A bladder eit three-dimensional imaging method fusing position shape constraint

Abstract

The application provides a bladder EIT three-dimensional imaging method fusing position shape constraints, comprising: acquiring a three-dimensional CT data set; acquiring position prior information of a bladder to be imaged and shape prior information of the bladder to be imaged according to the three-dimensional CT data set; constructing a position constraint term and a shape constraint term according to the position prior information and the shape prior information respectively; constructing a target function of a sparse representation framework according to the position constraint term and the shape constraint term; and solving the target function by using a Gauss-Newton iteration method to obtain a three-dimensional image of the bladder to be imaged. The application solves the problem in the prior art that a bladder EIT three-dimensional imaging result based on a single-layer EIT sensor only realizes bladder volume monitoring and cannot guarantee the accuracy of the bladder position and shape.

Description

Technical Field

[0001] This invention relates to the field of electrical impedance tomography (EIT) technology, and in particular to a three-dimensional imaging method for bladder EIT that incorporates position and shape constraints. Background Technology

[0002] Cancer is one of the leading causes of death in humans. Many tumors are concentrated in the lower abdomen, including colorectal cancer, bladder cancer, cervical cancer, and prostate cancer. Surgery is a first-line treatment, but the cure rate for single-therapy is not high. Radiation therapy is a primary treatment method, especially for patients with invasive cancers. Medical research indicates that minimizing organ positional variations during planning and treatment phases is crucial for accurate radiotherapy. The bladder is a mobile, hollow organ located in the lower abdomen. Its volume, shape, and position are highly variable due to factors such as urine inflow, rectal filling, and patient posture, posing challenges to precise radiotherapy. To ensure adequate radiation coverage of the lesion while minimizing its toxicity to healthy tissues, it is necessary to monitor the bladder's three-dimensional parameters, including volume, shape, and position.

[0003] Commonly used bladder monitoring methods in clinical practice include cone-beam CT, MRI, ultrasound, and infrared optical methods. CT and MRI, due to radiation damage and equipment cost limitations, cannot provide continuous monitoring, potentially causing missed opportunities for optimal radiotherapy. Therefore, various small-sized, low-cost, and low-harm detection methods have emerged. For example, invention patents CN201711401220, CN201880023352, CN202010540660, CN202111055300, and CN202310219424 all investigate non-invasive methods for detecting bladder volume using ultrasound technology; invention patents CN201310636585 and CN201880082747 investigate non-invasive methods for detecting bladder volume using infrared optical sensors. However, ultrasound imaging typically requires rotating or moving the ultrasound transducer to obtain information about different directions or positions of the target, leading to the development of mechanical scanning and phased array scanning structures. Regardless of the approach, it's difficult to balance cost, portability, and measurement accuracy. Therefore, most current portable ultrasound bladder detection instruments only offer volume measurement capabilities. The basic principle of infrared optical monitoring is that infrared light can penetrate the superficial layers of the body, and the intensity of the reflected signal is correlated with bladder volume. This method typically acquires a single signal, resulting in limited effective information and preventing imaging; similarly, it can only obtain bladder volume information.

[0004] Electrical Impedance Tomography (EIT) inherently possesses advantages such as safety and non-invasiveness, high real-time performance, three-dimensional imaging capability, simple structure, low cost, and suitability for wearable continuous measurement, making it highly suitable for continuous monitoring of comprehensive bladder parameters. In recent years, this method has received widespread attention from scholars. For example, invention patents CN201811514735, CN201911275429, CN201911292494, CN202111521075, and CN202210938276 all propose EIT-based methods for bladder volume monitoring. However, due to the inherent disadvantages of pathological and ill-posedness in EIT technology, its images have low physical resolution and are prone to target displacement and distortion; therefore, most research has only focused on bladder volume monitoring. Although the invention patent CN202210938276 achieves three-dimensional imaging of EIT, its imaging results cannot guarantee the accuracy of bladder position and shape, so its method only achieves bladder volume monitoring. Summary of the Invention

[0005] To overcome the shortcomings of the prior art, the purpose of this invention is to provide a bladder EIT three-dimensional imaging method that integrates position and shape constraints. This invention solves the problem that the prior art cannot guarantee the accuracy of bladder position and shape in the bladder three-dimensional imaging results.

[0006] To achieve the above objectives, the present invention provides the following solution:

[0007] A three-dimensional imaging method for bladder EIT that integrates position and shape constraints includes:

[0008] Obtain the patient's lower abdominal 3D CT dataset;

[0009] Based on the three-dimensional CT dataset, obtain the prior information on the location and shape of the bladder to be imaged;

[0010] Based on the prior position information and the prior shape information, position constraint terms and shape constraint terms are constructed respectively;

[0011] Construct the objective function of the sparse representation framework based on the position and shape constraints;

[0012] The objective function is solved using the Gauss-Newton iteration method to obtain a three-dimensional image of the bladder to be imaged.

[0013] Preferably, the shape prior information of the bladder to be imaged is obtained by a bladder shape contour compression representation algorithm based on spherical coordinate transformation.

[0014] Preferably, the calculation process for obtaining the prior shape information of the bladder to be imaged through the bladder shape contour compression representation algorithm based on spherical coordinate transformation is as follows:

[0015] The boundary point set of the bladder to be tested is obtained based on the 3D CT dataset, and then stacked layer by layer to form a point cloud set of the bladder's 3D boundary.

[0016] The coordinates of the bladder centroid are calculated based on the point cloud set of the bladder's three-dimensional boundary.

[0017] Based on the bladder centroid coordinates, the bladder contour point cloud is transformed into spherical coordinates to obtain the spherical coordinates of the bladder boundary points;

[0018] Based on the spherical coordinates, the original polar distance distribution is fitted and downsampled at fixed sampling angles with the same interval angles to obtain the first shape representation information in the form of a spherical coordinate polar distance vector.

[0019] The shape prior information of the bladder to be imaged is obtained based on the first shape characterization information.

[0020] Preferably, the method for constructing the position constraint term is as follows:

[0021] The weighted center of the basis points is calculated by combining the values ​​of the basis function coefficient vector and the positions of the basis function center points in the sparse representation framework.

[0022] Use the weighted center as the centroid of the reconstructed bladder region in the current iteration step, and construct the L2 norm by subtracting it from the prior bladder position.

[0023] The position constraint term is constructed based on the L2 norm.

[0024] Preferably, the expression for the position constraint term is:

[0025]

[0026] Where R3 represents the position constraint term, P c P represents the target position in the current iteration step. c * Indicates the prior position of the shape.

[0027] Preferably, the method for constructing shape constraints is as follows:

[0028] First, construct the function B(u), and establish a relationship between the basis function coefficient vector u under the sparse representation framework and the target shape vector r. sn Mapping;

[0029] The output value of the above function B(u), the difference between the shape vector of the reconstructed bladder region in the current iteration step and the prior shape vector of the bladder, and the L2 norm are used to construct the shape constraint term.

[0030] Preferably, the expression for the shape constraint term is:

[0031]

[0032] Where R4 is the shape constraint term, B(u) is the function, and r * Let be the prior bladder shape vector.

[0033] Preferably, the calculation process of the function B(u) is as follows:

[0034] Segment the target region in the current iteration step;

[0035] Extract the coordinates of the boundary points of the target region to form a boundary point cloud set;

[0036] Based on the boundary point cloud set, the bladder shape contour compression representation algorithm based on spherical coordinate transformation calculates the shape vector of the bladder region in the current iteration step.

[0037] Preferably, the objective function for constructing the sparse representation framework based on the positional and shape constraints is expressed as follows:

[0038]

[0039] The objective function is to construct a sparse representation framework based on position and shape constraints; U represents the voltage measurement of EIT, λ i Indicates the corresponding R i The regularization parameter.

[0040] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0041] This invention provides a three-dimensional imaging method for bladder EIT that integrates position and shape constraints, comprising:

[0042] A 3D CT dataset is acquired; prior information about the location and shape of the bladder to be imaged is obtained from the 3D CT dataset; positional constraints and shape constraints are constructed based on the positional and shape constraints, respectively; an objective function of a sparse representation framework is constructed based on the positional and shape constraints; the objective function is solved using the Gauss-Newton iteration method to obtain a 3D image of the bladder to be imaged. This invention considers positional and shape constraints when reconstructing bladder images in 3D, improving the accuracy of 3D imaging. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 A flowchart of a bladder EIT three-dimensional imaging method with fused position and shape constraints provided in an embodiment of the present invention;

[0045] Figure 2 This is a schematic diagram of the bladder shape contour compression representation process based on spherical coordinates provided in an embodiment of the present invention, wherein (a) is a schematic diagram of the original bladder boundary point cloud; (b) is a schematic diagram of the distribution of point coordinates in (a) after transformation to the spherical coordinate domain; (c) is a schematic diagram of the representation of the bladder shape contour after sampling - polar distance; (d) is a schematic diagram of the intersection point of the sampling angle ray and the unit circle when the sampling interval is 15°; and (e) is a schematic diagram of the bladder boundary point cloud after sampling.

[0046] Figure 3 A schematic diagram of the finite element model used for simulation verification in the embodiments of the present invention;

[0047] Figure 4 A schematic diagram showing the true distribution and reconstruction results of four algorithms for simulation verification provided in this embodiment of the invention;

[0048] Figure 5 This is a schematic diagram of the evaluation index corresponding to the reconstruction result in the simulation verification provided in the embodiments of the present invention. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] The purpose of this invention is to provide a bladder EIT three-dimensional imaging method that integrates position and shape constraints. This invention solves the problem in the prior art that the accuracy of bladder position and shape cannot be guaranteed in the three-dimensional imaging results of the bladder.

[0051] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0052] like Figure 1As shown, this invention provides a bladder EIT three-dimensional imaging method that integrates position and shape constraints, comprising:

[0053] Step 100: Obtain the patient's lower abdominal 3D CT dataset;

[0054] Step 200: Obtain the prior information on the location and shape of the bladder to be imaged based on the three-dimensional CT dataset;

[0055] Step 300: Construct position constraint terms and shape constraint terms based on the position prior information and the shape prior information, respectively;

[0056] Step 400: Construct the objective function of the sparse representation framework based on the position constraint and shape constraint terms;

[0057] Step 500: Solve the objective function using the Gauss-Newton iteration method to obtain a three-dimensional image of the bladder to be imaged.

[0058] The above method can be summarized into two steps:

[0059] First: Acquisition of the relative position of the bladder-electrode plane and the approximate shape of the bladder. The electrode plane refers to the plane where the electrode array is located, and the relative position mainly refers to the height difference between the electrode plane and the center of the bladder. Bladder shape information was extracted from three-dimensional CT images of the lower abdomen of multiple patients, and the shape information was compressed and represented in an appropriate form to facilitate the construction of constraint terms and their integration into the image reconstruction process.

[0060] Second: The implementation of position and shape constraints based on the sparse representation framework, and the image reconstruction process. Sparse representation is a class of processing methods applicable to image reconstruction problems. Its core idea is to map the parameters to be reconstructed to another sparse domain, using fewer unknowns to represent the original parameters to be reconstructed, thereby reducing the number of variables to be reconstructed and reducing the difficulty of reconstruction.

[0061] Furthermore, the EIT sensor is configured as a semi-circular, single-layer electrode array. This minimizes the sensor's size and structural complexity, thereby reducing the impact of the electrodes on radiation during radiotherapy.

[0062] Unlike industrial applications of EIT such as multiphase flow imaging, the imaging target in bladder monitoring is the bladder, whose position and shape are relatively fixed. Therefore, approximate information about the bladder's position and shape can be obtained in advance as prior information to assist in iterative image reconstruction.

[0063] The target application of this invention is the continuous monitoring of comprehensive three-dimensional bladder information before radiotherapy. During the treatment cycle of cancer patients, multiple three-dimensional CT scans are typically required. The three-dimensional CT images of the patient's lower abdomen can be compiled into a dataset, from which general information about the bladder's location and shape can be extracted and learned.

[0064] The bladder's anatomical location is within the pelvis and supported by it. As the bladder grows, the position of its base remains relatively fixed, while its center gradually shifts upwards. The approximate location of the bladder can be obtained from the position of the pelvis and the current bladder volume in CT images.

[0065] Specifically, the shape information of the bladder is more complex and high-dimensional, and is strictly coupled with the bladder volume and position in CT images. Therefore, appropriate preprocessing methods are needed to decouple the information and compress and represent the individual shape information. Bladder region boundary points are extracted from layered 3D CT images and stacked layer by layer to form a point cloud representing the bladder's three-dimensional boundary. Considering that the bladder is mostly a closed membranous organ similar to a balloon, its shape changes during urine volume increase, resembling a gradually expanding balloon under external pressure, which is a typical three-dimensional convex shape.

[0066] Furthermore, the shape prior information of the bladder to be imaged is obtained through a bladder shape contour compression representation algorithm based on spherical coordinate transformation.

[0067] Furthermore, the calculation process for obtaining the prior shape information of the bladder to be imaged through the bladder shape contour compression representation algorithm based on spherical coordinate transformation is as follows:

[0068] The boundary point set of the bladder to be tested is obtained based on the 3D CT dataset, and then stacked layer by layer to form a point cloud set of the bladder's 3D boundary.

[0069] The coordinates of the bladder centroid are calculated based on the point cloud set of the bladder's three-dimensional boundary.

[0070] Based on the bladder centroid coordinates, the bladder contour point cloud is transformed into spherical coordinates to obtain the spherical coordinates of the bladder boundary points;

[0071] Based on the spherical coordinates, the original polar distance distribution is fitted and downsampled at fixed sampling angles with the same interval angles to obtain the first shape representation information in the form of a spherical coordinate polar distance vector; based on the first shape representation information, the shape prior information of the bladder to be imaged is obtained.

[0072] Specifically, a normalized second shape representation is obtained based on the first shape representation information. This second shape representation information is the shape prior information of the bladder to be imaged. A Cartesian coordinate system transformation is performed on the second shape representation information to obtain a sampled bladder contour point cloud. This contour point cloud can be used to confirm and verify that the main shape information is basically not lost after compression and representation using this method. The final shape prior information is a vector. Converting this shape prior vector back to the Cartesian coordinate system is to verify whether the main shape information is damaged or lost.

[0073] like Figure 2 As shown, (a) is a schematic diagram of the original bladder boundary point cloud; (b) is a schematic diagram of the distribution of point coordinates in (a) after transformation to the spherical coordinate domain; (c) is a schematic diagram of the representation of the bladder shape contour after sampling - polar distance; (d) is a schematic diagram of the intersection point of the sampling angle ray and the unit circle when the sampling interval is 15°; (e) is a schematic diagram of the bladder boundary point cloud after sampling; the bladder shape contour compression representation algorithm based on spherical coordinate transformation based on the above method is specifically expressed as follows:

[0074] 1) Computing the bladder centroid coordinates C from the bladder's three-dimensional boundary points. g =[x cg ,y cg ,z cg Using this coordinate as the centroid, perform a spherical coordinate transformation on the bladder contour point cloud:

[0075] [α ps ,β ps r ps ] = F cart2sph (x ps y ps , z ps )

[0076] Where α, β, and r represent the azimuth, elevation, and radial distance, respectively.

[0077] 2) Set a fixed sampling angle θ with the same interval angle. s =[α s ,β s (In this example, the interval is 15°), fit and downsample the original polar distance distribution:

[0078] r s =F sample (α ps ,β ps r ps α s ,β s )

[0079] This yields the shape representation information r in spherical coordinate polar moment vector form. s ;

[0080] 3) Normalize the sampling polar distance vector:

[0081] r sn =r s / sum(r s )·size(r s )

[0082] This decouples information about the shape and size of the bladder.

[0083] 4) The compressed representation information can be converted back to a rectangular coordinate system to obtain the sampled bladder contour point cloud.

[0084] This achieves complete decoupling of bladder size, shape, and location information, and fully refines and compresses the shape information, laying the foundation for its subsequent introduction as prior information into the iterative reconstruction process.

[0085] The target bladder to be imaged has typical sparse characteristics, so using a sparse representation framework can significantly reduce the difficulty of image reconstruction.

[0086] The sparse characterization framework assumes that both the background Ω\D and the target D in the measurement domain Ω have uniformly distributed conductivity:

[0087]

[0088] In the formula, x represents the coordinates of a three-dimensional point.

[0089] Then, the target and background are separated by an eigenfunction composed of basis functions, and the conductivity distribution can be characterized as follows:

[0090]

[0091] In the formula, g i (x) is the set of basis functions; specifically, Gaussian radial basis functions (GRBFs) are used here. u = [u1, ..., u2] N ] is the basis coefficient vector, H ε It is a smoothing sign function, and c is a threshold parameter that can be freely set within a reasonable range.

[0092] The objective function of the sparse representation framework is:

[0093]

[0094] In the formula, R1 and R2 are the regularization terms added to the original framework, and λ1 and λ2 are their corresponding regularization parameters.

[0095] Furthermore, the method for constructing the position constraint terms is as follows:

[0096] The method for constructing position constraint terms is as follows:

[0097] The weighted center of the basis points is calculated by combining the values ​​of the basis function coefficient vector and the positions of the basis function center points in the sparse representation framework.

[0098] Use the weighted center as the centroid of the reconstructed bladder region in the current iteration step, and construct the L2 norm by subtracting it from the prior bladder position.

[0099] The position constraint term is constructed based on the L2 norm.

[0100] Specifically, the physical meaning of the sparse representation process can be interpreted as downsampling the original complete image according to the distribution of the basis function center points, and then blurring the image according to the basis function type. This invention uses a Gaussian radial basis function, resulting in Gaussian blur, which does not change the centroid of the target. Therefore, the nonlinear stage of the sparse representation process can be omitted, and the centroid of the target region can be replaced by the centroid of the Gaussian basis points. The target center can then be represented as: P c =P b T ·u / N;

[0101] This allows us to directly use the target function's parameter u to represent the current target position, thereby constructing the position constraint term: the expression for the position constraint term is:

[0102]

[0103] Where R3 represents the position constraint term, P c P represents the current target position. c * Indicates the prior position of the shape.

[0104] Furthermore, the method for constructing shape constraint terms is as follows:

[0105] First, construct the function B(u), and establish a relationship between the basis function coefficient vector u under the sparse representation framework and the target shape vector r. sn Mapping;

[0106] The output value of the above function B(u), the difference between the shape vector of the reconstructed bladder region in the current iteration step and the prior shape vector of the bladder, and the L2 norm are used to construct the shape constraint term.

[0107] Specifically, using the shape contour compression representation algorithm proposed in this invention, the shape of the bladder is represented as a normalized sampling polar distance vector r with a fixed dimension. sn If the parameters to be reconstructed in the objective function can be used to represent r... snThen, a regularization term in the form of L2 can be easily constructed. Therefore, a function B(u) is proposed to establish a regularization term from the sparse vector u of the basis functions (which is also one of the parameters to be reconstructed in the objective function) to the shape vector r. sn The mapping process is as follows:

[0108] 1) Based on the formula, the target region with conductivity σ1 is segmented;

[0109] 2) Extract the coordinates of the boundary points of the target area and form a point cloud;

[0110] 3) Calculate r using the shape contour compression representation algorithm described above. sn .

[0111] This enables the construction of shape constraint regularization terms:

[0112]

[0113] Where R4 is the shape constraint term, B(u) is the function, and r * Let be the prior bladder shape vector.

[0114] Furthermore, the calculation process of the function B(u) is as follows:

[0115] Segment the target region in the current iteration step;

[0116] Extract the coordinates of the boundary points of the target region to form a boundary point cloud set;

[0117] Based on the boundary point cloud set, the bladder shape contour compression representation algorithm based on spherical coordinate transformation calculates the shape vector of the bladder region in the current iteration step.

[0118] Furthermore, the expression for the objective function of constructing the sparse representation framework based on the positional and shape constraints is as follows:

[0119]

[0120] The objective function is to construct a sparse representation framework based on position and shape constraints; U represents the voltage measurement of EIT, λ i Indicates the corresponding R i The regularization parameter.

[0121] The objective function is solved using the Gauss-Newton iterative method, which requires calculating the Jacobian and Hessian matrices of the objective function. However, since the function B(u) is calculated forwards rather than analytically, its Jacobian matrix cannot be calculated using conventional methods. Therefore, a perturbation method is introduced to separately calculate the Jacobian matrix J of the B(u) part. BuThat is, if a small perturbation Δu is applied to each parameter contained in u, and the change in the output ΔB is recorded, then J Bu =ΔB / Δu. Since the objective function is in polynomial summation form, the Jacobian and Hessian matrices can be calculated separately for the fidelity term and the four regularization terms. The final iterative formula is:

[0122]

[0123] In the formula, α represents the iteration step size. Δ represents the Jacobian matrix and the Hessian matrix, respectively:

[0124]

[0125]

[0126]

[0127]

[0128]

[0129]

[0130] H R1 =I u °

[0131] H R2 =I σ °

[0132] H R3 =I u °J Bu J Bu T I u ° T

[0133] H R4 =I u °(P b P b T )I u ° T / N

[0134] In the formula, J is the Jacobian matrix of the standard EIT positive problem. A 0-1 diagonal matrix representing the positions of control parameters.

[0135] This embodiment also demonstrates the effectiveness of the method through simulation testing:

[0136] Use COMSOL to build a finite element model and perform simulation calculations. For example... Figure 3 As shown, the measurement domain uses a stretched model of a cross-section 8 cm below the navel; the target bladder uses two shapes: an ellipsoidal bladder commonly used in scientific research and a bladder with the actual shape inside the human body. To simulate the bladder volume increase during urination, simulations were performed with four volumes (220 mL, 310 mL, 400 mL, and 490 mL) and eight different distributions. The EIT measurements from the eight simulations were recorded, and three-dimensional image reconstruction was performed using the method proposed in this patent (hereinafter referred to as the PSC algorithm). Three existing algorithms were also used as comparisons. The three comparison algorithms include:

[0137] SR: Family of algorithms within a sparse representation framework;

[0138] SBL: Sparse Bayesian Learning Algorithm, another state-of-the-art (SOTA) algorithm outside of the sparse representation framework;

[0139] L1: L1 regularization algorithm, a traditional algorithm.

[0140] To quantitatively compare different reconstruction results, four evaluation metrics were introduced:

[0141] a. Structural Similarity Index (SSIM): A comprehensive evaluation index of reconstruction accuracy; the higher the accuracy, the higher the value.

[0142] b. Image Error: Another comprehensive evaluation index of reconstruction accuracy; the higher the accuracy, the lower the value.

[0143] c. Target Volume: The volume of the reconstructed target (bladder) region;

[0144] d. Position Error: An evaluation index of position accuracy; the higher the accuracy, the lower the value.

[0145] e. Shape Error: An evaluation metric for shape accuracy; the higher the accuracy, the lower the value.

[0146] The ground truth distribution (GT) of the ellipsoidal bladder model (Ellipsoid Model, -e) and the real human bladder model (UNIBladder Model, -u) and the reconstruction results of the four algorithms are shown below. Figure 4 The evaluation indicators corresponding to the reconstruction results are as follows: Figure 5As shown, the prefix indicates the algorithm name, and the suffix indicates the model type: -e indicates the ellipsoidal model (Ellipsold). The three parameters—target volume, position error, and shape error—are only used to compare the PSC and SR algorithms. The results show that the target reconstructed by the proposed PSC algorithm can better cover the real bladder distribution area of ​​the semi-transparent green color, the target volume estimation error is within an acceptable range, and it has lower position and shape errors. Therefore, the feasibility of this method can be proven.

[0147] The beneficial effects of this invention are as follows:

[0148] This invention provides a 3D imaging method for the bladder using a single-layer semi-circular electrode array (EIT) sensor, achieving higher accuracy in position and shape reconstruction. This method offers advantages such as simple structure, low cost, safety, and real-time monitoring. The employed EIT sensor structure minimizes the impact of electrodes on radiation during radiotherapy. Position and shape constraints improve the accuracy of the reconstructed target's position and shape, enabling EIT-based monitoring of comprehensive bladder parameters.

[0149] Compared with existing technologies such as MRI, CT, ultrasound, and infrared optical methods, this invention has a simple structure, low cost, is non-invasive, and allows for continuous monitoring. Compared with existing EIT technology, this invention achieves higher precision three-dimensional imaging of the bladder, effectively monitoring the bladder's position and shape while ensuring the accuracy of bladder volume monitoring.

[0150] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0151] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A three-dimensional imaging method for bladder EIT that integrates position and shape constraints, characterized in that, include: Obtain the patient's lower abdominal 3D CT dataset; Based on the three-dimensional CT dataset, obtain the prior information on the location and shape of the bladder to be imaged; Based on the prior position information and the prior shape information, position constraint terms and shape constraint terms are constructed respectively; Construct the objective function of the sparse representation framework based on the position and shape constraints; The objective function is solved using the Gauss-Newton iteration method to obtain a three-dimensional image of the bladder to be imaged; The method for constructing position constraint terms is as follows: The weighted center of the basis points is calculated by combining the values ​​of the basis function coefficient vector and the positions of the basis function center points in the sparse representation framework. Use the weighted center as the centroid of the reconstructed bladder region in the current iteration step, and construct the L2 norm by subtracting it from the prior bladder position. Construct position constraint terms based on the aforementioned L2 norm; The expression for the position constraint term is: ； in, R 3 Indicates position constraint terms. P c Indicates the target position in the current iteration step. Indicates the prior position of the shape; The method for constructing shape constraints is as follows: First, construct the function. Establish the basis function coefficient vector under the sparse representation framework. To target shape vector Mapping; The above function The output value, and the difference between the shape vector of the reconstructed bladder region in the current iteration step and the prior shape vector of the bladder, are used to construct the L2 norm and complete the construction of the shape constraint term; The expression for the shape constraint term is: ； in, R 4 For shape constraints, B(u) For functional purposes, The prior bladder shape vector; The function The calculation process is as follows: Segment the target region in the current iteration step; Extract the coordinates of the boundary points of the target region to form a boundary point cloud set; Based on the boundary point cloud set, the bladder shape contour compression representation algorithm based on spherical coordinate transformation calculates the shape vector of the bladder region in the current iteration step; The objective function for constructing the sparse representation framework based on the position and shape constraints is expressed as follows: ； The objective function for constructing a sparse representation framework based on position and shape constraints; U This indicates the voltage measurement value of EIT. Indicates correspondence The regularization parameter, Electrical conductivity; R 1 and R 2 These are the first and second regularization terms added to the original framework, respectively.

2. The bladder EIT three-dimensional imaging method with fused position and shape constraints according to claim 1, characterized in that, The shape prior information of the bladder to be imaged is obtained by a bladder shape contour compression representation algorithm based on spherical coordinate transformation.

3. The bladder EIT three-dimensional imaging method with fused position and shape constraints according to claim 2, characterized in that, The calculation process for obtaining the prior shape information of the bladder to be imaged using the bladder shape contour compression representation algorithm based on spherical coordinate transformation is as follows: The boundary point set of the bladder to be tested is obtained based on the 3D CT dataset, and then stacked layer by layer to form a point cloud set of the bladder's 3D boundary. The coordinates of the bladder centroid are calculated based on the point cloud set of the bladder's three-dimensional boundary. Based on the bladder centroid coordinates, the bladder contour point cloud is transformed into spherical coordinates to obtain the spherical coordinates of the bladder boundary points; Based on the spherical coordinates, the original polar distance distribution is fitted and downsampled at fixed sampling angles with the same interval angles to obtain the first shape representation information in the form of a spherical coordinate polar distance vector. The shape prior information of the bladder to be imaged is obtained based on the first shape characterization information.

Images