Method and system for modeling and signal attenuation compensation of thoracic conductor of magnetocardiogram imaging

Abstract

The application relates to the technical field of biomagnetic signal detection and imaging, and discloses a thoracic conductor modeling and signal attenuation compensation method and system for magnetocardiogram imaging. The method comprises five steps of multi-modal data synchronous acquisition and registration, personalized thoracic conductor model construction, individualized magnetic field attenuation transfer matrix construction, signal inverse compensation and general high-precision source imaging. The application performs front-end pre-compensation on original magnetocardiogram signals by constructing a high-precision personalized thoracic conductor model, generates a quasi-ideal signal graph that has been corrected for individual conduction attenuation, so that any subsequent standard source imaging algorithm works on the basis of fair and true signals, and the accuracy and consistency of final imaging are improved from the data source.

Description

Technical Field

[0001] This application relates to the field of biomagnetic signal detection and imaging technology, specifically to a method and system for thoracic conductor modeling and signal attenuation compensation in magnetocardiography imaging. Background Technology

[0002] Magnetocardiography (MCG) is a non-invasive technique for detecting the weak magnetic field generated by the electrical activity of the heart. It has potential value in the early diagnosis and localization of cardiac diseases such as arrhythmias and myocardial ischemia. The accuracy of MCG source imaging heavily depends on the accuracy of the electromagnetic forward model from the cardiac source to the surface sensor. The attenuation and distortion of the magnetic field by non-uniform conductors within the thoracic cavity (such as fat, muscle, bone, and lung tissue) are the main factors affecting the accuracy of the forward model.

[0003] Currently, constructing thoracic geometry models using individual medical imaging data (such as MRI and CT scans) to improve forward models has become a mainstream research direction for enhancing the accuracy of MCG imaging. However, existing technical approaches suffer from the following common limitations and unresolved core issues:

[0004] 1. The disconnect between "static" models and "dynamic" compensation: Existing methods typically use individualized geometric models directly in subsequent source imaging algorithms as a more accurate forward model input for solving the inverse problem. This essentially treats individual differences as a fixed, passive background parameter provided to the inverse problem algorithm (e.g., dipole fitting, beamforming). The inverse problem algorithm itself still needs to handle the coupling effect of "source localization" and "model error" simultaneously in a complex solution space, resulting in a heavy algorithmic burden and incomplete compensation.

[0005] 2. Lack of "front-end correction" for the signal: The current technical approach lacks a crucial intermediate step—physical compensation or correction of the original acquired magnetic signal using a personalized conduction model before source imaging. Individual attenuation differences in the original signal are not eliminated, resulting in inherent distortions in the data input to the inverse problem solver that are related to individual anatomical structures.

[0006] 3. Lack of a systematic workflow: There is a lack of a standardized system method and control logic that seamlessly integrates and efficiently operates the three stages of "automatic construction from image to computable electromagnetic model", "model-based signal physical compensation", and "standard source imaging". Existing research is mostly offline and discrete processing workflows, which are difficult to apply to real-time or near-real-time clinical scenarios.

[0007] Existing technologies have failed to optimally address the signal attenuation problem caused by individual thoracic conductors, resulting in a bottleneck in improving source imaging accuracy. There is an urgent need for a new technological paradigm that moves the application node of personalized models from within the "inverse problem solver" to the "raw signal preprocessing stage," enabling proactive and direct compensation for signal distortion. Summary of the Invention

[0008] To address the aforementioned issues, this application provides a method and system for thoracic conductor modeling and signal attenuation compensation in magnetocardiography (MCG) imaging. It utilizes a personalized model to perform "pre-compensation" processing on the original MCG signal, generating a "quasi-ideal" signal image with corrected individual conduction attenuation. This allows any subsequent standard source imaging algorithm to operate on a fairer and more realistic signal basis, improving the accuracy and consistency of the final image from the data source. Simulation experiments verify that the method provided in this application can significantly reduce the dipole localization error from approximately 12 mm under traditional methods to approximately 4 mm, and improve the correlation between the reconstructed result and the real source from 0.78 to over 0.95.

[0009] The technical solution adopted in this application is as follows:

[0010] In a first aspect, this application provides a method for thoracic conductor modeling and signal attenuation compensation in magnetocardiography imaging, including:

[0011] Step S10, Multimodal data synchronous acquisition and registration: Simultaneously or simultaneously acquire the three-dimensional thoracic structure image data and multi-channel raw magnetic cardiomyocyte signals of the target individual, and perform fusion registration of image space, magnetic cardiomyocyte sensor array space and individual body surface space;

[0012] Step S20, Personalized Thoracic Conductor Model Construction: The registered 3D thoracic structure image data is automatically segmented and classified into tissues, and complex conductivity parameters related to magnetocardiogram frequency are assigned to various tissues. A numerical calculation engine is used to generate a thoracic conductor model exclusive to the target individual.

[0013] Step S30, Construction of the transfer matrix for individualized magnetic field attenuation: Based on the thoracic conductor model, the heart region is set as the source space and the position of the cardiac magnetosensor array is set as the measurement space. Through electromagnetic field forward problem simulation, a transfer matrix characterizing individual-specific magnetic field attenuation is generated.

[0014] Step S40, inverse signal compensation: a compensation operator is constructed using the transfer matrix to transform the original multi-channel magnetocardiogram signal to obtain the attenuation-corrected magnetocardiogram signal vector;

[0015] Step S50, General High-Precision Source Imaging: Input the cardiac magnetic signal vector into the standard cardiac magnetic source imaging algorithm to solve the inverse problem and output the spatial distribution result of cardiac power source.

[0016] Secondly, this application also provides a system for thoracic conductor modeling and signal attenuation compensation in magnetocardiography (MCG) imaging, which implements the aforementioned method for thoracic conductor modeling and signal attenuation compensation in MCG imaging, including:

[0017] The synchronous acquisition and control unit coordinates and controls the data acquisition process of medical imaging equipment and magnetocardiography equipment and integrates an optical tracking system to achieve synchronous acquisition and registration of multimodal data.

[0018] The automated modeling and calculation engine includes: an image segmentation and tissue classification module, an electromagnetic parameter allocation module, a numerical model generator, and a forward simulation calculator for transfer matrices, used to realize the construction of personalized thoracic conductor models and the construction of transfer matrices for individualized magnetic field attenuation.

[0019] An online signal compensation processor includes an algorithm module and high-speed digital signal processing hardware for implementing inverse signal compensation.

[0020] A standardized source imaging and display terminal is used to achieve universal high-precision source imaging and fusion visualization on a target individual's three-dimensional cardiac model;

[0021] The system workflow manager is used to automatically manage data flow and control flow according to a fixed logical sequence.

[0022] Thirdly, this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for modeling the thoracic conductor and compensating for signal attenuation in magnetocardiography imaging.

[0023] Fourthly, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described method for modeling the thoracic conductor and compensating for signal attenuation in magnetocardiography imaging.

[0024] The above-mentioned technical solution adopted in this application can achieve the following beneficial effects:

[0025] 1. Paradigm Innovation, Leap in Accuracy: The application of personalized models shifts from "optimizing the inverse problem background" to "correcting the input signal itself." This is equivalent to "deconvolution" the data before it enters the ill-conditioned inverse problem, reducing the uncertainty of the solution at its source. Experiments have shown that this can significantly reduce positioning errors and achieve a step-by-step improvement in accuracy.

[0026] 2. Decoupling Complex Problems: The two coupled challenges of "handling individual conduction effects" and "solving cardiac power sources" are decoupled. First, the compensation step specifically addresses the conduction attenuation problem, and then a mature, general-purpose source imaging algorithm focuses on power source localization. This simplifies the problem and improves the robustness and efficiency of the overall solution.

[0027] 3. Enhanced algorithm universality and comparability: The compensated magnetocardiogram signal vector reduces interference caused by differences in thoracic anatomy among different subjects, making the same source imaging algorithm perform more consistently across different individuals and resulting in more comparable results. It also reduces the requirements for the source imaging algorithm's robustness against model errors.

[0028] 4. Achieving a closed loop for clinical application: The complete and standardized system process defined in this application, from data acquisition to final imaging, encapsulates cutting-edge personalized modeling and compensation technologies into automated and repeatable clinical tools, greatly promoting the transition of high-precision MCG from laboratory research to routine clinical application. Attached Figure Description

[0029] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0030] Figure 1 A flowchart illustrating a method for thoracic conductor modeling and signal attenuation compensation in magnetocardiography according to an embodiment of this application is shown.

[0031] Figure 2 A schematic diagram illustrating the principle of constructing a personalized thoracic conductor model according to an embodiment of this application is shown;

[0032] Figure 3 A schematic diagram illustrating the principle of constructing a transfer matrix for individualized magnetic field attenuation according to an embodiment of this application is shown.

[0033] Figure 4 A schematic diagram illustrating the principle of signal inverse compensation according to an embodiment of this application is shown;

[0034] Figure 5 A schematic diagram of a thoracic conductor modeling and signal attenuation compensation system for magnetocardiography imaging according to an embodiment of this application is shown.

[0035] Figure 6 A block diagram of an architecture for a thoracic conductor modeling and signal attenuation compensation system for magnetocardiography imaging according to an embodiment of this application is shown.

[0036] Figure 7 A schematic diagram of the structure of an electronic device according to an embodiment of this application is shown. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0038] Figure 1 A schematic flowchart of a method for thoracic conductor modeling and signal attenuation compensation in magnetocardiography according to an embodiment of this application is shown. (Refer to...) Figure 1 As shown, this embodiment includes steps S10 to S50:

[0039] Step S10, Multimodal Data Synchronous Acquisition and Registration: Acquire the three-dimensional thoracic structure image data and multi-channel raw magnetic cardiomyocyte signals of the target individual simultaneously or in the same session, and perform fusion registration of the image space, magnetic cardiomyocyte sensor array space and individual body surface space.

[0040] This embodiment first acquires three-dimensional thoracic structural image data (from MRI, CT, or 3D structured light scan) and multi-channel raw magnetocardiogram signals of the target individual simultaneously or in the same session. A coordinate positioning system is then used to achieve precise fusion and registration of the image space, the magnetocardiogram sensor array space, and the individual's body surface space.

[0041] In some optional implementations, step S10 includes: acquiring three-dimensional thoracic structural image data of the target individual using MRI, CT, or 3D structured light scanning; acquiring multi-channel magnetocardiogram (MCC) signals of the target individual at rest for several heartbeat cycles using a multi-channel OPM MCC system, and performing bandpass filtering on the multi-channel MCC signals to obtain the target individual's original multi-channel MCC signals; wherein the sampling frequency is not less than 1 kHz; attaching positioning markers to bony landmarks and corner points of the MCC sensor array on the target individual, recording the coordinates of the positioning markers using an optical tracking system, calculating the spatial rigid body transformation matrix using a point set registration algorithm, and transforming the image spatial coordinate system to the MCC sensor array spatial coordinate system; wherein the registration error is less than 2 mm.

[0042] Thoracic 3D structural image data acquisition:

[0043] The subject lay supine, and a 3T magnetic resonance imaging (MRI) scanner was used to perform a T1-weighted three-dimensional gradient echo sequence scan, covering the entire thoracic region. The acquired DICOM format data contained high-contrast soft tissue anatomical information, making it easy to distinguish between fat, muscle, and internal organs. Scan parameters: TR / TE = 5.2 ms / 2.5 ms; voxel resolution: 1.0 mm × 1.0 mm × 1.0 mm.

[0044] Multi-channel raw magnetocardiogram signal acquisition:

[0045] In the same experimental session, subjects were moved to a multi-channel (64-channel) OPM (atomic magnetometer) magnetocardiography system in a shielded room, ensuring their position was as consistent as possible with that during the MRI scan. Magnetocardiographic signals were acquired at rest for several heartbeat cycles (typically no less than 100 cycles) at a sampling rate of at least 1 kHz, and bandpass filtered (e.g., 0.1 Hz to 300 Hz) to remove baseline drift and high-frequency noise, yielding the raw multi-channel magnetocardiographic signals. The signals were stored in femtotes (fT).

[0046] Spatial registration:

[0047] MRI-visible but non-magnetically permeable positioning markers were affixed to bony landmarks such as the suprasternal notch, the midpoints of the left and right clavicles, and the spinous process of the 10th thoracic vertebra, as well as to the corners of the MCG sensor array. Using an optical tracking system (such as Polaris or NDI), the three-dimensional coordinates of these positioning markers in their respective device coordinate systems were recorded during MRI scans and MCG measurements. An optimal spatial rigid body transformation matrix (including rotation and translation) was calculated using a point set registration algorithm (ICP iterative nearest point algorithm) to accurately transform the MRI image spatial coordinate system to the MCG sensor array spatial coordinate system. The registration error was required to be less than 2 mm.

[0048] Step S20, Personalized Thoracic Conductor Model Construction: The registered 3D thoracic structure image data is automatically segmented and classified, and complex conductivity parameters related to magnetocardiogram frequency are assigned to various tissues. A numerical calculation engine is used to generate a thoracic conductor model exclusive to the target individual.

[0049] This embodiment then performs automated tissue segmentation and classification on the registered 3D thoracic structure image data, distinguishing tissues with different electromagnetic properties such as skin, subcutaneous fat, muscle, ribs / sternum, lung cavity, and heart. Complex conductivity parameters related to magnetocardiogram frequencies are assigned to each type of tissue, and a numerical calculation engine (such as the FEM finite element method) is used to generate a target individual-specific, non-uniform thoracic conductor model suitable for electromagnetic field simulation.

[0050] Figure 2 A schematic diagram illustrating the principle of constructing a personalized thoracic conductor model according to an embodiment of this application is shown. (Refer to...) Figure 2 As shown, in some optional implementations, step S20 includes: using a deep learning segmentation model pre-trained on a public dataset to segment and classify the registered thoracic 3D structural image data; assigning complex conductivity parameters corresponding to the main frequencies of the multi-channel original magnetic cardiomyography signals to various tissues; wherein the complex conductivity parameters are based on a biological tissue dielectric properties database; importing the segmented and parameterized 3D label map into finite element analysis software to generate an unstructured tetrahedral mesh, and forming a target individual-specific thoracic conductor model after mesh convergence analysis; wherein the mesh of the heart region and the magnetic cardiomyography sensor region is more refined than that of other regions.

[0051] Image segmentation and tissue classification:

[0052] The registered MRI 3D thoracic structure image data is imported into medical image processing software. A deep learning segmentation model (such as the nnUnet network pre-trained on a public dataset) is used to automatically segment the thoracic structure into the following tissue categories: skin, subcutaneous fat, bones (ribs, sternum, spine), skeletal muscles, lungs (distinguishing between left and right), heart (myocardium and blood pool), and the remaining area (default is homogeneous soft tissue).

[0053] Electromagnetic parameter assignment:

[0054] For each segmented tissue category, a complex conductivity parameter corresponding to the dominant frequency (approximately 1 Hz to 100 Hz) of the multichannel raw magnetic cardiomyography signal was assigned. The complex conductivity parameters were referenced from published databases of biological tissue dielectric properties (such as the ITIS Foundation database). The complex conductivity parameters are as follows: subcutaneous fat 0.05 S / m, bone 0.02 S / m, skeletal muscle 0.35 S / m, lung (end-tidal) 0.1 S / m, blood 0.7 S / m, myocardium 0.2 S / m, etc.

[0055] Finite element model generation:

[0056] The segmented and parameterized 3D labeled image is imported into the finite element analysis software (COMSOL Multiphysics), which automatically generates an unstructured tetrahedral mesh. To ensure computational accuracy and efficiency, mesh convergence analysis is performed: the mesh is refined (element edge length can be as small as 1 mm) in the heart region and near the magnetocardiogram sensor, while a coarser mesh (element edge length can be up to 5 mm) is used in other regions far from the region of interest. The final result is an individualized thoracic conductor model file containing approximately 2 to 5 million tetrahedral elements.

[0057] Step S30, Construction of the transfer matrix for individualized magnetic field attenuation: Based on the thoracic conductor model, the heart region is set as the source space and the position of the magnetocardiogram sensor array is set as the measurement space. Through electromagnetic field forward problem simulation, a transfer matrix characterizing individual-specific magnetic field attenuation is generated.

[0058] This embodiment then uses the aforementioned individualized thoracic conductor model as a basis, setting the heart region as the source space and the position of the magnetocardiogram sensor array as the measurement space. Through electromagnetic field forward problem simulation, the magnetic field transfer relationship between the unit current dipole at the discrete grid point in the source space and the probes of each magnetocardiogram sensor in the measurement space is calculated, ultimately forming a transfer matrix H characterizing the individual-specific magnetic field attenuation. This transfer matrix H fully encodes the spatial attenuation mode of the magnetic field caused by the unique anatomical structure of the subject.

[0059] Figure 3 A schematic diagram illustrating the principle of constructing a transfer matrix for individualized magnetic field attenuation according to an embodiment of this application is shown. (Refer to...) Figure 3 As shown, in some optional implementations, the source space in step S30 is set as a regular three-dimensional grid of points in the heart region of the thoracic conductor model; wherein each grid point in the three-dimensional grid represents the position of a current dipole; the measurement space in step S30 is set as the spatial coordinates corresponding to the center point of each probe in the magnetocardiogram sensor array; the generation of the transfer matrix characterizing individual-specific magnetic field attenuation through electromagnetic field forward problem simulation in step S30 includes: using a quasi-static current solver to calculate the magnetic field transfer relationship from each current dipole in the source space to each probe in the measurement space; wherein the quasi-static current solver is based on the physical field assumption of ignoring displacement current at low frequencies and satisfies the following equation: ;in, Indicates electrical conductivity. It represents electric potential.

[0060] Define source space:

[0061] Within the cardiac region (including the left and right ventricles and atria) of a thoracic conductor model obtained based on MRI segmentation, a regular three-dimensional grid is defined with a grid spacing of 3 mm. Each grid point represents the location of a current dipole. Let the number of grid points (source points) be Ns.

[0062] Define the measurement space:

[0063] It precisely corresponds to the spatial coordinates of the center point of each OPM probe in the 64-channel MCG magnetocardiogram sensor array.

[0064] Forward problem solving:

[0065] In COMSOL, the quasi-static current solver (Electric Currents interface in the AC / DC module) is used. Based on the physical field assumption of neglecting displacement current at low frequencies, the following formula (1) is satisfied:

[0066] , formula (1);

[0067] in, Indicates electrical conductivity. It represents electric potential.

[0068] In some optional implementations, a quasi-static current solver is used to calculate the magnetic field transfer relationship between each current dipole in the source space and each probe in the measurement space, including: for each grid point in the source space, a unit current dipole is placed sequentially in the three orthogonal directions of x, y, and z to form a current dipole; for a unit current dipole in one direction, the normal component of the magnetic flux density generated by the unit current dipole at all probes of the magnetocardiogram sensor array is calculated; for a three-dimensional grid lattice with Ns grid points and a magnetocardiogram sensor array with M probes, a transfer matrix H with dimension M×(3Ns) is obtained; where each column of H corresponds to the magnetic field distribution generated by a unit current dipole in a specific grid point and a specific direction on the magnetocardiogram sensor array.

[0069] For a grid point i in the source space, a unit current dipole (mathematically represented as a two-point current source) is placed sequentially in the three orthogonal directions x, y, and z. Maxwell's equations are solved using the finite element method to calculate the normal component (perpendicular to the sensor coil plane) of the magnetic flux density generated by the unit current dipole in one direction at all probes of the M=64 magnetocardiogram sensor array. Recordings are made for each direction.

[0070] For a three-dimensional grid lattice with Ns source points and a magnetocardiogram sensor array with M probes, a transfer matrix H of dimension M×(3Ns) is obtained. Each column of H corresponds to the magnetic field distribution (i.e., derived field) generated by a unit current dipole at a specific grid point and in a specific direction on the entire magnetocardiogram sensor array.

[0071] Step S40, inverse signal compensation: a compensation operator is constructed using the transfer matrix to transform the original multi-channel magnetic field signals and obtain the attenuation-corrected magnetic field signal vector.

[0072] This embodiment then utilizes the transfer matrix H or its generalized inverse (or pseudo-inverse) H. + The actual acquired multi-channel raw magnetic resonance imaging (MRI) signals Perform linear or nonlinear transformations: . This is a compensation operator based on the H design. The physical meaning of this step is to mathematically reverse the attenuation effect described by the transfer matrix H, outputting a set of magnetocardiogram signal vectors whose conduction attenuation has been corrected. .

[0073] Figure 4 A schematic diagram illustrating the principle of signal inverse compensation according to an embodiment of this application is shown. (Refer to...) Figure 4 As shown, in some optional implementations, step S40 includes: constructing a compensation operator for the transfer matrix using the Tikhonov regularization method, with the following formula: ;in, Represents the compensation operator. Represents the transfer matrix. Indicates transpose. Represents the identity matrix. Represents the regularization parameter. The L-curve method or generalized cross-validation method is used to automatically determine the signal; a compensation operator is applied to transform the multi-channel raw magnetocardiogram signal, and the transformation formula is as follows: ;in, This represents the attenuation-corrected magnetic field signal vector at a single time point. This represents the raw magnetic field signals from multiple channels acquired at a single time point.

[0074] Constructing the compensation operator:

[0075] Directly inverting the transfer matrix H is not feasible because H is typically a wide matrix with severe ill-conditioned behavior. This application employs the Tikhonov regularization method to construct the compensation operator. The formula for constructing the compensation operator is as follows (2):

[0076] , formula (2);

[0077] in, H represents the compensation operator, T represents the transfer matrix, I represents the identity matrix, and λ represents the regularization parameter. λ is used to balance the spatial smoothness of data fitting and solution. λ is automatically determined by the L-curve method or generalized cross-validation.

[0078] The dimension is (3Ns) × M. Mathematically, It is the regularized pseudoinverse of H, i.e., H + .

[0079] Signal compensation calculation:

[0080] Suppose the multi-channel raw magnetocardiogram signal acquired at a single time point is represented as an M×1 column vector. Applying compensation operators Perform the linear transformation as shown in formula (3):

[0081] , formula (3);

[0082] in, This represents the attenuation-corrected magnetic field signal vector at a single time point. It is a (3Ns)×1 vector. The physical meaning can be interpreted as follows: a "virtual" current dipole moment intensity is reconstructed at each grid point and in each direction in the source space. This intensity distribution map has reversed the individualized attenuation effect from the heart to the body surface, and can be regarded as a set of "attenuation-corrected equivalent source distribution maps".

[0083] In practical implementation, the construction of the compensation operator (which involves a large amount of computation) can be completed in advance before measurement. The signal compensation calculation (which involves a small amount of computation) can be processed in real time (millisecond level) through a graphics processor or field-programmable gate array integrated into the system, and online compensation can be performed on the continuously acquired magnetocardiogram signal stream.

[0084] Step S50, General High-Precision Source Imaging: Input the cardiac magnetic signal vector into the standard cardiac magnetic source imaging algorithm to solve the inverse problem and output the spatial distribution result of cardiac power source.

[0085] Finally, this embodiment will display the attenuation-corrected magnetocardiogram signal vector. As input, it is fed into any standard cardiac magnetic source imaging algorithm (such as ECD equivalent current dipole, current density imaging, beamforming imaging, etc.) to solve the inverse problem. Due to the input signal... Individual attenuation differences have been pre-corrected, so even when using general source imaging algorithms, significantly improved reconstruction results that are closer to the true distribution of cardiac electrical activity can be obtained.

[0086] In some alternative implementations, the standard cardiac magnetic source imaging algorithm in step S50 includes: equivalent current dipole, current density imaging, or beamforming imaging.

[0087] In some optional implementations, step S50 includes: using current density imaging to extract the x, y, and z directional components of all grid points in the source space from the magnetic field signal vector, and obtaining vector current density distribution maps in the three directions respectively; calculating the magnitude of the vector current density distribution maps in the three directions at each time point to obtain the current density amplitude map of each grid point in the source space; and visually presenting the spatiotemporal propagation process of cardiac excitation by setting a threshold.

[0088] Input signal preparation:

[0089] Attenuation-corrected single-time-point magnetocardiogram signal vector By selecting the x-direction components corresponding to all grid points in the source space, a corrected x-direction current density distribution map with dimension Ns×1 is reconstructed. Similarly, we can obtain and . , and The three components together characterize the vector current density distribution in the source space.

[0090] Applications of standard source imaging algorithms:

[0091] For each time point t, calculate , and The current density amplitude map of each grid point in the source space is obtained by using the modulus. By setting a threshold, the spatiotemporal propagation process of cardiac excitation can be visualized.

[0092] In some alternative implementations, step S50 includes: employing beamforming imaging, inputting the magnetocardiogram signal vector into a normalized sLORETA algorithm, and solving the following optimization problem: ;in, This represents the current density distribution vector in the source space at a single time point. Let L represent the magnetocardiogram signal vector, L represent the simplified forward model of a homogeneous medium, and μ represent the regularization parameter.

[0093] Input signal preparation:

[0094] Directly convert the attenuation-corrected single-time-point magnetic signal vector As input, a "virtual, high-dimensional, corrected magnetic field signal".

[0095] Applications of standard source imaging algorithms:

[0096] Will The input is fed into the standardized sLORETA algorithm. The core of sLORETA is to solve the optimization problem of the following formula (4):

[0097] , formula (4);

[0098] in, Let L represent the current density distribution vector of the source space at a single time point, and L represent the simplified forward model of the homogeneous medium (because individual decays have already been...). (The compensation is in the middle), μ represents the regularization parameter.

[0099] Due to input It is "cleaner" and even when using the simplified model L, the algorithm can output power source localization results with better focus.

[0100] Result verification:

[0101] In the simulation experiment, a simulated dipole source with a known position and orientation was set up, and reconstructed using both the "traditional method with a uniform sphere model" and the "method of this application". The positioning error of the traditional method was 12±3mm, while the positioning error of the method of this application was reduced to 4±1mm.

[0102] In a physical simulation experiment using a real human thoracic cavity model, the method in this application improved the correlation coefficient between the reconstructed dipole position and the real position from 0.78 to 0.95 compared to the method of directly using a personalized model for the weighted minimum norm solution.

[0103] Figure 5 A schematic diagram of a thoracic conductor modeling and signal attenuation compensation system for magnetocardiography imaging according to an embodiment of this application is shown. Figure 6 A block diagram of a thoracic conductor modeling and signal attenuation compensation system for magnetocardiography imaging according to one embodiment of this application is shown. (Refer to...) Figure 5 and Figure 6 As shown, the thoracic conductor modeling and signal attenuation compensation system 600 for magnetocardiography imaging includes:

[0104] The synchronous acquisition and control unit 610 coordinates and controls the data acquisition process of medical imaging equipment and magnetocardiography equipment and integrates an optical tracking system to achieve synchronous acquisition and registration of multimodal data.

[0105] The Automated Modeling and Calculation Engine 620 includes: an image segmentation and tissue classification module, an electromagnetic parameter allocation module, a numerical model generator, and a forward simulation calculator for transfer matrices, used to realize the construction of personalized thoracic conductor models and the construction of transfer matrices for individualized magnetic field attenuation.

[0106] The online signal compensation processor 630 includes an algorithm module and high-speed digital signal processing hardware for implementing inverse signal compensation.

[0107] Standardized source imaging and display terminal 640 is used to achieve universal high-precision source imaging and fusion visualization on a target individual heart 3D model;

[0108] The System Workflow Manager 650 is used to automatically manage data flow and control flow according to a fixed logical sequence.

[0109] In this embodiment, the synchronous acquisition and control unit 610 coordinates and controls the data acquisition process of medical imaging equipment (such as MRI) and magnetocardiography equipment, and integrates an optical tracking system to achieve synchronous acquisition and registration of multimodal data; the automated modeling and calculation engine 620 includes: an image segmentation and tissue classification module, an electromagnetic parameter allocation module, a numerical model (such as FEM mesh) generator, and a forward simulation calculator for the transfer matrix, used to realize the construction of personalized thoracic conductor models and the construction of individualized magnetic field attenuation transfer matrices; the core of the online signal compensation processor 630 is an algorithm module that calculates compensation operators in real time based on the transfer matrix and high-speed digital signal processing hardware (such as FPGA or GPU) that performs inverse signal compensation calculations; the standardized source imaging and display terminal 640 receives the attenuation-corrected magnetocardiogram signal vector, calls multiple built-in or external source imaging algorithms for solving, and fuses and visualizes the results on an individual heart 3D model, generates a clinical report, and realizes navigation docking; the system workflow manager 650 manages the data flow and control flow of the entire system automatically according to a fixed logical sequence of "registration → modeling → calculation of transfer matrix → compensation → imaging", ensuring the continuity and repeatability of the process.

[0110] In some optional embodiments, in the above system, the synchronous acquisition and control unit 610 is used to: acquire three-dimensional thoracic structural image data of the target individual using MRI, CT, or 3D structured light scanning; acquire multi-channel magnetocardiogram (MCC) signals of the target individual at rest for several heartbeat cycles using a multi-channel OPM MCC system, and perform bandpass filtering on the multi-channel MCC signals to obtain the target individual's multi-channel raw MCC signals; wherein the sampling frequency is not less than 1 kHz; attach positioning markers to bony landmarks and corner points of the MCC sensor array on the target individual, record the coordinates of the positioning markers using an optical tracking system, calculate the spatial rigid body transformation matrix using a point set registration algorithm, and transform the image spatial coordinate system to the MCC sensor array spatial coordinate system; wherein the registration error is less than 2 mm.

[0111] In some optional implementations, in the above system, the automated modeling and computation engine 620 is used to: segment and classify the registered thoracic 3D structural image data using a deep learning segmentation model pre-trained on a public dataset; assign complex conductivity parameters corresponding to the main frequencies of the multi-channel original magnetic cardiomyography signals to various tissues; wherein the complex conductivity parameters are based on a biological tissue dielectric properties database; import the segmented and parameterized 3D label map into finite element analysis software to generate an unstructured tetrahedral mesh, and form a thoracic conductor model specific to the target individual after mesh convergence analysis; wherein the mesh of the heart region and the magnetic cardiomyography sensor region is more refined than that of other regions.

[0112] In some alternative implementations, in the above system, the automated modeling and calculation engine 620 is used to: define the source space as a regular three-dimensional grid of points representing the heart region of the thoracic conductor model; wherein each grid point in the three-dimensional grid represents the position of a current dipole; define the measurement space as the spatial coordinates corresponding to the center point of each probe in the magnetocardiogram sensor array; and employ a quasi-static current solver to calculate the magnetic field transfer relationship between each current dipole in the source space and each probe in the measurement space; wherein the quasi-static current solver is based on the physical field assumption of neglecting displacement current at low frequencies and satisfies the following equation: ;in, Indicates electrical conductivity. It represents electric potential.

[0113] In some optional implementations, in the above system, the automated modeling and calculation engine 620 is used to: for each grid point in the source space, place a unit current dipole in the three orthogonal directions x, y, z to form a current dipole; for a unit current dipole in one direction, calculate the normal component of the magnetic flux density generated by the unit current dipole at all probes of the magnetocardiogram sensor array; for a three-dimensional grid lattice with Ns grid points and a magnetocardiogram sensor array with M probes, obtain a transfer matrix H with dimension M×(3Ns); wherein each column of H corresponds to the magnetic field distribution generated by a unit current dipole in a specific grid point and a specific direction on the magnetocardiogram sensor array.

[0114] In some alternative implementations, in the above system, the online signal compensation processor 630 is used to: construct a compensation operator for the transfer matrix using the Tikhonov regularization method, with the following construction formula: ;in, Represents the compensation operator. Represents the transfer matrix. Indicates transpose. Represents the identity matrix. Represents the regularization parameter. The L-curve method or generalized cross-validation method is used to automatically determine the signal; a compensation operator is applied to transform the multi-channel raw magnetocardiogram signal, and the transformation formula is as follows: ;in, This represents the attenuation-corrected magnetic field signal vector at a single time point. This represents the raw magnetic field signals from multiple channels acquired at a single time point.

[0115] In some alternative implementations, in the above system, the standardized source imaging and display terminal 640 is used to employ a standard cardiac magnetic source imaging algorithm including equivalent current dipole, current density imaging, or beamforming imaging.

[0116] In some alternative implementations, in the above system, the standardized source imaging and display terminal 640 is used to: use current density imaging to extract the x, y, and z directional components of all grid points in the source space from the magnetic heart signal vector, and obtain vector current density distribution maps in the three directions respectively; calculate the magnitude of the vector current density distribution maps in the three directions at each time point to obtain the current density amplitude map of each grid point in the source space; and visualize the spatiotemporal propagation process of cardiac excitation by setting a threshold.

[0117] In some alternative implementations, in the above system, the standardized source imaging and display terminal 640 is used to: employ beamforming imaging, input the magnetocardiogram signal vector into a standardized sLORETA algorithm, and solve the following optimization problem: ;in, This represents the current density distribution vector in the source space at a single time point. Let L represent the magnetocardiogram signal vector, L represent the simplified forward model of a homogeneous medium, and μ represent the regularization parameter.

[0118] It should be noted that the aforementioned thoracic conductor modeling and signal attenuation compensation system 600 for magnetocardiography imaging can implement the aforementioned thoracic conductor modeling and signal attenuation compensation methods for magnetocardiography imaging, which will not be elaborated further.

[0119] Figure 7 This invention illustrates a schematic diagram of the structure of an electronic device according to an embodiment of the present application. Figure 7 As shown, the electronic device includes a processor, internal memory, a network interface, and a non-volatile storage medium connected via a system bus. The processor provides computational and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used for communication with external devices via a network connection. When executed by the processor, the computer program implements the functions or steps of a method for thoracic conductor modeling and signal attenuation compensation in magnetocardiography imaging.

[0120] In one embodiment, the electronic device provided in this application includes a memory and a processor. The memory stores a database and a computer program that can run on the processor. When the processor executes the computer program, it implements the steps of a method for modeling thoracic conductors and compensating for signal attenuation in magnetocardiography imaging.

[0121] The above is as stated in this application. Figure 5The method for thoracic conductor modeling and signal attenuation compensation system execution in magnetocardiography imaging disclosed in the illustrated embodiment can be applied to a processor or implemented by a processor. During implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by software instructions. The processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The steps of the method disclosed in the embodiments of this application can be directly embodied as being executed by a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0122] In one embodiment, a computer-readable storage medium is also provided, on which a computer program is stored, wherein the computer program, when executed by a processor, implements the steps of a method for modeling thoracic conductors and compensating for signal attenuation in magnetocardiography imaging.

[0123] It should be noted that the functions or steps that the above-mentioned electronic devices or computer-readable storage media can achieve can be referred to the relevant descriptions in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0124] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0125] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0126] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for modeling the thoracic conductor and compensating for signal attenuation in magnetocardiography imaging, characterized in that, include: Step S10, Multimodal data synchronous acquisition and registration: Simultaneously or simultaneously acquire the three-dimensional thoracic structure image data and multi-channel raw magnetic cardiomyocyte signals of the target individual, and perform fusion registration of image space, magnetic cardiomyocyte sensor array space and individual body surface space; Step S20, Personalized Thoracic Conductor Model Construction: The registered 3D thoracic structure image data is automatically segmented and classified into tissues, and complex conductivity parameters related to magnetocardiogram frequency are assigned to various tissues. A numerical calculation engine is used to generate a thoracic conductor model exclusive to the target individual. Step S30, Construction of individualized magnetic field attenuation transfer matrix: Based on the thoracic conductor model, the heart region is set as the source space and the position of the cardiac magnetosensor array is set as the measurement space. Through electromagnetic field forward problem simulation, a transfer matrix characterizing individual-specific magnetic field attenuation is generated. Step S40, inverse signal compensation: a compensation operator is constructed using the transfer matrix to transform the original multi-channel magnetocardiogram signal and obtain the attenuation-corrected magnetocardiogram signal vector. Step S50, General High-Precision Source Imaging: Input the cardiac magnetic signal vector into the standard cardiac magnetic source imaging algorithm to solve the inverse problem and output the spatial distribution result of cardiac power source.

2. The method for thoracic conductor modeling and signal attenuation compensation in magnetocardiography imaging according to claim 1, characterized in that, Step S10 includes: Three-dimensional thoracic structural image data of the target individual are obtained using MRI, CT, or 3D structured light scanning. A multi-channel OPM magnetocardiography system was used to acquire multi-channel magnetocardiogram signals of the target individual during several heartbeat cycles at rest. The multi-channel magnetocardiogram signals were then bandpass filtered to obtain the original multi-channel magnetocardiogram signals of the target individual. The sampling frequency was not less than 1 kHz. Positioning markers were affixed to the bony landmarks and corners of the magnetocardiogram sensor array of the target individual. The coordinates of the positioning markers were recorded using an optical tracking system. A point set registration algorithm was used to calculate the spatial rigid body transformation matrix, transforming the image spatial coordinate system to the magnetocardiogram sensor array spatial coordinate system. The registration error was less than 2 mm.

3. The method for thoracic conductor modeling and signal attenuation compensation in magnetocardiography imaging according to claim 1, characterized in that, Step S20 includes: A deep learning segmentation model pre-trained on a public dataset was used to perform tissue segmentation and classification on the registered 3D thoracic structure image data; Complex conductivity parameters corresponding to the main frequencies of multi-channel raw magnetic cardiomyocyte signals are assigned to various tissues; the complex conductivity parameters are based on a database of biological tissue dielectric properties. The segmented and parameterized 3D label map is imported into finite element analysis software to generate an unstructured tetrahedral mesh. After mesh convergence analysis, a thoracic conductor model specific to the target individual is formed. Among them, the mesh of the heart region and the magnetocardiogram sensor region is more refined than that of other regions.

4. The method for thoracic conductor modeling and signal attenuation compensation in magnetocardiography imaging according to claim 1, characterized in that, The source space in step S30 is set as follows: A regular three-dimensional grid of points in the heart region of the thoracic conductor model; wherein each grid point in the three-dimensional grid represents the location of a current dipole; The measurement space in step S30 is set as follows: Corresponding to the spatial coordinates of the center point of each probe in the magnetocardiogram sensor array; Step S30, which generates a transfer matrix characterizing individual-specific magnetic field attenuation through electromagnetic field forward problem simulation, includes: A quasi-static current solver is used to calculate the magnetic field transfer relationship between each current dipole in the source space and each probe in the measurement space. The quasi-static current solver is based on the physical field assumption of neglecting displacement current at low frequencies and satisfies the following equation: ； in, Indicates electrical conductivity. It represents electric potential.

5. The method for thoracic conductor modeling and signal attenuation compensation in magnetocardiography imaging according to claim 4, characterized in that, The method employs a quasi-static current solver to calculate the magnetic field transfer relationship between each current dipole in the source space and each probe in the measurement space, including: For each grid point in the source space, a unit current dipole is placed sequentially in the three orthogonal directions of x, y, and z to form a current dipole; For a unit current dipole in one direction, calculate the normal component of the magnetic flux density produced by the unit current dipole at all probes of the magnetocardiogram sensor array. For a three-dimensional grid lattice with Ns grid points and a magnetocardiogram sensor array with M probes, a transfer matrix H with dimension M×(3Ns) is obtained; where each column of H corresponds to the magnetic field distribution generated on the magnetocardiogram sensor array by a unit current dipole at a specific grid point and in a specific direction.

6. The method for thoracic conductor modeling and signal attenuation compensation in magnetocardiography imaging according to claim 1, characterized in that, Step S40 includes: The compensation operator for the transfer matrix is ​​constructed using the Tikhonov regularization method, and the construction formula is as follows: ； in, Represents the compensation operator. Represents the transfer matrix. Indicates transpose. Represents the identity matrix. Represents the regularization parameter. Automatically determined using the L-curve method or generalized cross-validation; The multi-channel raw magnetocardiogram signal is transformed using a compensation operator. The transformation formula is as follows: ； in, This represents the attenuation-corrected magnetic field signal vector at a single time point. This indicates the multi-channel raw magnetic field signal acquired at a single time point.

7. The method for thoracic conductor modeling and signal attenuation compensation in magnetocardiography imaging according to claim 1, characterized in that, The standard cardiac magnetic source imaging algorithm in step S50 includes: equivalent current dipole, current density imaging, or beamforming imaging.

8. The method for thoracic conductor modeling and signal attenuation compensation in magnetocardiography imaging according to claim 7, characterized in that, Step S50 includes: By using current density imaging, the x, y, and z components of all grid points in the source space are extracted from the magnetocardiogram signal vector, and the vector current density distribution maps in the three directions are obtained respectively. The magnitude of the vector current density distribution map in three directions is calculated for each time point to obtain the current density amplitude map of each grid point in the source space; By setting a threshold, the spatiotemporal propagation process of cardiac excitation can be visualized.

9. The method for thoracic conductor modeling and signal attenuation compensation in magnetocardiography imaging according to claim 7, characterized in that, Step S50 includes: Using beamforming imaging, the magnetocardiogram signal vector is input into the standardized sLORETA algorithm to solve the following optimization problem: ； in, This represents the current density distribution vector in the source space at a single time point. Let L represent the magnetic signal vector, L represent the simplified forward model of a homogeneous medium, and μ represent the regularization parameter.

10. A system for modeling thoracic conductors and compensating for signal attenuation in magnetocardiography imaging, characterized in that, The method for thoracic conductor modeling and signal attenuation compensation in magnetocardiography imaging according to any one of claims 1 to 9 includes: The synchronous acquisition and control unit coordinates and controls the data acquisition process of medical imaging equipment and magnetocardiography equipment and integrates an optical tracking system to achieve synchronous acquisition and registration of multimodal data; The automated modeling and calculation engine includes: an image segmentation and tissue classification module, an electromagnetic parameter allocation module, a numerical model generator, and a forward simulation calculator for transfer matrices, used to realize the construction of personalized thoracic conductor models and the construction of transfer matrices for individualized magnetic field attenuation. An online signal compensation processor includes an algorithm module and high-speed digital signal processing hardware for implementing inverse signal compensation. A standardized source imaging and display terminal is used to achieve universal high-precision source imaging and integrate visualization on a three-dimensional model of the target individual's heart. The system workflow manager is used to automatically manage data flow and control flow according to a fixed logical sequence.

Images