Vascular respiratory movement map construction method, device, equipment and medium

Abstract

The invention discloses a vascular respiratory movement map construction method, device and equipment and a medium, and relates to the technical field of medical treatment, the method establishes a vascular digraph based on a topological connection relationship in a vascular structure, and quantifies the overall displacement of a blood vessel and the posture of each branch through the extraction of structure control points and the calculation of branch bending energy. The propagation direction of motion between branch points is defined, and consistency constraint is provided for vascular motion modeling. Vascular deformation in motion modeling is quantified by adopting adaptive deformation compensation, and motion deformation of each section of a branch is represented by using rotation measurement and stretching measurement under a spherical coordinate system of a manifold space. Meanwhile, the bending energy weight is adaptively distributed for each branch based on the vascular graph and the SRVF constraint, and the learning efficiency of the model for local attitude deformation is improved.

Description

Technical Field

[0001] The present invention relates to the field of medical technology, and in particular to a method, device, equipment and medium for constructing a vascular respiratory motion atlas based on topological constraints and deformation compensation. Background Art

[0002] In hepatic vascular interventional surgery, the displacement and deformation of hepatic vessels caused by the patient's free breathing pose a major challenge to the accuracy and safety of the surgery. Usually, interventional physicians use preoperative computed tomography (CTA) to obtain the patient's three-dimensional (3D) vascular structure, and combine it with the enhanced vascular branches in the intraoperative X-ray angiography (XRA) sequence to complete the operation through spatial imagination and clinical experience [1]. This method of estimating vascular motion based on experience is prone to deviation in complex situations, thus affecting the surgical effect. In order to quantify the movement of vascular branches during surgery, contrast agent is usually continuously injected until the blood vessels are filled, and then a two-dimensional (2D) vascular motion model is established based on the periodic movement of the visible branches during the respiratory cycle. These models are then used to generate dynamic vascular roadmaps and superimpose them on real-time X-ray fluoroscopic image sequences to provide intraoperative navigation support for interventional operations. On this basis, Ma et al. compensated for the vascular motion caused by heartbeat and breathing by aligning 2D image frames with ECG signals and tracking the catheter tip in motion modeling. Wang et al. input high-resolution features into the detection network to extract more detailed vascular motion features from X-ray images; Yang et al. used an optical flow-based method to extract the displacement and deformation of angiographic vascular branches during respiratory motion.

[0003] At the same time, in order to complete the 2D motion modeling of blood vessels and reduce the use of contrast agents during surgery, many studies have used 3D / 2D registration methods to project the preoperative vascular structure onto the intraoperative X-ray image sequence, and then combined with 2D motion modeling to obtain the 2D dynamic posture of the blood vessel. Baka et al. proposed a method based on a directional Gaussian mixture model to reduce the 3D / 2D initial posture difference and image noise interference, thereby improving the accuracy of vascular motion estimation. Ambrosini et al. registered 3D blood vessels with 2D guidewires based on a hidden Markov model, and guided the projection of 3D vascular branches to corresponding positions through the trajectory of the guidewire between frames.

[11] proposed a population-based method to establish a coronary artery motion model to provide constraints for 2D+t / 3D registration, including the average motion range of the coronary artery and the change in the shape of the heart when the coronary artery deforms. In order to further improve the estimation accuracy of deformation motion, a neural network based on deep learning was proposed. Using a segment-by-segment registration method, the center line that retains the vascular topology information is represented by the origin tensor and the shape tensor, and the input network is used to predict the deformation field of each segment. Although the above methods can generate dynamic poses of blood vessels on 2D images, their outputs are still limited to planar information.

[0004] In order to complete the 3D respiratory motion modeling of blood vessels, it is necessary to quantify the 3D posture changes of blood vessels under different respiratory states. However, current research focuses on the 3D motion modeling of organs. Usually, such studies use 4D-CT scans to obtain organ structures at different phases and extract features to characterize the position and morphological changes of organs. Common methods include using principal component analysis (PCA) to implement statistical shape modeling (SSM), which extracts edge feature points of organs to sparsely represent dense shape networks to generate organ shape templates. The further developed statistical deformable model (SDM) establishes a deformation library for the posture changes of multiple organs during respiratory motion. It not only performs local statistics and quantification on different regions of the organs, but also incorporates the mutual motion association between multiple organs into the model. In order to make the motion template closer to the real 3D organ motion of the patient during surgery, the patient's intraoperative X-ray image sequence is associated with the 3D motion template through 2D / 3D deformation registration. Under the guidance of the diaphragm or lung contours, these methods can achieve personalized dynamic deformation prediction for the liver and lungs. In addition to organ structure-based analysis, there are also studies that directly model motion in 4D-CT images. Fu et al. proposed a group registration method based on global-local graph shrinkage to construct a 3D motion map of the liver.

[0005] Similarly, in order to obtain the liver vascular structure under different respiratory states, 4D-CTA scanning is usually used in clinical practice to obtain CT images under multiple respiratory phases, and then the vascular branch distribution of each phase is extracted using vascular segmentation technology. Some 3D point cloud registration algorithms have been used to calculate the transformation relationship between the initial vascular pose and the target pose. For example, the iterative closest point algorithm (ICP) using the nearest neighbor matching strategy, the coherent point drift algorithm (CPD) that completes the registration by probability density estimation, and the centerline-guided Gaussian mixture model-based registration algorithm (CG-GMM). In view of the rich characteristics of vascular branches, Zhu et al. quantified the matching score through a heuristic tree search strategy and optimized the node matching to obtain a registration matrix with a higher score. In addition, non-rigid registration gradually reduces the rigid weight through iterative optimization, so that the initial structure is locally deformed to adapt to the target structure. Khallaghi et al. used the finite element model (FEM) to generate a deformation field based on the prior knowledge of the biomechanical model to complete non-rigid registration. Although this method can accurately characterize complex deformation features, it has high computational complexity and is highly dependent on the alignment accuracy of the initial pose.

[0006] To this end, manifold regularization constraints based on Hilbert space transformation are introduced into the registration. This method can extract the intrinsic information of the structure from unmatched feature point pairs by balancing the weight constraints of structural deformation and shape preservation. However, how to quickly and accurately estimate the motion law of hepatic blood vessels and extract effective features from multi-patient and multi-phase vascular structures to establish a universal motion model remains a challenge that needs to be solved. Summary of the invention

[0007] In view of the above problems, the present invention provides a method, device, equipment and medium for constructing a vascular respiratory motion map for overcoming the above problems or at least partially solving the above problems.

[0008] The present invention provides the following scheme:

[0009] A method for constructing a vascular respiratory motion atlas, comprising:

[0010] Acquiring 4D-CTA image data of the patient, wherein the 4D-CTA image data includes a blood vessel image;

[0011] Inputting the 4D-CTA image data into the constructed respiratory motion model so that the respiratory motion model outputs a 4D vascular motion atlas;

[0012] The processing process of the respiratory motion model includes data preprocessing, topological structure constraints, adaptive deformation compensation constraints and overall structure preservation constraints;

[0013] The data preprocessing includes extracting the vascular structure and vascular centerline from the 4D-CTA image using a TotalSegmentator algorithm and a morphological algorithm, respectively, and calculating the lumen diameter at each centerline point;

[0014] The topological structure constraint includes establishing a directed graph of blood vessels based on the topological connection relationship within the blood vessel structure, and quantifying the overall displacement of the blood vessels and the posture of each branch by extracting structural control points and calculating branch bending energy;

[0015] The adaptive deformation compensation constraint includes using rotation metric and stretching metric in the spherical coordinate system of the manifold space to characterize the motion deformation of each segment of the branch to achieve quantification of the deformation of each branch, and adaptively assigning bending energy weights to each branch based on the vascular map and the SRVF constraint;

[0016] The overall structure preservation constraint includes jointly optimizing the skeleton integrity and lumen preservation during deformation using a loss minimization metric.

[0017] Preferably: the structural control points include vascular branch endpoints, bifurcation points and branch points, and distance vectors between the structural control points are established to characterize the direction and movement amplitude of each branch displacement vector to approximate the movement of a dense vascular point set;

[0018] The branch bending energy of the corresponding segment is quantified by calculating the curvature of the dense points on the centerline of the segment branch.

[0019] Preferably, the topological structure constraint is expressed by the following formula:

[0020]

[0021] Where: C represents the curvature matrix, represents the displacement matrix of the center line point set, and La represents the Laplacian matrix.

[0022] Preferably, the method of using rotation measurement and stretching measurement in the spherical coordinate system of the manifold space to characterize the motion deformation of each segment of the branch to achieve quantification of the deformation of each branch includes:

[0023] The branch structure β is transformed from the spatial rectangular coordinate system (x, y, z) to the spherical coordinate system (r, θ, φ). The rotation is quantified by measuring the change in the direction of the tangent vector of the curve f on the branch β along the deformation path, and the stretching is quantified by the change in the length of the curve along the deformation path.

[0024] Preferably, the adaptive deformation compensation constraint is expressed by the following formula:

[0025]

[0026] Where W represents the branch bending energy weight matrix, O represents the branch rotation metric matrix, and L represents the branch length metric matrix.

[0027] Preferably: the branch rotation metric matrix includes all branch rotation metrics, and each branch rotation metric is represented by the following formula:

[0028]

[0029] Where: a and b represent constant weights, represents the unit vector tangent to the curve f at point s, e φ(s) Then it represents the magnitude of the tangent vector at point s;

[0030] The branch length metric matrix includes all branch length metrics, and the branch length metric is expressed by the following formula:

[0031]

[0032] Where: is the velocity vector, and The inner product 〈·〉 serves as a deterministic metric operator.

[0033] Preferably: the overall structure holds the constraint represented by the following formula:

[0034]

[0035] Where: Ls(Δv) and Ls(Δr) represent the absolute values ​​of the differences in vascular diameter and volume at different respiratory phases, respectively. m is the bending energy weight corresponding to each point on the branch centerline, and num represents the number of branches in the current vascular structure.

[0036] A vascular respiratory motion atlas construction device, used to execute the above-mentioned vascular respiratory motion atlas construction method, the device comprising:

[0037] An image acquisition unit, used to acquire 4D-CTA image data of a patient, wherein the 4D-CTA image data includes a blood vessel image;

[0038] A motion atlas output unit, used for inputting the 4D-CTA image data into the constructed respiratory motion model, so that the respiratory motion model outputs a 4D motion atlas of blood vessels;

[0039] The processing process of the respiratory motion model includes data preprocessing, topological structure constraints, adaptive deformation compensation constraints and overall structure preservation constraints;

[0040] The data preprocessing includes extracting the vascular structure and vascular centerline from the 4D-CTA image using a TotalSegmentator algorithm and a morphological algorithm, respectively, and calculating the lumen diameter at each centerline point;

[0041] The topological structure constraint includes establishing a directed graph of blood vessels based on the topological connection relationship within the blood vessel structure, and quantifying the overall displacement of the blood vessels and the posture of each branch by extracting structural control points and calculating branch bending energy;

[0042] The adaptive deformation compensation constraint includes using rotation metric and stretching metric in the spherical coordinate system of the manifold space to characterize the motion deformation of each segment of the branch to achieve quantification of the deformation of each branch, and adaptively assigning bending energy weights to each branch based on the vascular map and the SRVF constraint;

[0043] The overall structure preservation constraint includes jointly optimizing the skeleton integrity and lumen preservation during deformation using a loss minimization metric.

[0044] A device for constructing a vascular respiratory motion atlas, the device comprising a processor and a memory:

[0045] The memory is used to store program code and transmit the program code to the processor;

[0046] The processor is used to execute the above-mentioned method for constructing a vascular respiratory motion map according to the instructions in the program code.

[0047] A computer-readable storage medium, characterized in that the computer-readable storage medium is used to store program code, and the program code is used to execute the above-mentioned vascular respiratory motion map construction method.

[0048] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0049] The embodiments of the present application provide a method, device, equipment and medium for constructing a vascular respiratory motion map. A directed vascular graph is established based on the topological connection relationship within the vascular structure. The overall displacement of the blood vessel and the posture of each branch are quantified by extracting the structural control points and calculating the branch bending energy. The propagation direction of the motion between the branch points is defined, providing a consistency constraint for vascular motion modeling. Adaptive deformation compensation is used to quantify the vascular deformation in motion modeling, and the rotation metric and stretching metric are used in the spherical coordinate system of the manifold space to characterize the motion deformation of each segment of the branch. At the same time, based on the vascular graph and SRVF constraints, bending energy weights are adaptively allocated to each branch, which improves the learning efficiency of the model for local posture deformation. In addition, a vascular structure preservation constraint is introduced. By designing a loss minimization metric, the skeleton integrity and lumen retention during the deformation process are jointly optimized to avoid topological structure fractures and vascular volume changes. This method can effectively capture the dynamic deformation characteristics of hepatic blood vessels and provide reliable motion modeling support for interventional surgical navigation under complex respiratory conditions.

[0050] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0052] Figure 1 is a flow chart of a method for constructing a vascular respiratory motion atlas provided by an embodiment of the present invention;

[0053] Figure 2 It is a framework diagram of a method for constructing a vascular respiratory motion atlas provided by an embodiment of the present invention;

[0054] Figure 3is a schematic diagram of a device for constructing a vascular respiratory motion atlas provided by an embodiment of the present invention;

[0055] Figure 4 It is a schematic diagram of a vascular respiratory motion atlas construction device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0057] See also Figure 1 , is a method for constructing a vascular respiratory motion map provided by an embodiment of the present invention, such as Figure 1 As shown, the method may include:

[0058] S101: Acquire 4D-CTA image data of a patient, wherein the 4D-CTA image data includes a vascular image;

[0059] S102: inputting the 4D-CTA image data into the constructed respiratory motion model so that the respiratory motion model outputs a 4D blood vessel motion atlas;

[0060] The processing process of the respiratory motion model includes data preprocessing, topological structure constraints, adaptive deformation compensation constraints and overall structure preservation constraints;

[0061] The data preprocessing includes extracting the vascular structure and vascular centerline from the 4D-CTA image using TotalSegmentator and morphological algorithm respectively, and calculating the lumen diameter at each centerline point;

[0062] The topological structure constraint includes establishing a directed graph of blood vessels based on the topological connection relationship within the blood vessel structure, and quantifying the overall displacement of the blood vessels and the posture of each branch by extracting the structural control points and calculating the branch bending energy; in specific implementation, the embodiment of the present application can provide that the structural control points include the endpoints, bifurcation points and branch points of blood vessel branches, and establish the distance vectors between the structural control points to characterize the direction and movement amplitude of the displacement vectors of each branch to approximate the movement of the dense blood vessel point set;

[0063] The branch bending energy of the corresponding segment is quantified by calculating the curvature of the dense points on the centerline of the segment branch.

[0064] Furthermore, the topological structure constraint is expressed by the following formula:

[0065]

[0066] Where: C represents the curvature matrix, represents the displacement matrix of the center line point set, and La represents the Laplacian matrix.

[0067] The adaptive deformation compensation constraint includes using a rotation metric and a stretching metric in a spherical coordinate system of a manifold space to characterize the motion deformation of each segment of the branch to achieve quantification of the deformation of each branch, and adaptively assigning a bending energy weight to each branch based on a vascular map and an SRVF constraint; in specific implementation, the embodiment of the present application can provide that the use of a rotation metric and a stretching metric in a spherical coordinate system of a manifold space to characterize the motion deformation of each segment of the branch to achieve quantification of the deformation of each branch includes:

[0068] The branch structure β is transformed from the spatial rectangular coordinate system (x, y, z) to the spherical coordinate system (r, θ, φ). The rotation is quantified by measuring the change in the direction of the tangent vector of the curve f on the branch β along the deformation path, and the stretching is quantified by the change in the length of the curve along the deformation path.

[0069] The adaptive deformation compensation constraint is expressed by the following formula:

[0070]

[0071] Where W represents the branch bending energy weight matrix, O represents the branch rotation metric matrix, and L represents the branch length metric matrix.

[0072] Furthermore, the branch rotation metric matrix includes all branch rotation metrics, and each branch rotation metric is represented by the following formula:

[0073]

[0074] Where: a and b represent constant weights, represents the unit vector tangent to the curve f at point s, and eφ(s) represents the magnitude of the tangent vector at point s;

[0075] The branch length metric matrix includes all branch length metrics, and the branch length metric is expressed by the following formula:

[0076]

[0077] Where: is the velocity vector, and The inner product 〈·〉 serves as a deterministic metric operator.

[0078] The overall structure preservation constraint includes jointly optimizing the skeleton integrity and lumen preservation during deformation using a loss minimization metric.

[0079] In a specific implementation, the embodiment of the present application can provide that the overall structure retention constraint is expressed by the following formula:

[0080]

[0081] Where: Ls(Δv) and Ls(Δr) represent the absolute values ​​of the differences in vascular diameter and volume at different respiratory phases, respectively. m is the bending energy weight corresponding to each point on the branch centerline, and num represents the number of branches in the current vascular structure.

[0082] The method for constructing a vascular respiratory motion map provided in the embodiment of the present application establishes a directed vascular graph based on the topological connection relationship within the vascular structure, and quantifies the overall displacement of the vascular and the posture of each branch by extracting the structural control points and calculating the branch bending energy. This defines the propagation direction of the motion between the branch points and provides consistency constraints for vascular motion modeling.

[0083] Adaptive deformation compensation is proposed to quantify the vascular deformation in motion modeling. The rotation metric and stretching metric are used to characterize the motion deformation of each branch segment in the spherical coordinate system of the manifold space. At the same time, the bending energy weight is adaptively assigned to each branch based on the vascular graph and SRVF constraints, which improves the learning efficiency of the model for local posture deformation. The vascular structure preservation constraint is introduced. By designing a loss minimization metric, the skeleton integrity and lumen retention during the deformation process are jointly optimized to avoid topological structure breaks and vascular volume changes.

[0084] The following is a detailed introduction to the method for constructing the vascular respiratory motion atlas provided in this application.

[0085] like Figure 2 As shown, the method for constructing a vascular respiratory motion atlas provided in the embodiment of the present application includes four parts, namely data preprocessing, topological connection constraints, adaptive deformation compensation and overall structure maintenance. The input data is 4D-CTA image data of multiple patients. After being processed by the respiratory motion model, a 4D motion atlas of the blood vessels is output. The atlas contains the vascular deformation field between multiple respiratory phases, which is used to predict the continuous vascular posture changes in the respiratory cycle.

[0086] 1. Data preprocessing.

[0087] In preprocessing, the present application uses TotalSegmentator and morphological algorithms to extract vascular structure and vascular centerline from 4D-CTA images, and calculates the lumen diameter at each centerline point. After completing sampling normalization and coordinate alignment of the vascular point set of each phase (0%-90%), the vascular data of two different respiratory phases are randomly selected from multiple phases in the 4D-CTA image for motion analysis, and the combination of all phase data is traversed to finally complete the respiratory motion modeling and obtain the 4D respiratory motion atlas of the blood vessels.

[0088] 2. Topological connection constraints.

[0089] The vascular topology exhibits geometric invariance during respiratory motion, which keeps the connection relationship between branches unchanged during motion. In order to ensure the accuracy of vascular deformation compensation and improve the computational efficiency of respiratory modeling, this application introduces topological connectivity constraints to maintain the structural consistency of blood vessels under different respiratory states.

[0090] First, the application constructs a topological structure based on the input vascular centerline point set. It can be defined as a curve obtained by treating the vascular bifurcation points and branch endpoints as vertices and connecting the vertices. It describes the connection relationship of the vascular centerline points and the direction of the blood vessels. The graph model is the most common way to express the vascular topological structure. After obtaining the coordinates of the vascular centerline point set, the root node, endpoint, bifurcation point, branch point and vascular segmentation curve of the blood vessel can be identified. The application defines these points as the structural control points of the blood vessel.

[0091] Through centerline point set The topological connectivity between the two nodes is established in this application. The adjacency matrix A∈[0,1] K×K and the metric matrix D K×K , where K represents the number of structural control points and defines the consistency constraints of the vascular structure based on topological connections in motion modeling. A(i, j) = 1 means that the i-th point is connected to the j-th point, and the rest of the states A(i, j) = 0. The value of D(i, j) represents the number of branches connected at the i-th centerline point. The movement of each branch node propagates only in one direction along the topological direction, and the other driven nodes and branch points can be uniquely determined by A and D.

[0092] At the same time, as the topological connectivity between nodes is established, the displacement of topological adjacent points becomes consistent. This application uses the topological regularization term To constrain these topological adjacencies:

[0093]

[0094] in, represents the displacement matrix of the centerline point set, Constrain the length change of the branch centerline. Through the definition of Laplace graph, It can be represented by a manifold, and the topological relationship can be expanded to the Laplacian matrix La = DA, where Therefore, formula (1) can be expressed in the following matrix form:

[0095]

[0096] Here, tr(·) represents the trace of the matrix.

[0097] Based on the identified vascular branch endpoints, bifurcation points, and branching points, the present application can construct a displacement matrix. Constructing the displacement matrix involves the vascular centerlines of two different respiratory phases, including the initial phase and the target phase. Each element τ in the displacement matrix ij Represents the spatial distance from the i-th point of the initial phase blood vessel centerline to the j-th point of the target phase blood vessel centerline. On this basis, this application establishes a distance vector between structural control points to characterize the direction and movement amplitude of each branch displacement vector to fully approximate the movement of the dense blood vessel point set.

[0098] During respiratory movement, the curvature of the segmented curve of the blood vessel is also an important feature in motion modeling. As the curvature of the branch increases, the accumulated bending elastic potential energy inside the branch structure will cause the local area to reach the maximum energy entropy (LME), causing this part to be more susceptible to deformation during respiratory movement. Therefore, this application is based on the constructed topological connectivity and quantifies the bending energy of the segment by calculating the curvature of the dense points on the center line of the segmented branch. Here, this application uses three-dimensional Bezier curve fitting to realize energy calculation. Assume that P0, P1, and P2 are any three points on the center line of the same branch, P0, P1, and P2∈X, and the Bezier curve equation is defined as follows:

[0099] B(μ)=(1-μ) 2 P0+2(1-μ)μP1+μ 2 P2 (3)

[0100] Where μ∈[0,1], P0, P1, P2 represent the three-dimensional coordinates. In order to obtain the curvature The first and second derivatives of a Bezier curve are calculated as follows:

[0101] B′(μ)=2(1-μ)(P1-P0)+2μ(P2-P1), B″(μ)=2(P2-2P1+P0). (4)

[0102] Then the curvature κ(μ) can be calculated by the following formula:

[0103]

[0104] Where × represents the vector cross product and |·| represents the vector modulus operation. The obtained κ(μ) of each branch of the multi-phase is statistically recorded as the curvature matrix C, where each element c ij ∈C represents the curvature value of the segment obtained by connecting the i-th point and the j-th point on the branch along the center line. In summary, the topological connection constraint constructed by the application based on the center line point set It includes branch structure consistency measurement, branch displacement measurement and branch bending measurement, which can be expressed by the following formula:

[0105]

[0106] 3. Adaptive deformation compensation.

[0107] During respiratory movement, due to inconsistent deformations in different areas of the liver parenchyma, the deformation of the hepatic blood vessels wrapped by the liver parenchyma is also diversely distributed on each branch. Here, the present application proposes adaptive deformation compensation to enhance the posture learning of the motion model for local branches and reduce the interference of deformation errors. The essence of branch deformation diversity lies in the inconsistency of rotation and stretching of various branches during breathing. To solve this problem, the present application uses non-rigid measurement to quantify bending features.

[0108] First, the present application transforms the branch structure β from the spatial rectangular coordinate system (x, y, z) to the spherical coordinate system (r, θ, φ), quantifies the rotation by measuring the change in the direction of the tangent vector of the curve f on the branch β along the deformation path, and quantifies the stretching by measuring the change in the length of the curve along the deformation path. Where f∈β, the deformation path α can be expressed as It means that during the respiratory phase t1 to t2, the branch Transform to and get Represents a Hilbert manifold space The branch β can be parameterized as Where D is a parameterized specific domain. This application restricts the branches β to be absolutely continuous on D, which is usually set to [0, 1].

[0109] Furthermore, the stretching of the branch should be related to the elongation of the branch curve f, which can be quantified by observing the magnitude of the change in the tangent vector along the curve f at each point s over the deformation path. The branch length metric L[β] can be defined as:

[0110]

[0111] in, is the velocity vector, and The inner product 〈·〉 serves as a deterministic metric operator.

[0112] In fact, the branch β can be obtained by the following formula:

[0113]

[0114] Then for a continuous curve f at every point s, we have:

[0115]

[0116] where θ(s) is the unit vector tangent to the curve f at point s, and e φ(s) Then it represents the size of the tangent vector, which is defined as:

[0117]

[0118] The branch rotation metric is defined as the weighted sum of φ and θ along the deformation path α:

[0119]

[0120] Where a and b represent constant weights. The first term of the above equation quantifies the change in elevation angle θ, and the second term quantifies the change in azimuth angle φ. The sum of these two terms is equivalent to The rotation metric between curves in space, combined with the branch length metric L[β], can effectively quantify the deformation process of the branch structure. In order to improve the computational efficiency of learning the branch β posture, this application establishes a mapping relationship for the branch β through the Square Root Velocity Function (SRVF) The specific form is:

[0121]

[0122] Here ‖·‖ is the Euclidean 2-norm norm of the real space, and F is a continuous mapping. Here, the present application uses the SRVF form of elastic metric to replace the posture of each vascular branch, and preferentially completes the SRVF mapping in the area near the structural control point (structural control area). Then, the deformation constraint of branch β is It can be expressed as:

[0123]

[0124] Among them, c is a constant weight, just like a and b.

[0125] After completing the quantification of the deformation of each branch, the present application needs to adaptively compensate for the deformation on each branch. Based on the Bezier curve, the present application calculates the bending energy of the segmented branches. For the entire vascular structure, the branch segments with larger bending energy are more likely to deform first during respiratory movement, and then drive other segments on the branch topology structure to deform. Therefore, the present application allocates weights based on the bending energy distribution of all branches in the current vascular structure. If the current vascular structure has num branches, the sum of all branch energies is The weight of the deformation constraint of each branch is w m =κ(μ) m / κ sum Therefore, the adaptive deformation compensation constraint for the branch can be updated as:

[0126]

[0127] Rewritten in matrix form:

[0128]

[0129] 4. Vascular structure is maintained.

[0130] After completing the adaptive deformation compensation of the vascular branches, the maintenance of the integrity of the vascular branch structure can further include the diverse vascular postures of multiple patients in motion modeling. The present application decomposes it into the maintenance of the integrity of the vascular skeleton structure and the maintenance of the invariance of the branch lumen volume, which can ensure that the topological structure control points are not lost, the topological connections are not broken, and the lumen diameter is not degraded during the movement. The present application performs a branch decomposition operation on the deformed vascular structure through the extracted vascular centerline, so that the deformation of each branch during the respiratory movement is relatively independent.

[0131] At the same time, in order to keep the lumen volume of the blood vessel unchanged during the deformation from the initial structure to the target structure, the present application uses the morphological method in the preprocessing process to calculate the diameter r and the total lumen volume v of the blood vessel at the corresponding centerline point before and after deformation, that is, the number of voxels occupied by the blood vessel, and designs a loss minimization metric to jointly optimize the diameter loss and volume loss during the deformation of the blood vessel posture. Transform to The diameter of a pipe at a point on the center line is given by becomes The total vascular volume is becomes And because the more severe the branch deformation is, the easier it is for the tube diameter to change, this application introduces the branch bending energy weight matrix W, so the overall structure retention constraint can be defined as:

[0132]

[0133] Among them, Ls(Δv) and Ls(Δr) represent the absolute values ​​of the differences in vascular diameter and volume at different respiratory phases, respectively. m is the bending energy weight corresponding to each point on the branch centerline. Therefore, the respiratory motion model constructed in this application based on topological structure constraints, adaptive deformation compensation constraints and overall structure preservation constraints can be defined as the following formula:

[0134]

[0135] Experimental results show that the proposed method is superior to the latest motion modeling methods and achieves vascular motion estimation errors in clinical free breathing mode: Hausdorff distance is 1.04±0.27 mm and root mean square error is 1.26±0.26 mm.

[0136] In summary, the method for constructing a vascular respiratory motion atlas provided in this application aims to explore the motion laws of hepatic blood vessels and construct their 4D motion atlas. First, the vascular graph structure is constructed based on the topological connection relationship between vascular branches, and the overall consistent displacement and branch posture of the blood vessels are quantified by extracting structural control points and calculating bending energy. Subsequently, the Euclidean coordinates of the vascular branches are converted into manifold coordinates using the spherical coordinate system of the Hilbert manifold space, and the deformation information of the blood vessels is characterized by quantifying the rotation and stretching of the branches. On this basis, the local deformation of each branch is adaptively compensated in combination with the vascular topology map and the square root velocity function (SRVF) constraint. Finally, the vascular structure preservation constraint is introduced to ensure the consistency and integrity of the vascular morphology by designing a loss function that minimizes topological structure breaks and lumen volume changes. The experimental results show that this method can effectively capture the dynamic deformation characteristics of hepatic blood vessels and provide reliable motion modeling support for interventional surgical navigation under complex respiratory conditions.

[0137] See also Figure 3 The embodiment of the present application can also provide a device for constructing a vascular respiratory motion map, such as Figure 3 As shown, the device may include:

[0138] An image acquisition unit 301 is used to acquire 4D-CTA image data of a patient, wherein the 4D-CTA image data includes a blood vessel image;

[0139] A motion map output unit 302, used for inputting the 4D-CTA image data into the constructed respiratory motion model so that the respiratory motion model outputs a vascular 4D motion map;

[0140] The processing process of the respiratory motion model includes data preprocessing, topological structure constraints, adaptive deformation compensation constraints and overall structure preservation constraints;

[0141] The data preprocessing includes extracting the vascular structure and vascular centerline from the 4D-CTA image using a TotalSegmentator algorithm and a morphological algorithm, respectively, and calculating the lumen diameter at each centerline point;

[0142] The topological structure constraint includes establishing a directed graph of blood vessels based on the topological connection relationship within the blood vessel structure, and quantifying the overall displacement of the blood vessels and the posture of each branch by extracting structural control points and calculating branch bending energy;

[0143] The adaptive deformation compensation constraint includes using rotation metric and stretching metric in the spherical coordinate system of the manifold space to characterize the motion deformation of each segment of the branch to achieve quantification of the deformation of each branch, and adaptively assigning bending energy weights to each branch based on the vascular map and the SRVF constraint;

[0144] The overall structure preservation constraint includes jointly optimizing the skeleton integrity and lumen preservation during deformation using a loss minimization metric.

[0145] The embodiment of the present application may also provide a device for constructing a vascular respiratory motion map, the device comprising a processor and a memory:

[0146] The memory is used to store program code and transmit the program code to the processor;

[0147] The processor is used to execute the steps of the above-mentioned vascular respiratory motion atlas construction method according to the instructions in the program code.

[0148] like Figure 4 As shown, a vascular respiratory motion map construction device provided in an embodiment of the present application may include: a processor 10, a memory 11, a communication interface 12 and a communication bus 13. The processor 10, the memory 11, and the communication interface 12 all communicate with each other through the communication bus 13.

[0149] In the embodiment of the present application, the processor 10 may be a central processing unit (CPU), an application specific integrated circuit, a digital signal processor, a field programmable gate array or other programmable logic devices, etc.

[0150] The processor 10 may call a program stored in the memory 11. Specifically, the processor 10 may execute operations in an embodiment of the method for constructing a vascular respiratory motion map.

[0151] The memory 11 is used to store one or more programs, which may include program codes, and the program codes include computer operation instructions. In the embodiment of the present application, the memory 11 at least stores programs for implementing the following functions:

[0152] Acquiring 4D-CTA image data of the patient, wherein the 4D-CTA image data includes a blood vessel image;

[0153] Inputting the 4D-CTA image data into the constructed respiratory motion model so that the respiratory motion model outputs a 4D vascular motion atlas;

[0154] The processing process of the respiratory motion model includes data preprocessing, topological structure constraints, adaptive deformation compensation constraints and overall structure preservation constraints;

[0155] The data preprocessing includes extracting the vascular structure and vascular centerline from the 4D-CTA image using a TotalSegmentator algorithm and a morphological algorithm, respectively, and calculating the lumen diameter at each centerline point;

[0156] The topological structure constraint includes establishing a directed graph of blood vessels based on the topological connection relationship within the blood vessel structure, and quantifying the overall displacement of the blood vessels and the posture of each branch by extracting structural control points and calculating branch bending energy;

[0157] The adaptive deformation compensation constraint includes using rotation metric and stretching metric in the spherical coordinate system of the manifold space to characterize the motion deformation of each segment of the branch to achieve quantification of the deformation of each branch, and adaptively assigning bending energy weights to each branch based on the vascular map and the SRVF constraint;

[0158] The overall structure preservation constraint includes jointly optimizing the skeleton integrity and lumen preservation during deformation using a loss minimization metric.

[0159] In one possible implementation, the memory 11 may include a program storage area and a data storage area, wherein the program storage area can store an operating system and application programs required for at least one function (such as a file creation function, a data reading and writing function), etc.; the data storage area can store data created during use, such as initialization data, etc.

[0160] In addition, the memory 11 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device or other volatile solid-state storage device.

[0161] The communication interface 12 may be an interface of a communication module, and is used to connect to other devices or systems.

[0162] Of course, it should be noted that Figure 4The structure shown does not constitute a limitation on the vascular respiratory motion map construction device in the embodiment of the present application. In actual applications, the vascular respiratory motion map construction device may include Figure 4 More or fewer components than shown, or combinations of certain components.

[0163] The embodiment of the present application may also provide a computer-readable storage medium, wherein the computer-readable storage medium is used to store program code, and the program code is used to execute the steps of the above-mentioned vascular respiratory motion map construction method.

[0164] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0165] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present application or certain parts of the embodiments.

[0166] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can refer to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, wherein 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 may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without creative work.

[0167] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. A method for constructing a vascular respiratory motion atlas, characterized in that: include: Acquiring 4D-CTA image data of the patient, wherein the 4D-CTA image data includes a blood vessel image; Inputting the 4D-CTA image data into the constructed respiratory motion model so that the respiratory motion model outputs a 4D vascular motion atlas; The processing process of the respiratory motion model includes data preprocessing, topological structure constraints, adaptive deformation compensation constraints and overall structure preservation constraints; The data preprocessing includes extracting the vascular structure and vascular centerline from the 4D-CTA image using a TotalSegmentator algorithm and a morphological algorithm, respectively, and calculating the lumen diameter at each centerline point; The topological structure constraint includes establishing a directed graph of blood vessels based on the topological connection relationship within the blood vessel structure, and quantifying the overall displacement of the blood vessels and the posture of each branch by extracting structural control points and calculating branch bending energy; The adaptive deformation compensation constraint includes using rotation metric and stretching metric in the spherical coordinate system of the manifold space to characterize the motion deformation of each segment of the branch to achieve quantification of the deformation of each branch, and adaptively assigning bending energy weights to each branch based on the vascular map and the SRVF constraint; The overall structure preservation constraint includes jointly optimizing the skeleton integrity and lumen preservation during deformation using a loss minimization metric.

2. The method for constructing a vascular respiratory motion atlas according to claim 1, characterized in that: The structural control points include the end points, bifurcation points and branch points of the blood vessel branches, and distance vectors between the structural control points are established to characterize the direction and movement amplitude of the displacement vectors of each branch to approximate the movement of the dense blood vessel point set; The branch bending energy of the corresponding segment is quantified by calculating the curvature of the dense points on the centerline of the segment branch.

3. The method for constructing a vascular respiratory motion atlas according to claim 1, characterized in that: The topology constraint is expressed as follows: Where: C represents the curvature matrix, represents the displacement matrix of the center line point set, and La represents the Laplacian matrix.

4. The method for constructing a vascular respiratory motion atlas according to claim 1, characterized in that: The method of using rotation metric and stretching metric in the spherical coordinate system of the manifold space to characterize the motion deformation of each segment of the branch and realize the quantification of the deformation of each branch includes: The branch structure β is transformed from the spatial rectangular coordinate system (x, y, z) to the spherical coordinate system (r, θ, φ). The rotation is quantified by measuring the change in the direction of the tangent vector of the curve f on the branch β along the deformation path, and the stretching is quantified by the change in the length of the curve along the deformation path.

5. The method for constructing a vascular respiratory motion atlas according to claim 1, characterized in that: The adaptive deformation compensation constraint is expressed by the following formula: Where W represents the branch bending energy weight matrix, O represents the branch rotation metric matrix, and L represents the branch length metric matrix.

6. The method for constructing a vascular respiratory motion atlas according to claim 5, characterized in that: The branch rotation metric matrix includes all branch rotation metrics, and each branch rotation metric is expressed by the following formula: Where: a and b represent constant weights, represents the unit vector tangent to the curve f at point s, e φ(s) Then it represents the magnitude of the tangent vector at point s; The branch length metric matrix includes all branch length metrics, and the branch length metric is expressed by the following formula: Where: is the velocity vector, and The inner product 〈·〉 serves as a deterministic metric operator.

7. The method for constructing a vascular respiratory motion atlas according to claim 1, characterized in that: The overall structure preservation constraint is expressed by the following formula: Where: Ls(Δv) and Ls(Δr) represent the absolute values ​​of the differences in vascular diameter and volume at different respiratory phases, respectively. m is the bending energy weight corresponding to each point on the branch centerline, and num represents the number of branches in the current vascular structure.

8. A device for constructing a vascular respiratory motion map, characterized in that: The device is used to execute the method for constructing a vascular respiratory motion atlas according to any one of claims 1 to 7, comprising: An image acquisition unit, used to acquire 4D-CTA image data of a patient, wherein the 4D-CTA image data includes a blood vessel image; A motion atlas output unit, used for inputting the 4D-CTA image data into the constructed respiratory motion model, so that the respiratory motion model outputs a 4D motion atlas of blood vessels; The processing process of the respiratory motion model includes data preprocessing, topological structure constraints, adaptive deformation compensation constraints and overall structure preservation constraints; The data preprocessing includes extracting the vascular structure and vascular centerline from the 4D-CTA image using a TotalSegmentator algorithm and a morphological algorithm, respectively, and calculating the lumen diameter at each centerline point; The topological structure constraint includes establishing a directed graph of blood vessels based on the topological connection relationship within the blood vessel structure, and quantifying the overall displacement of the blood vessels and the posture of each branch by extracting structural control points and calculating branch bending energy; The adaptive deformation compensation constraint includes using rotation metric and stretching metric in the spherical coordinate system of the manifold space to characterize the motion deformation of each segment of the branch to achieve quantification of the deformation of each branch, and adaptively assigning bending energy weights to each branch based on the vascular map and the SRVF constraint; The overall structure preservation constraint includes jointly optimizing the skeleton integrity and lumen preservation during deformation using a loss minimization metric.

9. A device for constructing a vascular respiratory motion map, characterized in that: The device comprises a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the vascular respiratory motion map construction method described in any one of claims 1-7 according to the instructions in the program code.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium is used to store program codes, and the program codes are used to execute the method for constructing a vascular respiratory motion atlas according to any one of claims 1-7.

Images