021] Cebral et al.'s study showed that a more concentrated injection flow and a smaller impact domain significantly increased the rupture probability of intracranial aneurysms. However, the current research has certain limitations. The characteristic analysis of the injection flow and the impact domain is not quantified, and the classification, evaluation and prediction of the rupture status of intracranial aneurysms cannot be performed by simple qualitative description. The present invention constructs a hemodynamic cloud feature calculation method based on the Pointnet model to extract the hemodynamic cloud features of the impact domain and the injection flow, and obtains 1024, 64, 32, 16, and 8-dimensional depth information features of the hemodynamic cloud of the injection flow and the impact domain, respectively. According to the computational efficiency and contribution weight, it is determined that the 8-dimensional depth information feature of the hemodynamic cloud is a more ideal quantitative information of depth features. The use of the 8-dimensional depth information feature of the hemodynamic cloud can effectively improve the classification, evaluation and prediction performance of the rupture status of intracranial aneurysms. The relevant vector machine algorithm is used to integrate the morphological and hemodynamic features of intracranial aneurysms and the 8-dimensional depth information features of the hemodynamic cloud of the impact domain and the injection flow to classify and evaluate the rupture status of intracranial aneurysms, thereby improving the robustness of the model.

[0022] The method provided by the present invention can provide a new rupture status classification and evaluation technology for clinical intervention of intracranial aneurysms, and has good clinical application value in clinical decision-making for intracranial aneurysms. The use of the 8-dimensional deep information features of the hemodynamic cloud can significantly improve the classification performance and provide a new computational analysis and evaluation technology for clinical treatment intervention decisions for intracranial aneurysms. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Schematic diagram of the computational model framework for hemodynamic cloud characteristics of the impact domain and injection flow based on Pointnet. DETAILED DESCRIPTION

[0024] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to be able to fully convey the scope of the present invention to those skilled in the art. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0025] The present invention discloses a new classification and evaluation method for the rupture state of intracranial aneurysms based on the 8-dimensional depth information characteristics of the impact domain and the injection flow hemodynamic cloud. The main feature of this method is to use the hemodynamic characteristic distribution cloud map in the impact domain and injection flow area of ​​the intracranial aneurysm to extract the 8-dimensional depth information characteristics of the cloud map, and further combine the morphological and hemodynamic characteristic parameters of the aneurysm to classify and evaluate the rupture state of the intracranial aneurysm. It mainly includes the following steps:

[0026] a. Computational fluid dynamics numerical simulation is used to perform hemodynamic analysis on intracranial aneurysms to obtain the hemodynamic characteristic distribution of aneurysms. The impact domain is determined based on the distribution of high wall shear stress on the surface of the aneurysm and the area where blood flow impacts the vascular wall, and the outflow and inflow are determined based on the direction of blood flow in the aneurysm area and the area with high flow velocity.

[0027] b. According to the spatial position of the impact domain and the injection flow, the hemodynamic characteristic distribution cloud map in these two areas is extracted, and the data is exported, mainly extracting four hemodynamic characteristic cloud maps such as wall shear stress, shear stress oscillation factor, pressure and velocity in the corresponding position.

[0028] c. Construct an 8-dimensional deep information feature calculation model of the hemodynamic cloud based on the Pointnet model, and further calculate and obtain the 8-dimensional deep information features of the hemodynamic cloud of the impact domain and the injection flow.

[0029] d. The morphological and hemodynamic parameters of intracranial aneurysms and the 8-dimensional depth information features of the impact domain and the injection flow hemodynamic cloud are integrated, and a classification model is established using the relevance vector machine algorithm to classify and evaluate the rupture status of intracranial aneurysms.

[0030] The technical solution of the present invention is further described below in conjunction with specific embodiments.

[0031] (1) The 3D-DSA clinical imaging data of intracranial aneurysms were acquired by a rotational digital subtraction angiography system (LCV+ digital subtraction angiography machine produced by GE, USA) and saved as image data files in DICOM format.

[0032] (2) Image segmentation based on the level set method was used to obtain the morphological characteristics of intracranial aneurysms. The 3D geometric morphology of intracranial aneurysms was reconstructed based on the Marching Tetrahedra algorithm to obtain a 3D geometric model of intracranial aneurysms. To improve the visualization of the image, the 3D reconstructed image surface was smoothed using a Gaussian filter. After 3D surface reconstruction, a surface model file in the STL (STereo Lithography) format was obtained.

[0033] (3) Geomagic Studio 2013 was used to measure and calculate the morphological parameters of intracranial aneurysms. The morphological parameters include aneurysm sac height, aneurysm sac width, aneurysm neck width, aneurysm volume, aneurysm area, height-to-width ratio, aspect ratio, and size ratio.

[0034] (4) Establish a computational fluid dynamics (CFD) simulation method based on imaging. The Navier-Stokes equation is usually used to simulate and calculate the hemodynamics of IAs based on imaging. The specific implementation includes establishing a three-dimensional geometric model of IAs, meshing the geometric model using tool software such as ANSYS ICEM-CFD, generating a solid model, and then using ANSYS CFX-Post to perform hemodynamic simulation calculations. Thus, a hemodynamic calculation method based on CFD numerical simulation of imaging features is established. Then, the hemodynamic characteristic parameters of IAs are obtained through simulation calculation.

[0035] (5) Four types of hemodynamic cloud maps were obtained by using hemodynamic calculation based on imaging CFD, including wall shear stress WSS cloud map, shear stress oscillation factor OSI cloud map, blood pressure P cloud map and blood flow velocity V cloud map.

[0036] (6) Based on the distribution cloud map of the wall shear stress on the surface of the tumor capsule, find the area with higher wall shear stress, and then observe the distribution of blood streamlines to find the spatial position where the blood flow impacts the blood vessel wall. Based on these two conditions, determine the area of ​​the impact domain. According to the direction and speed of the streamlines, the flow with a higher speed when the blood flow enters the tumor capsule area is determined as the injection flow.

[0037] (7) Extract the hemodynamic cloud map of the impact domain and the injection flow area and save the data.

[0038] (8) An 8-dimensional deep information feature calculation model of the hemodynamic cloud based on the Pointnet model was constructed. The model includes an input conversion module, two multi-layer perceptron MLP (8, 8) modules and a maximum pooling module. A feature conversion module is set between the two multi-layer perceptron MLP (8, 8) modules. The model is trained using sample data of ruptured and unruptured intracranial aneurysms.

[0039] (9) Input the data extracted in step (7) into the computational model trained in step (8) to perform calculations to obtain the 8-dimensional depth information features of the hemodynamic cloud of the impact domain and the injection flow. Figure 1 As shown in the figure, the extracted wall shear stress WSS cloud map, shear stress oscillation factor OSI cloud map, blood pressure P cloud map and blood flow velocity V cloud map of the impact domain and injection flow area are input into the 8-dimensional deep information feature calculation model of the point cloud. The T-Net 7×7 in the input conversion module aligns the 1024×7 input point cloud to ensure that the point cloud maintains rotation invariance. The first multi-layer perceptron MLP (8, 8) module with shared weights is used to extract features from each point in the cloud to obtain the corresponding feature matrix, thereby obtaining the local features of the point cloud and outputting 1024×8-dimensional features. Then, the 1024×8-dimensional features are aligned by T-Net 8×8 in the feature conversion module, and then the 1024×8-dimensional features are output by the second multi-layer perceptron MLP (8, 8) module with shared weights. The maximum pooling module gathers all point cloud features and performs maximum pooling to obtain the 1×8-dimensional deep information features of the point cloud.

[0040] (10) The morphological and hemodynamic parameters of intracranial aneurysms as well as the 8-dimensional depth information features of the impact domain and the injection flow hemodynamic cloud are integrated, and a classification model is established using the relevance vector machine algorithm to classify and evaluate the rupture status of intracranial aneurysms and determine whether the intracranial aneurysms are ruptured.

[0041] The present invention proposes a classification and assessment method for the rupture status of intracranial aneurysms based on the hemodynamic pattern characteristics of the injection flow and the impact domain. A method for calculating the 8-dimensional depth information characteristics of the hemodynamic cloud based on the Pointnet model was created. The 8-dimensional depth information characteristics of the hemodynamic cloud of the impact domain and the injection flow were calculated and obtained for the first time. The correlation vector machine algorithm was used to integrate the morphological characteristics and hemodynamic characteristics of intracranial aneurysms and the 8-dimensional depth information characteristics of the hemodynamic cloud of the impact domain and the injection flow to classify and assess the rupture status of intracranial aneurysms. The classification accuracy was 83.5% and the AUC was 93.2%, achieving good classification performance and improving the robustness of the model. It provides a new computational analysis technology for the clinical rupture risk assessment of intracranial aneurysms, and also provides an effective basis for clinical decision-making on aneurysms.

[0042] The present invention has been described in detail above through embodiments, but the contents described are only exemplary embodiments of the present invention and cannot be considered to limit the scope of implementation of the present invention. The protection scope of the present invention is defined by the claims. Anyone who utilizes the technical solution described in the present invention, or a technician in the field, inspired by the technical solution of the present invention, designs a similar technical solution within the essence and protection scope of the present invention to achieve the above technical effects, or makes equal changes and improvements to the scope of application, shall still belong to the scope of protection covered by the patent of the present invention. It should be noted that in order to make a clear statement, the description of some components and processes that have no direct and obvious connection with the scope of protection of the present invention but are known to those skilled in the art are omitted in the description of the present invention.

Claims

1. A method for assessing the rupture status of an intracranial aneurysm based on the hemodynamic pattern characteristics of the injection flow and the impact domain, characterized in that: The following steps are involved: S1: Computational fluid dynamics numerical simulation is used to perform hemodynamic analysis on intracranial aneurysms and obtain the hemodynamic characteristic distribution of aneurysms; S2: Determine the impact domain based on the distribution of high wall shear stress on the tumor surface and the area where blood flow impacts the blood vessel wall, and determine the outflow and inflow based on the direction of blood flow in the tumor area and the area with high flow velocity; S3: According to the spatial position of the impact domain and the injection flow, the hemodynamic characteristic distribution cloud map in these two areas is extracted and the data is exported; S4: Construct an 8-dimensional deep information feature calculation model of the hemodynamic cloud based on the Pointnet model. S5: input the data derived from step S3 into the model constructed in step S4 to calculate and obtain the 8-dimensional depth information features of the hemodynamic cloud of the impact domain and the injection flow; S6: The morphological and hemodynamic parameters of intracranial aneurysms as well as the 8-dimensional depth information features of the impact domain and the injection flow hemodynamic cloud are integrated, and a classification model is established using the relevance vector machine algorithm to classify and evaluate the rupture status of intracranial aneurysms.

2. A method for evaluating the rupture state of an intracranial aneurysm based on the hemodynamic pattern characteristics of the injection flow and the impact domain according to claim 1, characterized in that: In step S2, the area with higher wall shear stress is found based on the distribution cloud map of the wall shear stress on the surface of the tumor capsule, and then the distribution of blood streamlines is observed to find the spatial position where the blood flow impacts the blood vessel wall, and the area of ​​the impact domain is determined based on these two conditions.

3. A method for assessing the rupture status of an intracranial aneurysm based on the hemodynamic pattern characteristics of the injection flow and the impact domain according to claim 1, characterized in that: In step S2, according to the direction and speed of the streamline, the flow with a larger speed when the blood flows into the tumor sac area is determined as the injection flow.

4. A method for evaluating the rupture status of an intracranial aneurysm based on the hemodynamic pattern characteristics of the injection flow and the impact domain according to claim 1, characterized in that: The data exported in step S3 includes extracting four hemodynamic characteristic cloud maps of wall shear stress, shear stress oscillation factor, pressure and velocity in the impact domain and the corresponding position of the injection flow.

5. The method for evaluating the rupture status of an intracranial aneurysm based on the hemodynamic pattern characteristics of the injection flow and the impact domain according to claim 1, characterized in that: The specific structure of the 8-dimensional deep information feature calculation model of the hemodynamic cloud based on the Pointnet model constructed in step S4 is as follows: it includes an input conversion module, two multi-layer perceptron MLP (8, 8) modules and a maximum pooling module, and a feature conversion module is set between the two multi-layer perceptron MLP (8, 8) modules.

6. A method for evaluating the rupture status of an intracranial aneurysm based on the hemodynamic pattern characteristics of the injection flow and the impact domain according to claim 1, characterized in that: The morphological parameters include: aneurysm sac height, aneurysm sac width, aneurysm neck width, aneurysm volume, aneurysm area, height-to-width ratio, aspect ratio, and size ratio.

Images