Method and system for dynamic reconstruction of thoracic whole organ and respiratory monitoring driven by point cloud

Abstract

The application discloses a point cloud driven thoracic cavity whole organ dynamic reconstruction and respiratory monitoring method and system, and the method implementation comprises the following steps: based on the thoracic medical image data of a target object, constructing three-dimensional geometric models corresponding to each target structure in the thoracic cavity, and converting the three-dimensional geometric models into static point cloud data; based on the respiratory motion law of the target object, driving the static point cloud data to perform dynamic deformation simulation in a respiratory cycle, and obtaining dynamic point cloud data synchronized with the respiratory phase; synchronously associating the dynamic point cloud data with electrical characteristic parameters in time and space, and constructing a digital twin thoracic cavity model; deploying a virtual sensor assembly that dynamically deforms with the thoracic cavity in the digital twin thoracic cavity model, and simulating and monitoring the dynamic respiratory process of the target object, and generating a virtual physiological signal; and through signal processing on the virtual physiological signal, a dynamic reconstruction image reflecting the respiratory motion is obtained. The continuous simulation of the whole respiratory motion process is realized, and the time and space resolution and diversity of the virtual database are improved.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a point cloud-driven method and system for dynamic reconstruction of all organs in the thoracic cavity and respiratory monitoring. Background Technology

[0002] In the field of clinical respiratory function monitoring and imaging, EIT (electrical impedance tomography) combined with digital twin technology has become an important research direction, while the accurate acquisition and dynamic simulation of three-dimensional data is the core technical bottleneck in this field.

[0003] Traditional 3D data processing and related technologies have significant drawbacks: traditional methods often rely on reconstructing 3D models from 2D cross-sections, making it difficult to accurately capture the 3D deformation characteristics of the whole lung during respiration and failing to reconstruct the dynamic physiological state within the respiratory cycle; 3D models reconstructed from CT tomographic images have complex and unevenly distributed triangular patches on their surfaces, which can lead to a rapid increase in data volume and redundancy, affecting the efficiency of subsequent dynamic simulations and calculations; clinically measured respiratory data are limited by ethical constraints, operational risks, and other factors, making collection difficult and sample sizes small, which cannot meet the needs of AI algorithm training and digital twin model optimization for massive and diverse data.

[0004] Meanwhile, traditional respiratory monitoring and imaging technologies also have other limitations: physical electrode attachment can easily cause skin irritation, and the relative displacement of the electrodes and the chest cavity during breathing can lead to signal distortion; standardized chest cavity models are difficult to adapt to individual differences in anatomical features; ultimately affecting the accuracy of monitoring and imaging. Summary of the Invention

[0005] Therefore, it is necessary to provide a point cloud-driven dynamic reconstruction of all organs in the thoracic cavity and respiratory monitoring method and system to address the above-mentioned technical problems, so as to solve at least one of the problems existing in the prior art.

[0006] Firstly, a point cloud-driven method for dynamic reconstruction of all organs in the thoracic cavity and respiratory monitoring is provided, including:

[0007] Based on the chest medical imaging data of the target object, a three-dimensional geometric model corresponding to each target structure in the thoracic cavity is constructed, and the three-dimensional geometric model is converted into static point cloud data;

[0008] Based on the respiratory motion pattern of the target object, the static point cloud data is driven to perform dynamic deformation simulation within the respiratory cycle, thereby obtaining dynamic point cloud data synchronized with breathing.

[0009] The dynamic point cloud data and electrical characteristic parameters are spatiotemporally correlated to construct a digital twin thoracic cavity model;

[0010] Virtual sensing components that dynamically deform with the chest cavity are deployed in the digital twin chest cavity model to form a virtual sensing array;

[0011] The dynamic respiratory process of the target object is simulated and monitored based on the virtual sensor array to generate virtual physiological signals.

[0012] By processing the virtual physiological signals, a dynamic reconstructed image reflecting respiratory movements is obtained.

[0013] Secondly, a point cloud-driven dynamic reconstruction and respiratory monitoring system for all organs in the thoracic cavity is provided, including:

[0014] The static point cloud data generation unit is used to construct a three-dimensional geometric model corresponding to each target structure in the thoracic cavity based on the chest medical imaging data of the target object, and convert the three-dimensional geometric model into static point cloud data.

[0015] The dynamic point cloud data generation unit is used to drive the static point cloud data to perform dynamic deformation simulation within the respiratory cycle based on the respiratory motion law of the target object, so as to obtain dynamic point cloud data synchronized with breathing.

[0016] A digital twin thoracic cavity model construction unit is used to spatiotemporally correlate the dynamic point cloud data with electrical characteristic parameters to construct a digital twin thoracic cavity model.

[0017] A virtual sensor array generation unit is used to deploy virtual sensor components that dynamically deform with the chest cavity in the digital twin chest cavity model to form a virtual sensor array.

[0018] The virtual physiological signal generation unit is used to simulate and monitor the dynamic respiratory process of the target object based on the virtual sensor array, and generate virtual physiological signals.

[0019] The dynamic image reconstruction unit is used to obtain a dynamic reconstructed image reflecting respiratory movements by processing the virtual physiological signals.

[0020] The aforementioned point cloud-driven dynamic reconstruction and respiratory monitoring method and system for the entire thoracic cavity includes the following steps: Based on the target object's chest medical imaging data, constructing a three-dimensional geometric model corresponding to each target structure within the thoracic cavity, and converting the three-dimensional geometric model into static point cloud data; based on the respiratory motion patterns of the target object, driving the static point cloud data to perform dynamic deformation simulation within the respiratory cycle, obtaining dynamic point cloud data synchronized with respiration; spatiotemporally and synchronously associating the dynamic point cloud data with electrical characteristic parameters to construct a digital twin thoracic cavity model; deploying virtual sensing components that dynamically deform with the thoracic cavity in the digital twin thoracic cavity model to form a virtual sensing array; simulating and monitoring the dynamic respiratory process of the target object based on the virtual sensor array, generating virtual physiological signals; and processing the virtual physiological signals to obtain a dynamically reconstructed image reflecting respiratory motion. In this embodiment, a three-dimensional geometric model of the thoracic cavity is constructed using point cloud technology. This model is then expanded into multiple frames of dynamic expiratory data using single-frame point cloud data at the end of inspiration, fully capturing the three-dimensional deformation features of the entire lung and achieving continuous simulation of the entire respiratory process. This ensures that the deformation of each organ during the dynamic process conforms to the physiological changes of the human body's organs during normal breathing, effectively solving the problem of missing three-dimensional dynamic data in traditional technologies. The digital twin thoracic cavity model is constructed based on individual point cloud data, perfectly adapting to the anatomical features of the target object. Virtual sensing components are deployed synchronously with the dynamic deformation of the thoracic cavity, avoiding the safety risks and signal distortion problems caused by physical electrodes. This ensures that the generated virtual physiological signals are highly consistent with real signals in terms of amplitude and dynamic fluctuation patterns. The signal processing and imaging processes fully coordinate with the spatiotemporal characteristics of the point cloud dynamic model, effectively suppressing image artifacts and enhancing detail recognition. The resulting dynamically reconstructed image accurately reflects the differences in ventilation function in different lung regions during the respiratory cycle, achieving non-invasive and continuous respiratory monitoring and imaging without relying on a large amount of clinical experimental data and high-frequency physical examinations. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention 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.

[0022] Figure 1 This is a flowchart illustrating a point cloud-driven dynamic reconstruction and respiratory monitoring method for all organs in the thoracic cavity according to an embodiment of the present invention.

[0023] Figure 2 This is a schematic diagram of a point cloud-based scene of a three-dimensional geometric model in one embodiment of the present invention;

[0024] Figure 3aThis is a schematic diagram of the model structure of a case model of lung atelectasis simulation 1 in one embodiment of the present invention;

[0025] Figure 3b This is a schematic diagram of the model structure of a case model of lung atelectasis simulation 2 in one embodiment of the present invention;

[0026] Figure 3c This is a schematic diagram of the model structure of a simulated case model of pulmonary cysts in one embodiment of the present invention;

[0027] Figure 3d This is a schematic diagram of the model structure of a simulated case model of pulmonary pneumonia in one embodiment of the present invention;

[0028] Figure 3e This is a schematic diagram of the model structure of a simulated case 2 of pulmonary pneumonia in one embodiment of the present invention;

[0029] Figure 3f This is a schematic diagram of the structure of a simulated case model of an ellipsoidal cavity in the lung according to an embodiment of the present invention;

[0030] Figure 3g This is a schematic diagram of the model structure of a case model in which the corresponding inflammatory area of ​​the patient is substituted into an embodiment of the present invention;

[0031] Figure 4 This is a comparative schematic diagram of image reconstruction of multiple groups of original medical images of healthy lungs (FIT) and lungs with inflammation in each lung (PNEU1 / PNEU2) in one embodiment of the present invention.

[0032] Figure 5 This is a schematic diagram of a point cloud-driven dynamic reconstruction and respiratory monitoring system for all organs in the thoracic cavity according to an embodiment of the present invention.

[0033] Figure 6 This is a schematic diagram of a computer device according to an embodiment of the present invention. Detailed Implementation

[0034] 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, not all, of the embodiments of the present invention. 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.

[0035] In one embodiment, such as Figure 1 As shown, a point cloud-driven method for dynamic reconstruction of all organs in the thoracic cavity and respiratory monitoring is provided, including the following steps:

[0036] In step S110, based on the chest medical imaging data of the target object, a three-dimensional geometric model corresponding to each target structure in the thoracic cavity is constructed, and the three-dimensional geometric model is converted into static point cloud data;

[0037] Chest medical imaging data refers to clinical imaging data that can present the anatomical structure of the chest. Core types include CT (computed tomography) and MRI (magnetic resonance imaging), which can provide tomographic images and three-dimensional structural information of tissues and organs within the thoracic cavity.

[0038] Optionally, taking CT images of the chest as an example, professional software such as 3D Slicer and Wrap is used to process the CT images of the target object. First, organ segmentation algorithms are used to accurately separate multiple key organs and structures, including the lungs, heart, chest wall soft tissue, bones, liver, stomach, trachea, and esophagus (avoiding interference between different tissues). Then, based on the segmented organ contour data, 3D reconstruction technology is used to reproduce the spatial morphology, positional relationships, and anatomical details of each organ. Finally, surface smoothing optimization is used to eliminate imperfections such as jagged edges and holes that may occur during reconstruction, ultimately generating a structurally complete, morphologically accurate, and format-compatible (such as STL format, facilitating subsequent point cloud extraction) 3D geometric model of multiple chest organs. Then, based on this STL model, a point cloud extraction algorithm is used to collect the 3D coordinate point information of the model surface. After processing (such as deduplication and sorting of discrete coordinate points), the static point cloud data of each organ is obtained. Specifically, this can be done as follows: Figure 2 As shown.

[0039] In step S120, based on the respiratory motion pattern of the target object, the static point cloud data is driven to perform dynamic deformation simulation within the respiratory cycle to obtain dynamic point cloud data synchronized with breathing.

[0040] Optionally, the generated static point cloud data can be labeled with corresponding organ type tags to clarify its anatomical affiliation. Finally, a pre-built dynamic simulation algorithm (such as an organ-specific kinematic algorithm) is called, which has been personalized to suit the anatomical characteristics and motion differences of different target structures. The respiratory motion patterns of each target structure (such as the expansion and contraction of the lungs during ventilation, the rhythmic rise and fall of the thoracic cavity, and the linkage with surrounding tissues) are used as algorithm input to form targeted deformation driving instructions. Finally, these instructions drive the point cloud units in the static point cloud data labeled with organ type tags, allowing the point clouds of different organs to deform synchronously according to their own motion patterns (avoiding cross-target structure motion interference), ultimately accurately simulating the complete morphological change process of each organ from inhalation to exhalation during the respiratory cycle, restoring the dynamic linkage state of organs during real breathing.

[0041] It should be noted that, since the continuous simulation of organ morphology during respiration is achieved based on the parameters and trajectories of real respiratory movements, the generation of the dynamic process essentially only requires one frame of point cloud data. Therefore, only one frame of point cloud data at the end of inspiration or expiration needs to be extracted.

[0042] In step S130, the dynamic point cloud data and electrical characteristic parameters are spatiotemporally correlated to construct a digital twin thoracic cavity model;

[0043] Optionally, the temporal dimension (respiratory phase) and spatial dimension (organ location and morphology) information of the dynamic point cloud data are first clarified. Then, the electrical characteristic parameters (such as resistivity and conductivity, reflecting the electrical properties of tissues) corresponding to each organ are associated with the dynamic point cloud according to the principle of spatiotemporal synchronization. That is, under a specific respiratory phase, each point cloud unit in the thoracic cavity will be matched with the corresponding temporal electrical characteristic parameters of the corresponding tissue, ensuring that the electrical characteristic parameters are updated synchronously with the dynamic deformation of the organ. Through this deep fusion of multi-dimensional data, a digital twin thoracic cavity model that both restores the dynamic changes of the thoracic cavity anatomical structure and accurately maps the electrophysiological properties of tissues is finally constructed.

[0044] In step S140, virtual sensing components that dynamically deform with the chest cavity are deployed in the digital twin chest cavity model to form a virtual sensing array.

[0045] It should be noted that because the thoracic cavity constantly expands and contracts during respiration, the electrode pads attached to the thoracic cavity also move accordingly. To closely approximate reality, dynamic virtual sensing components can be formed based on the surface morphology and movement trends of the thoracic cavity in each frame. This is achieved by driving the virtual sensing array to move synchronously through the surface point cloud displacement transfer mechanism in the aforementioned digital twin thoracic cavity model. Specifically, in the constructed digital twin thoracic cavity model, virtual sensing components are deployed according to preset monitoring requirements. These components are not in fixed positions but are linked in real time with the dynamic deformation of the thoracic cavity, moving synchronously with the expansion and contraction of organs during respiration, always maintaining a stable relative position with the target monitoring area. At the same time, these dynamically adapted virtual sensing components are integrated according to specific arrangement rules to form a virtual sensing array covering key areas of the thoracic cavity and capable of simultaneously acquiring electrical signals from multiple sites.

[0046] The virtual sensing component can be an electrode pad, which can generate deployment points and geometric models that fit the chest cavity at different Z-axis height sections of the digital twin chest cavity model using layered positioning, contour sampling, and normal adaptation methods. These virtual electrode pads move synchronously with the expansion and contraction of the chest cavity during respiration, always maintaining a stable relative position with the target monitoring area. They are integrated into a virtual electrode array covering key areas of the chest cavity according to preset rules, simultaneously acquiring electrical signals from multiple sites, providing a high-fidelity sensing carrier for virtual monitoring simulations such as Electrical Impedance Tomography (EIT).

[0047] In step S150, the dynamic breathing process of the target object is simulated and monitored based on the virtual sensor array to generate virtual physiological signals;

[0048] Optionally, relying on the constructed virtual sensor array (such as a dynamic electrode array), the dynamic changes of the target object during the respiratory cycle are synchronously tracked in the digital twin chest cavity model. When the chest cavity expands or contracts with respiration, the virtual sensor moves synchronously with the organ deformation and maintains a stable relative position with the monitoring area. By simulating the process of applying excitation signals and collecting response data, the dynamic electrical characteristic parameters (such as electrical impedance) of the tissues in the chest cavity under different respiratory phases are captured in real time and finally converted into virtual physiological signals that are consistent with the real physiological signal characteristics output by the real sensor array.

[0049] In step S160, the virtual physiological signal is processed to obtain a dynamic reconstructed image reflecting respiratory movements.

[0050] Optionally, virtual physiological signals, such as voltage changes, collected through a virtual sensor array during respiration can be used. V (the weak voltage fluctuation modulated by respiration) is used as input data for image reconstruction, which is directly related to the dynamic changes in intrapulmonary conductivity (such as the decrease in local conductivity caused by gas filling during inspiration). Then, an inverse problem solving algorithm (such as the Tikhonov regularization method) or an AI algorithm can be used to train and retrieve the relative impedance distribution map of the cross section. The impedance differences in different regions of the map directly correspond to the regional heterogeneity of lung ventilation (such as some regions with adequate ventilation and low impedance, and some regions with insufficient ventilation and high impedance). At the same time, the changes in ventilation status can be dynamically tracked by combining the time dimension, so as to intuitively and quantitatively reflect the spatiotemporal distribution characteristics and heterogeneity of lung ventilation.

[0051] For example, if an inverse problem-solving algorithm (such as the Tikhonov regularization method) is used, a mapping between "voltage change" and "body impedance distribution" needs to be established through mathematical modeling. Regularization techniques are then used to suppress the ill-posedness of the inverse problem (avoiding divergence or distortion of the solution), and the impedance change pattern inside the thoracic cavity is inferred from the boundary voltage data. If an AI algorithm is used, the model needs to be trained based on a large number of labeled "voltage change-impedance distribution" samples, allowing the algorithm to autonomously learn the nonlinear relationship between the two and quickly output the reconstruction results through data-driven processing.

[0052] In this embodiment, a point cloud-driven method for dynamic reconstruction of all organs in the thoracic cavity and respiratory monitoring is provided, comprising: constructing a three-dimensional geometric model corresponding to each target structure in the thoracic cavity based on the chest medical imaging data of the target object, and converting the three-dimensional geometric model into static point cloud data; driving the static point cloud data to perform dynamic deformation simulation within the respiratory cycle based on the respiratory motion law of the target object, thereby obtaining dynamic point cloud data synchronized with respiration; spatiotemporally and synchronously associating the dynamic point cloud data with electrical characteristic parameters to construct a digital twin thoracic cavity model; deploying virtual sensing components that dynamically deform with the thoracic cavity in the digital twin thoracic cavity model to form a virtual sensing array; simulating and monitoring the dynamic respiratory process of the target object based on the virtual sensor array to generate virtual physiological signals; and obtaining a dynamic reconstructed image reflecting respiratory motion by signal processing of the virtual physiological signals. In this embodiment, a three-dimensional geometric model of the thoracic cavity is constructed using point cloud technology. This model is then expanded into multiple frames of dynamic expiratory data using single-frame point cloud data at the end of inspiration, fully capturing the three-dimensional deformation features of the entire lung and achieving continuous simulation of the entire respiratory process. This ensures that the deformation of each organ during the dynamic process conforms to the physiological changes of the human body's organs during normal breathing, effectively solving the problem of missing three-dimensional dynamic data in traditional technologies. The digital twin thoracic cavity model is constructed based on individual point cloud data, perfectly adapting to the anatomical features of the target object. Virtual sensing components are deployed synchronously with the dynamic deformation of the thoracic cavity, avoiding the safety risks and signal distortion problems caused by physical electrodes. This ensures that the generated virtual physiological signals are highly consistent with real signals in terms of amplitude and dynamic fluctuation patterns. The signal processing and imaging processes fully coordinate with the spatiotemporal characteristics of the point cloud dynamic model, effectively suppressing image artifacts and enhancing detail recognition. The resulting dynamically reconstructed image accurately reflects the differences in ventilation function in different lung regions during the respiratory cycle, achieving non-invasive and continuous respiratory monitoring and imaging without relying on a large amount of clinical experimental data and high-frequency physical examinations.

[0053] In one embodiment of this application, the step of constructing a three-dimensional geometric model corresponding to each target structure within the thoracic cavity based on the chest medical imaging data of the target object, and converting the three-dimensional geometric model into static point cloud data, includes:

[0054] Based on the chest medical imaging data, identify and segment the target structures within the thoracic cavity;

[0055] Three-dimensional modeling of each target structure within the thoracic cavity is performed, and the model surface is optimized through smoothing to obtain the three-dimensional geometric model corresponding to each target structure.

[0056] Discretize the surface of the three-dimensional geometric model corresponding to each target structure, extract the spatial location information of each sampling point, and generate static point cloud data corresponding to each target structure.

[0057] Optionally, the acquired chest medical image data of the target object is preprocessed using professional software such as 3DSlicer and Wrap. Multiple target structures, including the lungs, heart, chest wall soft tissue, bones, liver, stomach, trachea, and esophagus, are accurately identified using organ segmentation algorithms (avoiding interference between different tissues). Contour extraction and boundary delineation separate different target structures from the images, ensuring the independence and integrity of each structure. Then, based on the two-dimensional contour information of each target structure obtained after segmentation, three-dimensional modeling technology is used for stereo reconstruction, forming a preliminary three-dimensional geometric model. Subsequently, surface smoothing technology is used to eliminate imperfections such as jagged edges, depressions, and holes on the model surface, optimizing the model's regularity and simulation realism, making the model more closely resemble the actual morphology of human anatomy. Finally, high-density discretization sampling is performed on the surface of each optimized three-dimensional geometric model. The system extracts the three-dimensional spatial coordinate information of each sampling point, transforming the continuous three-dimensional geometric model into a discrete point cloud set, ultimately generating static point cloud data corresponding one-to-one with each target structure, completely preserving the spatial morphological characteristics of the original structure.

[0058] It should be noted that during the model training stage based on point cloud data (such as dynamic deformation simulation models and image reconstruction algorithm training), Gaussian noise (with the noise standard deviation set to 5%-10% of the point cloud spacing) can be introduced into the point cloud data to simulate errors such as equipment noise and motion artifacts that may occur during the acquisition of clinical medical images. This can enhance the diversity and robustness of the training data, thereby improving the model's adaptability to noise interference in real-world scenarios and ensuring the accuracy of the training results.

[0059] In one embodiment of this application, the step of driving the static point cloud data to perform dynamic deformation simulation within the respiratory cycle based on the respiratory motion pattern of the target object, thereby obtaining dynamic point cloud data synchronized with respiration, includes:

[0060] Based on the respiratory motion characteristics of each target structure during the respiratory cycle, the dynamic driving parameters corresponding to each target structure are determined.

[0061] Based on the dynamic driving parameters, the static point cloud data is dynamically adjusted at each phase of the respiratory cycle to generate a dynamic point cloud corresponding to each phase.

[0062] The dynamic point clouds of each time phase are integrated into a complete dynamic sequence according to the breathing time sequence to obtain the dynamic point cloud data.

[0063] Optionally, by analyzing the respiratory motion characteristics of each target structure during the respiratory cycle, such as the expansion and contraction amplitude of the lungs, the rise and fall trajectory of the thoracic cavity, and the coordinated rhythm of the heart and breathing, dynamic driving parameters are obtained by assigning individual values ​​to each target structure based on these characteristics. These parameters may include deformation rate, displacement range, and direction of motion, ensuring that the parameters are precisely matched with the physiological motion laws of the structure. Then, based on the set dynamic driving parameters, the static point cloud data of each target structure is dynamically adjusted for each key phase of the respiratory cycle (such as the beginning and end of inspiration, the beginning and end of expiration, etc.), so that the point cloud deforms according to the motion state of the corresponding phase, thereby generating a dynamic point cloud that conforms to the real motion characteristics in each phase. Finally, the dynamic point clouds of all phases can be arranged and spliced ​​together sequentially according to the respiratory sequence (the complete process from inspiration to expiration) to form a continuous and coherent dynamic point cloud sequence, ultimately obtaining dynamic point cloud data that is precisely synchronized with the respiratory process of the target object, completely restoring the dynamic change process of each structure during the respiratory cycle.

[0064] The following provides a detailed description of the respiratory characteristics and dynamic parameter determination process for multiple target structures, including the lungs, heart, bones, thoracic cavity, static liver, stomach, esophagus, and trachea.

[0065] Lung respiratory motion characteristics: Lung deformation is simulated using a four-axis displacement synthesis mechanism. Through the vector superposition of four components—radial contraction and expansion, vertical displacement, anteroposterior diameter contraction and expansion, and lateral diameter contraction and expansion—the complex motion patterns of lung tissue during respiration are realistically reproduced.

[0066] Radial constriction-extension displacement constitutes the core mechanical basis for changes in lung ventilation volume, which can be specifically expressed by the following formula:

[0067] ;

[0068] in, Represents the radial displacement vector; This is the radial displacement amplitude coefficient, which controls the foundation displacement strength. Define the radial motion direction as the unit direction vector from the lung center of mass to the current point cloud cell; The respiratory phase parameter varies continuously in the range of [-1,1] to achieve a smooth transition from exhalation to inhalation; The highly correlated intensity function characterizes the gradient distribution of displacement intensity along the vertical direction, where, The normalized height coordinates of the points are (0 for the base of the lung and 1 for the apex of the lung). Ensure that the apex region of the lungs maintains minimal exercise intensity. Control the maximum adjustment range of displacement strength by height. The function that determines the exponential decay rate of displacement intensity with increasing height accurately reproduces the regional heterogeneity where the amplitude of movement in the lower lobe is significantly greater than that in the upper lobe during diaphragmatic-dominated breathing. It is a lung displacement scaling factor, which is proportional to the average lung diameter, to ensure that the range of motion matches the organ size; The weighting coefficients in the anteroposterior diameter direction of the two lungs are used to regulate the displacement difference between the anterior and posterior regions, ensuring that the lungs near the anterior chest move significantly while the lungs near the posterior back remain basically fixed. The global calibration coefficients for lung displacement are determined through an adaptive iterative mechanism. Maintain a physiologically reasonable rate of volume change. To change lung volume for the target This represents the current change in lung volume.

[0069] Vertical displacement simulates the overall displacement of lung tissue caused by diaphragmatic contraction, and can be expressed by the following formula:

[0070] ;

[0071] in, Define the vertical displacement direction, accurately reflecting the head-to-tail movement of lung tissue caused by the rise and fall of the diaphragm along the Z-axis; It is the vertical displacement amplitude coefficient, which controls the displacement intensity in the vertical direction.

[0072] The anteroposterior diameter expansion displacement simulates the traction effect of the anterior-posterior movement of the thoracic cage on the lung tissue, which can be expressed by the following formula:

[0073] ;

[0074] in, It is the displacement amplitude coefficient of the front and rear diameters, which controls the displacement intensity in the front and rear diameter directions.

[0075] The lateral diameter expansion displacement characterizes the effect of lateral thoracic expansion on lung tissue, and can be specifically expressed by the following formula:

[0076] ;

[0077] in, These are the displacement amplitude coefficients for the left and right radial directions, controlling the displacement intensity in those directions; in the weighting function for the left and right directions, As a lateral weighting factor, and These are weight constraint boundaries used to ensure that the weights fall within a reasonable range. This represents the normalized position of a point in the left-right direction. This mechanism achieves a distribution pattern where movement is weak in the midline region and significant in the outer regions.

[0078] The final displacement of the lungs is represented by the superposition of components in four directions, which can be expressed by the following formula:

[0079] ;

[0080] Cardiac and respiratory motion characteristics: Cardiac motion is simulated using a biphasic coupling model of respiratory displacement and cardiac motion, which accurately simulates the coordinated motion of different cardiac chambers through zonal regulation and weighted smooth transition mechanisms. Specifically, values ​​can be assigned according to real physiological parameters (respiratory rate 12-20 breaths / minute, heart rate 60-100 beats / minute), with the cardiac cycle nested within the respiratory cycle, as shown below:

[0081] Respiratory phase shift characterizes the overall effect of changes in intrathoracic pressure on the heart, and can be specifically expressed by the following formula:

[0082] ;

[0083] in, This is the respiratory displacement coefficient. For heart characteristic dimensions, such as The displacement is 1% of the heart size; This indicates movement along the Z-axis to accurately simulate the traction effect of the diaphragm's rise and fall on the heart.

[0084] The phase of the heartbeat is determined by the heartbeat phase parameters. BT precise control, among which... bt is calculated based on time and heart rate and is used to describe the phase position of the cardiac cycle at the current moment. Ventricular contraction intensity Sven and ventricular contraction intensity Satr are expressed as follows:

[0085] ;

[0086] ;

[0087] Where A and B are the contraction amplitude parameters of the ventricle and atrium, respectively; This represents the phase delay of the atria relative to the ventricles.

[0088] To simulate the dynamic changes in heart morphology during the cardiac cycle, periodic scaling is applied along the long axis of the heart, which can be represented as:

[0089] ;

[0090] in, This is the axial expansion / contraction ratio. It represents the original anatomical length of the heart along its long axis in a resting state (such as end-diastole), serving as a reference scale for scaling calculations.

[0091] Torsion angle of cardiac torsional motion It can be represented as:

[0092] ;

[0093] in, The angle of twist between the apex and base of the heart, generally The angle should not exceed 15°; π / 180 is the coefficient for converting angles to radians, which is convenient for direct use in model calculations.

[0094] Skeletal Motion Characteristics: Employing a multi-zone control mechanism based on biomechanical principles, the thoracic skeleton is divided into the spinal, rib, and sternal zones. Through precise spatial division, differentiated motion rules, and transitional connections at the boundaries of each zone, physiologically realistic thoracic cage motion simulation is achieved, as detailed below:

[0095] The spinal anchoring zone is defined as a fixed area with the midline of the chest as the reference, and can be represented as:

[0096] ;

[0097] in, Threshold for the spinal region. This is the total width of the left and right diameters of the thoracic cavity. This area remains completely still, providing a stable biomechanical fulcrum for rib movement.

[0098] The transition weight function achieves a smooth motion transition from the spinal region to the activity region:

[0099] ;

[0100] in, The width of the transition area. The function restricts the weights to the [0,1] interval. This mechanism ensures motion continuity through distance-dependent weight gradients.

[0101] The coordinates of the head and tail points (along the Z-axis) are obtained through the motion. The rib area is further divided into the upper rib area, the lower rib area, and the middle rib area.

[0102] upper rib area Implementing a forward lifting movement mechanism:

[0103] ;

[0104] in, and Control the intensity of movement in the forward and backward and vertical directions respectively; This is the skeletal displacement scale factor. The ratio of skeletal displacement to lung displacement is constant to ensure coordination with lung movement; This is the overall amplitude enhancement coefficient; The weighting of the back attenuation significantly reduces the range of motion in the back region. Let be the normalized coordinates of the point to be moved in the forward and backward directions.

[0105] lower rib area Implementing a lateral deployment movement mechanism:

[0106] in, and The intensity of movement in the lateral and vertical directions is controlled separately.

[0107] Middle rib area Adopting a hybrid motion mode:

[0108] ;

[0109] Among them, the mixed weight Achieve a smooth transition from the front of the chest to the back, ensuring that the pump handle movement is dominated in the front chest area and the barrel handle movement is dominated in the lateral area.

[0110] As for the sternal region Implement a comprehensive proactive mechanism:

[0111] ;

[0112] in, and Controlling specific displacement characteristics of the sternal region.

[0113] Radial expansion component simulates overall thoracic deformation:

[0114] ;

[0115] in, Let be the radial unit vector originating from the center of mass of the thoracic cavity; expansion coefficient Ensure that the expansion at the bottom is greater than that at the top.

[0116] The final displacement of the skeletal system is achieved by the vector superposition of the displacements of each zone:

[0117] ;

[0118] This zonal regulation model realistically reproduces the coordinated three-dimensional movement of the thoracic cavity during respiration through precise anatomical constraints and biomechanical rules, providing a reliable skeletal movement basis for constructing a high-fidelity digital twin of the thoracic cavity.

[0119] Thoracic respiratory motion characteristics: Thoracic motion adopts a biaxial displacement model based on anatomical constraints. By precisely controlling the displacement components of the anterior-posterior diameter and the lateral diameter, the three-dimensional deformation characteristics of the thoracic cavity during the breathing process are realistically reproduced.

[0120] The anteroposterior displacement simulates the coordinated movement of the sternum and costal cartilage region, and its mathematical expression is:

[0121] ;

[0122] in, The anteroposterior diameter displacement direction coefficient is used to adjust the anteroposterior diameter movement amplitude and control the anterior chest region to move forward during inhalation and backward during exhalation. These are respiratory phase parameters used to control the timing of movement. The original anteroposterior diameter of the thoracic cavity is used to ensure that the displacement range matches the anatomical structure. The front and rear diaphragm displacement calibration coefficient is obtained by... Iterative optimization The relative change rate of the front and rear diameters of the target The current relative change rate of the anterior and posterior diameters is used to maintain the target deformation rate within the physiological range of 1.8-2.2%.

[0123] The forward and backward weighting functions adopt an S-shaped distribution:

[0124] ;

[0125] in, Controlling the steepness of the weight distribution, Determine the transition point between the preceding and following movements; the value is between 0 and 1. These are the normalized coordinates of the point to be moved in the forward and backward directions. This function ensures the anterior chest region... Significant movement, back area The movement is weak.

[0126] The height adjustment function characterizes the vertical gradient of the displacement intensity:

[0127] ;

[0128] in, Ensure that the top maintains basic exercise intensity. Controlling the range of adjustment of exercise intensity based on height. The coordinates are normalized height coordinates. This mechanism accurately reproduces the physiological characteristic that the lower chest has a greater range of motion than the upper chest.

[0129] Lateral displacement simulates rib abduction and retraction movements:

[0130]

[0131] in, The left and right radial displacement amplitude coefficients; Determine the direction of displacement. The coordinates of the moving point along the left and right radial directions, The coordinates of the center of the overall point cloud in the left and right radial directions are used to achieve bilateral symmetrical motion; This represents the original dimensions of the left and right diameters of the thoracic cavity. For the left and right radial displacement calibration coefficients, through The adaptive mechanism maintains the deformation rate within the physiological range of 0.7-1.2%. The relative change rate of the target left and right diameters, This represents the current relative rate of change between the left and right diameters.

[0132] The left and right weighting functions are subject to linear constraints.

[0133] ;

[0134] in, As a lateral weighting factor, and For the weighted boundary, The normalized position of the point in the left-right direction. This distribution ensures that the motion is weak in the midline region and significant in the outer region.

[0135] The final displacement of the thoracic cavity is achieved through vector synthesis:

[0136] ;

[0137] This model accurately reproduces the characteristic changes in the thoracic cavity diameter during calm breathing through precise matching of biaxial displacement components, providing an accurate geometric basis for the subsequent synchronous movement of the virtual electrode array.

[0138] It should be noted that for the four organ regions of liver, stomach, esophagus and trachea, the deformation and displacement are minimal and low during actual respiratory movements (e.g., the liver only undergoes a 1-2 cm vertical displacement with the rise and fall of the diaphragm, with no obvious morphological changes; the esophagus and trachea maintain stable luminal morphology), and the interference with respiratory imaging (e.g., lung ventilation distribution) is negligible. Therefore, static processing is performed. Specific requirements are: preserve the relative anatomical positions of the organs with the thoracic cavity and lungs (e.g., the liver is located below the right lung, and the trachea runs through the thoracic cavity in the middle) to ensure the accuracy of the overall spatial relationship of the model; static point cloud data does not participate in dynamic deformation calculations, but needs to be translated synchronously with the overall displacement of the thoracic cavity (e.g., slight vertical movement with the rise and fall of the diaphragm) to avoid spatial misalignment with dynamic organs.

[0139] By designing differentiated motion for the thoracic cavity (biaxial + gradient dynamic model), lungs / heart / bones (region-specific dynamic parameters), liver, etc. (static processing), the transformation of single-frame static point cloud (such as end of inspiration) into multi-frame dynamic point cloud of the entire respiratory cycle was successfully achieved. This fully reproduces the spatial position changes and morphological dynamics of the core organs in the thoracic cavity during respiration, solving the problem of missing three-dimensional dynamic data in traditional technologies.

[0140] In one embodiment of this application, the step of spatiotemporally and synchronously associating the dynamic point cloud data with electrical characteristic parameters to construct a digital twin thoracic cavity model includes:

[0141] Based on the aforementioned chest medical imaging data, the extent of the thoracic cavity was determined;

[0142] Based on the aforementioned thoracic cavity space range, a three-dimensional parameter matrix including electrical characteristic parameters is established;

[0143] Determine the corresponding electrical characteristic parameters according to the anatomical divisions of each target structure within the thoracic cavity;

[0144] Determine the anatomical affiliation of each spatial point to be tested in the three-dimensional parameter matrix and assign corresponding electrical characteristic parameters to generate a single-frame electrical characteristic parameter model.

[0145] By spatiotemporally aligning the dynamic point cloud data of each respiratory phase with the corresponding single-frame electrical characteristic parameter model, a digital twin thoracic cavity model covering the entire respiratory cycle is constructed.

[0146] Optionally, such as Figure 4 As shown, based on chest medical imaging data, the three-dimensional spatial range of the thoracic cavity is accurately determined by analyzing the anatomical boundary information in the images (such as the lateral border of the thoracic cage, the top of the diaphragm, and the thoracic cavity inlet). Within the defined thoracic cavity space, a three-dimensional parameter matrix is ​​constructed, the core of which includes the conductivity matrix (σ) and the relative permittivity matrix. There are two types. The first is the matching of the matrix's spatial resolution with dynamic point cloud data (e.g., each matrix element corresponds to the spatial position of a point cloud unit), used to store electrical characteristic parameters (conductivity, resistivity, etc.) for different spatial points. Based on the anatomical divisions of the target structures within the thoracic cavity (lungs, heart, bones, chest wall soft tissue, etc.), and in conjunction with clinical data, each structure is assigned unique electrical characteristic parameters. For example, the conductivity of lung tissue and bone differs significantly, requiring specific values ​​to be defined separately to ensure the parameters conform to the tissue's physiological characteristics. The second method iterates through each spatial point in the three-dimensional parameter matrix, determining its anatomical affiliation (belonging to the lungs, heart, or bones, etc.) through spatial coordinate matching (e.g., comparing point coordinates with the coordinates of the anatomical divisions in the image), and assigning the corresponding electrical characteristic parameters to that point. This completes the parameter filling of the entire thoracic cavity space, ultimately generating a single-frame electrical characteristic parameter model (corresponding to a specific respiratory phase, such as end of inspiration or end of expiration). The dynamic point cloud data (including the spatial morphology of organs at each time phase) within the respiratory cycle is precisely spatiotemporally aligned with the single-frame electrical characteristic parameter model of the corresponding time phase. This ensures that within the same respiratory phase, each unit of the dynamic point cloud can be matched with the electrical characteristic parameter of the corresponding spatial location in the parameter model. After integrating the alignment results of all time phases, a digital twin thoracic cavity model covering the entire respiratory cycle and possessing both dynamic anatomical morphology and precise electrical characteristic parameters is finally constructed.

[0147] It should be noted that, to improve the generalization ability of subsequent algorithms (such as the EIT image reconstruction algorithm), multi-dimensional constraints can be applied to the three-dimensional conductivity matrix during the training phase to increase the diversity of case models. Specifically, this is achieved through the following methods: Physiological constraints: Apply principal strain constraints to the lung basal region of the matrix region corresponding to lung tissue (simulating the physiological characteristic of minimal deformation of the lung basal region due to gravity during real breathing) to prevent excessive deformation in the model that does not conform to physiological laws. Pathological model construction: Simulate common lung pathological states in the parameter matrix, such as mapping the actual inflammation / lesion areas of clinical patients (the location and size of lesions diagnosed through imaging) to the three-dimensional parameter matrix. After modifying the parameters of the corresponding regions, generate individualized pathological models and supplement them to the case database used for algorithm training, improving the algorithm's adaptability to real pathological scenarios.

[0148] like Figures 3a-3g As shown, the simulations are 1 for atelectasis, 2 for atelectasis, 3 for cysts, 1 for pneumonia, 2 for pneumonia, 1 for ellipsoidal cavity, and a case model with the inflammatory area of ​​a real patient. The green part in the figure represents the area with electrical properties that are consistent with healthy tissue, and the red part represents the lesion area with electrical properties that have been pathologically corrected. The figure intuitively presents the differences in parameter distribution under different pathological conditions.

[0149] In one embodiment of this application, the step of determining the anatomical affiliation of each spatial point in the three-dimensional parameter matrix and assigning corresponding electrical characteristic parameters to generate a single-frame electrical characteristic parameter model includes:

[0150] The spatial points to be measured are extracted sequentially from the three-dimensional parameter matrix according to a preset order;

[0151] Based on the spatial location information of the test point, the contour information of the height section of each target structure at the test point is extracted;

[0152] Determine whether the spatial point to be measured is within the contour range of any target structure;

[0153] If the space point to be tested is within the contour range of any target structure, then the space point to be tested is assigned the electrical characteristic parameters of the corresponding target structure;

[0154] If the spatial point to be tested is located within the pleural cavity, then the corresponding electrical characteristic parameters of the pleural cavity are assigned.

[0155] If the spatial point to be measured is outside the thoracic cavity, then a preset reference parameter is assigned;

[0156] Once all spatial points to be tested have been assigned values, the single-frame electrical characteristic parameter model is generated.

[0157] Optionally, a two-parameter three-dimensional parameter matrix system can be established based on the three-dimensional spatial range of the thoracic cavity, i.e., the anatomical dimensions of the thoracic cavity. Its spatial resolution can be M×N×L voxels (length×width×height); the parameter types can be the conductivity matrix (σ) and the relative permittivity matrix. The coordinate system can be a Cartesian coordinate system, spatially registered with the original acquired chest medical image data. Then, referring to a pre-defined bio-battery database (such as the ITIS Foundation Bioelectromagnetic Database), electromagnetic properties can be classified according to anatomical structures, assigning different values ​​of conductivity and relative permittivity to different anatomical regions.

[0158] Specifically, a three-dimensional spatial occupancy matrix can be established using the xyz range of the smallest cuboid just enclosed by each frame of the thoracic cavity. Then, the spatial points to be tested are extracted sequentially from the M×N×L matrix according to a preset order (e.g., voxel traversal order from the x-axis to the y-axis, and from the bottom layer to the top layer). Based on the Z coordinate values ​​of the extracted spatial points, the two-dimensional contour information of each target structure in the Z-height plane is extracted (the contour lattice is parallel to the xOy plane). Then, a layered judgment is performed using the two-dimensional ray method (the number of intersections between rays emitted from points within the contour line and the contour line must be odd, otherwise even): First, it is determined whether the point is within the thoracic cavity contour line. If it is, it is further determined whether it belongs to the contour range of a certain target structure (lung, heart, bone, etc.). Ray detection is performed on the contours of each structure sequentially to determine the specific anatomical affiliation of the point to be tested (e.g., left lung, right lung, myocardium, etc.). If it is not within the thoracic cavity contour line, the conductivity and dielectric constant are both assigned to 0. If it is inside the thoracic cavity but not within any target structure, it is determined to be inside the thoracic cavity, and the corresponding baseline value of the thoracic cavity can be assigned.

[0159] In one embodiment of this application, the deployment of virtual sensing components that dynamically deform with the chest cavity in the digital twin chest cavity model to form a virtual sensing array includes:

[0160] Based on the spatial morphological features of the thoracic cavity in the dynamic point cloud data, the spatial distribution position of the virtual sensing components is determined according to a preset distribution rule.

[0161] The dynamic point cloud of the thoracic cavity at the corresponding time phase is cut by cross-section, and the surface contour information of the thoracic cavity at each spatial distribution location is extracted.

[0162] Based on the chest cavity surface contour information, the deployment pose of the virtual sensing component is determined;

[0163] Based on the deployment pose of the virtual sensing component, a geometric model of the virtual sensing component that fits the surface of the chest cavity is constructed to form a single-frame virtual sensing array.

[0164] After the dynamic point cloud of all phases within the respiratory cycle has generated a corresponding single-frame virtual sensor array, the single-frame virtual sensor arrays are combined according to the respiratory sequence into a virtual sensor array that dynamically changes synchronously with the movement of the chest cavity.

[0165] It should be noted that, because the chest cavity constantly expands and contracts during respiration, the electrode pads attached to the chest cavity also move accordingly. To closely resemble reality, a dynamic virtual sensing component can be created based on the surface morphology and movement trend of the chest cavity in each frame. This is achieved by driving the virtual sensing array to move synchronously through a surface point cloud displacement transfer mechanism, as described in the chest cavity model above. This allows the virtual sensing component to not only fit snugly against the chest cavity but also to move with it.

[0166] Specifically, based on the thoracic cavity spatial morphological features presented by dynamic point cloud data (such as the size and curvature of the thoracic cavity at different respiratory phases), combined with preset monitoring requirements (such as covering key ventilation areas of the lungs and being evenly distributed across the thoracic cavity), and following preset distribution rules (such as layered arrangement and uniform circumferential distribution), the initial distribution position of each virtual sensing component in the thoracic cavity space is determined. For a specific phase within the respiratory cycle, the dynamic point cloud of the thoracic cavity at that phase is cut with a plane parallel to the corresponding plane to obtain a thoracic cavity cross-section at the corresponding height. The contour data of the thoracic cavity surface at that phase (such as the coordinates and curvature of the contour line) is extracted from the cross-section. Based on the extracted thoracic cavity surface contour information, the contour normal direction (perpendicular to the contour tangent and pointing towards the outer side of the thoracic cavity) at the distribution position of each virtual sensing component is calculated. To meet the requirements of monitoring signal acquisition (e.g., electrodes need to be in contact with the body surface to ensure signal quality), the deployment angle (e.g., aligned with the normal direction) and spatial orientation of the virtual sensing components are determined. Based on the determined deployment orientation, a geometric model is constructed for each virtual sensing component in the digital twin chest cavity model (e.g., multiple cylindrical models are created in COMSOL, these cylinders are positioned and oriented so that their sides are tangent to the chest cavity contour at that point, and these cylinders are processed through COMSOL geometric operations (e.g., Boolean operations, interpolation, etc.) to finally generate a single-layer electrode array geometric model that fits the chest cavity morphology of that frame). By adjusting the position and angle of the model, it is made to completely fit the chest cavity surface of the corresponding time phase; the geometric models of all virtual sensing components are integrated to form a single-frame virtual sensing array corresponding to a certain respiratory phase. Repeat the above steps to generate a corresponding single-frame virtual sensor array for each phase of the respiratory cycle (such as the beginning of inhalation, the end of inhalation, the beginning of exhalation, and the end of exhalation). Finally, connect all the single-frame arrays in series according to the respiratory sequence (the complete process from inhalation to exhalation) to form a virtual sensor array that moves and adjusts its posture synchronously with the dynamic deformation of the chest cavity, ensuring that the sensing components always fit the chest cavity and monitor stably throughout the entire respiratory process.

[0167] It should be noted that when performing virtual physiological signal calculations in COMSOL, it is necessary to identify the geometric entity number of each virtual sensing component model and grounding point. Therefore, the functions provided by COMSOL's MATLAB interface can be used, with the previously calculated electrode deployment point coordinates as input. This allows you to query and retrieve the number of the model's geometric entity (point, edge, boundary, or domain) corresponding to that location. For the grounding point number, you can first calculate the point with the lowest Z-coordinate on the geometric boundary of the chest cavity model in the current frame. Then, using the same COMSOL and MATLAB interface functions, input the coordinates of this lowest point, query and retrieve its corresponding geometric entity number, and set this number as the grounding point of the model in that frame.

[0168] In one embodiment of this application, determining the deployment pose of the virtual sensing component based on the chest cavity surface contour information includes:

[0169] On the contour line corresponding to the contour information of the thoracic cavity surface, uniform sampling is performed from the starting point along a preset direction;

[0170] A sampling point is selected at every preset arc length as the reference position of the virtual sensing component;

[0171] Calculate the normal direction of the contour line corresponding to the reference position of each virtual sensing component, and determine the angle corresponding to the normal direction as the deployment angle of the virtual sensing component;

[0172] The virtual sensing component is moved a preset distance away from the center of the chest cavity along the normal direction to obtain the final deployment position.

[0173] Optionally, on the closed contour line corresponding to the chest cavity surface contour information, first determine the sampling starting point (e.g., the rightmost point of the contour line, ensuring consistent sampling references across different time phases), and then uniformly sample along a preset direction (e.g., counterclockwise, to avoid chaotic sampling order). The sampling density needs to match the preset number of virtual sensing components to ensure uniform component distribution and coverage of key monitoring areas. Then, target points can be selected from the sampling points according to a preset arc length (e.g., 5mm arc lengths per interval, calculated based on the contour perimeter and the number of components), serving as the reference positions for the virtual sensing components. For each reference position, the normal direction of the contour line is calculated using a geometric algorithm (e.g., solving for the tangent direction of the contour line at that point, and then taking the perpendicular direction). The normal direction needs to point outwards away from the center of the chest cavity (ensuring the components are deployed on the chest cavity surface), and the angle corresponding to this normal direction (e.g., the angle with the X-axis of the coordinate system) is determined as the deployment angle of the virtual sensing components. Finally, along the determined normal direction, the reference position is moved outward (away from the center of the chest cavity) by a preset distance (e.g., 2mm, simulating the thickness of the real sensing component) to obtain the final deployment position of the virtual sensing component. This distance must be fixed and less than the component's geometric dimensions (e.g., 1 / 2 the diameter of the virtual sensing component) to avoid the component detaching from the chest cavity surface or being over-embedded.

[0174] For example, the virtual sensing component layer number (LayerNum) and the height range of a single frame of the thoracic cavity on the Z-axis can be determined according to the preset virtual sensing component layer number (LayerNum). ),Will The system is approximately divided into (LayerNum+1) equal division points. The Z values ​​corresponding to these division points (e.g., when LayerNum=3, Z_min~Z_max are divided into 4 equal division points, and the middle 3 points are taken as the Z coordinates of the 3-layer component) are the Z coordinates of the electrode sheets in each layer. For each layer (e.g., For example: using The thoracic cavity point cloud data of the frame is cut along the horizontal plane (parallel to the xOy plane) corresponding to the value, obtaining the cross-sectional contour line of the thoracic cavity at that height. For example, a two-dimensional coordinate system (X'Y' plane) is established with the geometric center point of a contour line along a Z-axis section (parallel to the XoY plane) as the origin. Then, any layer section can be arbitrarily selected (e.g., ...). Starting from an arbitrary point on the contour line, samples are uniformly taken counterclockwise along the contour line. At equal intervals of contour line arc length, a point is selected as the reference position for the virtual sensing component. The normal direction of the contour line at each reference position is calculated, and its angle is recorded. (That is, the deployment angle of the electrode pads). Along this normal direction, move a fixed distance outward from the reference point (away from the center of the chest cavity) to obtain the final deployment point coordinates of the electrodes. Repeat this process for each layer, generating a total of [number] layers along the contour line. such virtual sensing component deployment points and their corresponding deployment angles .

[0175] Furthermore, based on the generated deployment point coordinates and placement angle In COMSOL, multiple cylindrical models are created. The axes of the cylinders must be aligned with the deployment angle (normal direction), and their sides must be tangent to the chest cavity contour at the reference position to ensure the models fit the chest cavity surface. The electrode sheet model shape is optimized using COMSOL's Boolean operations (such as performing an "intersection" operation between the cylinders and the chest cavity surface model to remove portions extending beyond the chest cavity surface) and difference operations (removing possible overlapping areas between models). After integrating all layers of electrode sheet models, a multi-layered virtual sensor array geometric model that fits the chest cavity shape of that frame is generated. The above layered deployment and modeling process is repeated for the dynamic point cloud of the chest cavity at each phase of the respiratory cycle to generate a single-frame virtual sensor array for the corresponding phase. Subsequently, through the surface point cloud displacement transfer mechanism, the array at each phase is updated synchronously with the chest cavity deformation to achieve a dynamically adapted virtual sensor array.

[0176] In one embodiment of this application, the step of simulating and monitoring the dynamic respiratory process of the target object based on the virtual sensor array to generate virtual physiological signals includes:

[0177] Electrical characteristic parameters are imported into a digital twin thoracic cavity model including the virtual sensor array to construct a virtual monitoring scenario with electromagnetic response properties.

[0178] According to the preset traversal sequence, different virtual sensing components in the virtual sensing array are selected as excitation sources in sequence.

[0179] Based on the preset excitation rules, an excitation signal is applied to the excitation source, and the voltage signal is acquired using the remaining virtual sensing components;

[0180] After the traversal is completed, all the collected voltage signals are used as virtual physiological signals covering the respiratory cycle. The virtual physiological signals include dynamic change components related to thoracic respiratory movements, and the number of virtual physiological signals is adapted to the configuration parameters of the virtual sensing components.

[0181] It should be noted that this is based on the electrophysiological coupling effect during lung respiration, specifically: when a safe amplitude (typically 3mA and -3mA) AC excitation current is injected through a virtual sensor array attached to the chest, During inhalation, the regional dynamic changes in conductivity caused by alveolar ventilation (the decrease in conductivity due to gas filling during inhalation) alter the distribution of current in the body, thereby generating a weak voltage signal on the boundary electrode. and its respiratory modulation fluctuations V. Therefore, virtual physiological signals can be obtained in the following way:

[0182] First, the previously generated three-dimensional parameter matrices (three-dimensional conductivity matrix and three-dimensional dielectric constant matrix) are imported into the thoracic cavity geometry model with added electrodes using COMSOL's interpolation function. This operation is equivalent to assigning realistic electromagnetic properties to different anatomical structures within the model (such as lungs, heart, bones, etc.). Then, using the standard method of differential excitation of adjacent electrode pairs and measurement of voltage differences for the remaining electrode pairs, virtual physiological signals, i.e., voltage measurements (potential difference data), are generated.

[0183] Specifically, a first virtual sensing component is identified as the first excitation source, and a first excitation current is applied. A second virtual sensing component is identified as the second excitation source, and a second excitation current is applied. Then, the remaining virtual sensing components are paired in pairs (typically, electrodes that are physically adjacent or close together form measurement pairs), and the potential difference (voltage) between each pair of electrodes is measured. After completion, the second virtual sensing component is used as the next first excitation source, and the third virtual sensing component is used as the third excitation source. The process of pairing the remaining virtual sensing components in pairs and measuring the potential difference (voltage) between each pair of electrodes is repeated until all possible starting excitation positions have been traversed. For example, the loop can start from the first virtual sensing component on the front of the chest cavity: a +3mA excitation current (I=3mA) is applied to virtual sensing component 1; a -3mA excitation current (I=-3mA) is applied to the adjacent virtual sensing component 2. At this point, all the remaining electrode plates (except for the excited plates 1 and 2) are paired up in pairs (usually, the electrodes that are physically adjacent or close to each other form a measurement pair), and the potential difference (voltage) between each pair of electrodes is measured.

[0184] Then, apply a +3mA current (I=3mA) to virtual sensing component 2; apply a -3mA current (I=-3mA) to virtual sensing component 3. Measure the voltage difference between the paired electrodes of all remaining electrode plates (except 2 and 3) again. Continue this process in this manner (current excitation electrode pair is [n, n+1], apply [+3mA, -3mA], measure the voltage between all other adjacent electrode pairs) until all possible initial excitation positions have been traversed (considering symmetry).

[0185] For a single-layer array containing multiple (e.g., 16) virtual sensing components, after cycling through the above excitation and measurement modes, 104 independent voltage measurements (potential difference data) will be obtained. For multi-layer (n-layer electrode sheets), since auxiliary cross-layer excitation is also added, [104*n+120*(n-1)] independent voltage measurements (potential difference data) will be obtained, which serve as virtual physiological signals.

[0186] It should be noted that during the training phase, Gaussian noise is introduced into the voltage signal calculated by COMSOL, with the noise standard deviation set to 1%~5% of the signal amplitude (simulating the acquisition error caused by circuit interference and contact impedance changes in clinical EIT equipment). By adding noise, the robustness of subsequent algorithms (such as EIT image reconstruction algorithms) is improved, making them more adaptable to the fluctuation characteristics of real clinical data.

[0187] In one embodiment of this application, the step of obtaining a dynamic reconstructed image reflecting respiratory motion by processing the virtual physiological signal includes:

[0188] Based on the voltage signal, the voltage change is calculated and normalized.

[0189] Based on the normalized voltage change, a sensitive field matrix is ​​constructed to complete the forward problem modeling. The sensitive field matrix is ​​used to reflect the physical mapping relationship between conductivity distribution and boundary voltage.

[0190] The distribution of intrathoracic conductivity variation was obtained by solving the inverse problem using a regularization algorithm.

[0191] The conductivity variation distribution is post-processed and optimized to generate the dynamic reconstructed image reflecting respiratory motion.

[0192] Optionally, the change in the boundary voltage (voltage measurement value) generated above can be used as the core input data for image reconstruction. The relative impedance distribution map of the cross section can be obtained by training a unified inverse problem solving algorithm (such as Tikhonov regularization) or an AI algorithm, which can intuitively reflect the spatiotemporal heterogeneity of lung ventilation.

[0193] Taking the inverse problem solving algorithm as an example, the image reconstruction process is explained in detail. The specific implementation process is as follows: First, data preprocessing is performed, calculating the boundary voltage change and normalizing it to eliminate measurement system errors. Next, a sensitive field matrix J containing complete electromagnetic field information such as geometry, electrode position, and background conductivity is constructed using the finite element method, completing the forward problem modeling and accurately describing the physical relationship between conductivity distribution and boundary voltage. Then, the inverse problem solving stage is entered, using a regularization method to solve the optimization problem min||JΔσ-ΔV||²+λ²||LΔσ||² (where J is the sensitive field matrix, Δσ is the conductivity change distribution, ΔV is the boundary voltage change, λ is the regularization parameter, and L is the regularization matrix). The regularization parameter λ is adaptively determined using the L-curve method or generalized cross-validation to balance the data fit and solution smoothness. Finally, post-processing optimization is performed, applying image filtering and dynamic range adjustment to enhance the reconstructed image quality and eliminate artifacts introduced by numerical calculations, ultimately obtaining a reconstructed image reflecting the changes in conductivity within the body. Figure 4 As shown, this diagram illustrates the reconstruction of multiple sets of raw thoracic medical image data with lung health (Functional Imaging Test, FIT) and inflammation in both lungs (PNEU1 / PNEU2) using the Tikhonov regularization algorithm. The diagram shows the reconstruction of images at different phases within the respiratory cycle (such as end of inspiration, 20% of expiration, 40% of expiration, 60% of expiration, 80% of expiration, and 100% of expiration) from the raw point cloud.

[0194] In this embodiment, a three-dimensional geometric model of the thoracic cavity is constructed using point cloud technology. Utilizing single-frame point cloud data at the end of inspiration, and combining it with respiratory motion patterns, multi-frame dynamic expiratory data is generated to fully capture the three-dimensional deformation features of the entire lung. Combined with real respiratory motion parameters and trajectory-driven simulation, the continuous morphological changes of the lung from the end of inspiration to the end of expiration within the respiratory cycle can be accurately reproduced, achieving continuous simulation of the entire respiratory process. Furthermore, the spatiotemporal resolution is enhanced through point cloud data. Combined with multiple pathological models and respiratory pattern control, a standardized EIT dataset covering anatomical variations, dynamic deformations, and pathological features is generated, effectively solving the problem of missing three-dimensional dynamic data in traditional techniques. The digital twin thoracic cavity model is constructed based on individual point cloud data, fully... The virtual sensing components are deployed synchronously with the dynamic deformation of the thoracic cavity, adapting to the anatomical features of the target object. This avoids the safety risks and signal distortion problems caused by physical electrodes, and ensures that the generated virtual physiological signals are highly consistent with real signals in terms of amplitude and dynamic fluctuation patterns. By combining point cloud simulation data with clinically measured EIT signals, the dynamic calibration capability of the digital twin system is optimized, allowing the signal processing and imaging processes to fully coordinate with the spatiotemporal characteristics of the point cloud dynamic model. This effectively suppresses image artifacts and enhances detail recognition. The resulting dynamically reconstructed images can accurately reflect the differences in ventilatory function in lung regions during the respiratory cycle, enabling non-invasive and continuous respiratory monitoring and imaging without relying on a large amount of clinical measured data and high-frequency physical examinations.

[0195] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0196] In one embodiment, a point cloud-driven dynamic reconstruction and respiratory monitoring system for all organs of the thoracic cavity is provided, which corresponds one-to-one with the point cloud-driven dynamic reconstruction and respiratory monitoring method for all organs of the thoracic cavity described in the above embodiments. For example... Figure 5 As shown, the point cloud-driven dynamic reconstruction and respiratory monitoring system for all organs of the thoracic cavity includes a static point cloud data generation unit 10, a dynamic point cloud data generation unit 20, a digital twin thoracic cavity model construction unit 30, a virtual sensor array generation unit 40, a virtual physiological signal generation unit 50, and a dynamic image reconstruction unit 60. Detailed descriptions of each functional module are as follows:

[0197] The static point cloud data generation unit 10 is used to construct a three-dimensional geometric model corresponding to each target structure in the thoracic cavity based on the chest medical imaging data of the target object, and convert the three-dimensional geometric model into static point cloud data.

[0198] The dynamic point cloud data generation unit 20 is used to drive the static point cloud data to perform dynamic deformation simulation within the respiratory cycle based on the respiratory motion law of the target object, so as to obtain dynamic point cloud data synchronized with breathing.

[0199] The digital twin thoracic cavity model construction unit 30 is used to spatiotemporally correlate the dynamic point cloud data with electrical characteristic parameters to construct a digital twin thoracic cavity model.

[0200] The virtual sensor array generation unit 40 is used to deploy virtual sensor components that dynamically deform with the chest cavity in the digital twin chest cavity model to form a virtual sensor array.

[0201] The virtual physiological signal generation unit 50 is used to simulate and monitor the dynamic respiratory process of the target object based on the virtual sensor array, and generate virtual physiological signals.

[0202] The dynamic image reconstruction unit 60 is used to obtain a dynamic reconstructed image reflecting respiratory motion by performing signal processing on the virtual physiological signal.

[0203] In one embodiment of this application, the static point cloud data generation unit 10 further includes:

[0204] Based on the chest medical imaging data, identify and segment the target structures within the thoracic cavity;

[0205] Three-dimensional modeling of each target structure within the thoracic cavity is performed, and the model surface is optimized through smoothing to obtain the three-dimensional geometric model corresponding to each target structure.

[0206] Discretize the surface of the three-dimensional geometric model corresponding to each target structure, extract the spatial location information of each sampling point, and generate static point cloud data corresponding to each target structure.

[0207] In one embodiment of this application, the dynamic point cloud data generation unit 20 is further configured to:

[0208] Based on the respiratory motion characteristics of each target structure during the respiratory cycle, the dynamic driving parameters corresponding to each target structure are determined.

[0209] Based on the dynamic driving parameters, the static point cloud data is dynamically adjusted at each phase of the respiratory cycle to generate a dynamic point cloud corresponding to each phase.

[0210] The dynamic point clouds of each time phase are integrated into a complete dynamic sequence according to the breathing time sequence to obtain the dynamic point cloud data.

[0211] In one embodiment of this application, the digital twin thoracic cavity model construction unit 30 is further used for:

[0212] Based on the aforementioned chest medical imaging data, the extent of the thoracic cavity was determined;

[0213] Based on the aforementioned thoracic cavity space range, a three-dimensional parameter matrix including electrical characteristic parameters is established;

[0214] Determine the corresponding electrical characteristic parameters according to the anatomical divisions of each target structure within the thoracic cavity;

[0215] Determine the anatomical affiliation of each spatial point to be tested in the three-dimensional parameter matrix and assign corresponding electrical characteristic parameters to generate a single-frame electrical characteristic parameter model.

[0216] By spatiotemporally aligning the dynamic point cloud data of each respiratory phase with the corresponding single-frame electrical characteristic parameter model, a digital twin thoracic cavity model covering the entire respiratory cycle is constructed.

[0217] In one embodiment of this application, the digital twin thoracic cavity model construction unit 30 is further used for:

[0218] The spatial points to be measured are extracted sequentially from the three-dimensional parameter matrix according to a preset order;

[0219] Based on the spatial location information of the test point, the contour information of the height section of each target structure at the test point is extracted;

[0220] Determine whether the spatial point to be measured is within the contour range of any target structure;

[0221] If the space point to be tested is within the contour range of any target structure, then the space point to be tested is assigned the electrical characteristic parameters of the corresponding target structure;

[0222] If the spatial point to be tested is located within the pleural cavity, then the corresponding electrical characteristic parameters of the pleural cavity are assigned.

[0223] If the spatial point to be measured is outside the thoracic cavity, then a preset reference parameter is assigned;

[0224] Once all spatial points to be tested have been assigned values, the single-frame electrical characteristic parameter model is generated.

[0225] In one embodiment of this application, the virtual sensor array generation unit 40 is further configured to:

[0226] Based on the spatial morphological features of the thoracic cavity in the dynamic point cloud data, the spatial distribution position of the virtual sensing components is determined according to a preset distribution rule.

[0227] The dynamic point cloud of the thoracic cavity at the corresponding time phase is cut by cross-section, and the surface contour information of the thoracic cavity at each spatial distribution location is extracted.

[0228] Based on the chest cavity surface contour information, the deployment pose of the virtual sensing component is determined;

[0229] Based on the deployment pose of the virtual sensing component, a geometric model of the virtual sensing component that fits the surface of the chest cavity is constructed to form a single-frame virtual sensing array.

[0230] After the dynamic point cloud of all phases within the respiratory cycle has generated a corresponding single-frame virtual sensor array, the single-frame virtual sensor arrays are combined according to the respiratory sequence into a virtual sensor array that dynamically changes synchronously with the movement of the chest cavity.

[0231] In one embodiment of this application, the virtual physiological signal generation unit 50 is further configured to:

[0232] On the contour line corresponding to the contour information of the thoracic cavity surface, uniform sampling is performed from the starting point along a preset direction;

[0233] A sampling point is selected at every preset arc length as the reference position of the virtual sensing component;

[0234] Calculate the normal direction of the contour line corresponding to the reference position of each virtual sensing component, and determine the angle corresponding to the normal direction as the deployment angle of the virtual sensing component;

[0235] The virtual sensing component is moved a preset distance away from the center of the chest cavity along the normal direction to obtain the final deployment position.

[0236] In one embodiment of this application, the virtual physiological signal generation unit 50 is further configured to:

[0237] Electrical characteristic parameters are imported into a digital twin thoracic cavity model including the virtual sensor array to construct a virtual monitoring scenario with electromagnetic response properties.

[0238] According to the preset traversal sequence, different virtual sensing components in the virtual sensing array are selected as excitation sources in sequence.

[0239] Based on the preset excitation rules, an excitation signal is applied to the excitation source, and the voltage signal is acquired using the remaining virtual sensing components;

[0240] After the traversal is completed, all the collected voltage signals are used as virtual physiological signals covering the respiratory cycle. The virtual physiological signals include dynamic change components related to thoracic respiratory movements, and the number of virtual physiological signals is adapted to the configuration parameters of the virtual sensing components.

[0241] In one embodiment of this application, the dynamic image reconstruction unit 60 is further configured to:

[0242] Based on the voltage signal, the voltage change is calculated and normalized.

[0243] Based on the normalized voltage change, a sensitive field matrix is ​​constructed to complete the forward problem modeling. The sensitive field matrix is ​​used to reflect the physical mapping relationship between conductivity distribution and boundary voltage.

[0244] The distribution of intrathoracic conductivity variation was obtained by solving the inverse problem using a regularization algorithm.

[0245] The conductivity variation distribution is post-processed and optimized to generate the dynamic reconstructed image reflecting respiratory motion.

[0246] In this embodiment, a three-dimensional geometric model of the thoracic cavity is constructed using point cloud technology. Utilizing single-frame point cloud data at the end of inspiration, and combining it with respiratory motion patterns, multi-frame dynamic expiratory data is generated to fully capture the three-dimensional deformation features of the entire lung. Combined with real respiratory motion parameters and trajectory-driven simulation, the continuous morphological changes of the lung from the end of inspiration to the end of expiration within the respiratory cycle can be accurately reproduced, achieving continuous simulation of the entire respiratory process. Furthermore, the spatiotemporal resolution is enhanced through point cloud data. Combined with multiple pathological models and respiratory pattern control, a standardized EIT dataset covering anatomical variations, dynamic deformations, and pathological features is generated, effectively solving the problem of missing three-dimensional dynamic data in traditional techniques. The digital twin thoracic cavity model is constructed based on individual point cloud data, fully... The virtual sensing components are deployed synchronously with the dynamic deformation of the thoracic cavity, adapting to the anatomical features of the target object. This avoids the safety risks and signal distortion problems caused by physical electrodes, and ensures that the generated virtual physiological signals are highly consistent with real signals in terms of amplitude and dynamic fluctuation patterns. By combining point cloud simulation data with clinically measured EIT signals, the dynamic calibration capability of the digital twin system is optimized, allowing the signal processing and imaging processes to fully coordinate with the spatiotemporal characteristics of the point cloud dynamic model. This effectively suppresses image artifacts and enhances detail recognition. The resulting dynamically reconstructed images can accurately reflect the differences in ventilatory function in lung regions during the respiratory cycle, enabling non-invasive and continuous respiratory monitoring and imaging without relying on a large amount of clinical measured data and high-frequency physical examinations.

[0247] Specific limitations regarding the point cloud-driven dynamic reconstruction and respiratory monitoring system for all thoracic organs can be found in the limitations of the point cloud-driven dynamic reconstruction and respiratory monitoring method for all thoracic organs described above, and will not be repeated here. Each module in the aforementioned point cloud-driven dynamic reconstruction and respiratory monitoring system for all thoracic organs can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0248] In one embodiment, a computer device is provided, which may be a terminal device, and its internal structure diagram may be as follows: Figure 6As shown, the computer device includes a processor, memory, and network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a readable storage medium storing computer-readable instructions. The network interface communicates with external terminals via a network connection. When executed by the processor, the computer-readable instructions implement a point cloud-driven method for dynamic reconstruction of all thoracic organs and respiratory monitoring. The readable storage medium provided in this embodiment includes both non-volatile and volatile readable storage media.

[0249] In this application embodiment, a computer device is provided, including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor. When the processor executes the computer-readable instructions, it implements the steps of the point cloud-driven dynamic reconstruction of all organs in the thoracic cavity and respiratory monitoring method described above.

[0250] In this embodiment of the application, a readable storage medium is provided, which stores computer-readable instructions. When the computer-readable instructions are executed by a processor, they implement the steps of the point cloud-driven dynamic reconstruction of the whole thoracic organs and respiratory monitoring method described above.

[0251] 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 instructing related hardware with computer-readable instructions. These computer-readable instructions can be stored in a non-volatile readable storage medium or a volatile readable storage medium. When executed, these computer-readable instructions can include the processes of the embodiments of the methods described above. 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.

[0252] 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.

[0253] The above 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 point cloud-driven method for dynamic reconstruction of all organs in the thoracic cavity and respiratory monitoring, characterized in that, The method includes: Based on the chest medical imaging data of the target object, a three-dimensional geometric model corresponding to each target structure in the thoracic cavity is constructed, and the three-dimensional geometric model is converted into static point cloud data; Based on the respiratory motion pattern of the target object, the static point cloud data is driven to perform dynamic deformation simulation within the respiratory cycle, including: determining the corresponding dynamic driving parameters based on the specific respiratory motion characteristics of each target structure in the thoracic cavity within the respiratory cycle; driving the static point cloud data of the whole organs in the thoracic cavity to perform dynamic deformation simulation within the respiratory cycle through the dynamic driving parameters; dynamically adjusting the static point cloud data at each phase of the respiratory cycle and integrating it according to the respiratory time sequence to obtain dynamic point cloud data synchronized with the respiratory time phase. The electrical characteristic parameters corresponding to each target structure are spatiotemporally associated with the respiratory phase, organ location and morphological information of the dynamic point cloud data, so that each point cloud unit in the thoracic cavity under a specific respiratory phase is matched with the corresponding temporal electrical characteristic parameters of the corresponding tissue, thus constructing a digital twin thoracic cavity model. Virtual sensing components that dynamically deform with the chest cavity are deployed in the digital twin chest cavity model to form a virtual sensing array; The dynamic respiratory process of the target object is simulated and monitored based on the virtual sensor array to generate virtual physiological signals. By processing the virtual physiological signals, a dynamic reconstructed image reflecting respiratory movements is obtained.

2. The point cloud-driven dynamic reconstruction and respiratory monitoring method for all organs of the thoracic cavity as described in claim 1, characterized in that, The method involves constructing a three-dimensional geometric model corresponding to each target structure within the thoracic cavity based on the chest medical imaging data of the target object, and converting the three-dimensional geometric model into static point cloud data, including: Based on the chest medical imaging data, identify and segment the target structures within the thoracic cavity; Three-dimensional modeling of each target structure within the thoracic cavity is performed, and the model surface is optimized through smoothing to obtain the three-dimensional geometric model corresponding to each target structure. Discretize the surface of the three-dimensional geometric model corresponding to each target structure, extract the spatial location information of each sampling point, and generate static point cloud data corresponding to each target structure.

3. The point cloud-driven dynamic reconstruction and respiratory monitoring method for all organs of the thoracic cavity as described in claim 1, characterized in that, The step of spatiotemporally and synchronously associating the dynamic point cloud data with electrical characteristic parameters to construct a digital twin thoracic cavity model includes: Based on the aforementioned chest medical imaging data, the extent of the thoracic cavity was determined; Based on the aforementioned thoracic cavity space range, a three-dimensional parameter matrix including electrical characteristic parameters is established; Determine the corresponding electrical characteristic parameters according to the anatomical divisions of each target structure within the thoracic cavity; Determine the anatomical affiliation of each spatial point to be tested in the three-dimensional parameter matrix and assign corresponding electrical characteristic parameters to generate a single-frame electrical characteristic parameter model. By spatiotemporally aligning the dynamic point cloud data of each respiratory phase with the corresponding single-frame electrical characteristic parameter model, a digital twin thoracic cavity model covering the entire respiratory cycle is constructed.

4. The point cloud-driven dynamic reconstruction and respiratory monitoring method for all organs of the thoracic cavity as described in claim 3, characterized in that, The step of determining the anatomical affiliation of each spatial point in the three-dimensional parameter matrix and assigning corresponding electrical characteristic parameters to generate a single-frame electrical characteristic parameter model includes: The spatial points to be measured are extracted sequentially from the three-dimensional parameter matrix according to a preset order; Based on the spatial location information of the test point, the contour information of the height section of each target structure at the test point is extracted; Determine whether the spatial point to be measured is within the contour range of any target structure; If the space point to be tested is within the contour range of any target structure, then the space point to be tested is assigned the electrical characteristic parameters of the corresponding target structure; If the spatial point to be tested is located within the pleural cavity, then the corresponding electrical characteristic parameters of the pleural cavity are assigned. If the spatial point to be measured is outside the thoracic cavity, then a preset reference parameter is assigned; Once all spatial points to be tested have been assigned values, the single-frame electrical characteristic parameter model is generated.

5. The point cloud-driven dynamic reconstruction and respiratory monitoring method for all organs of the thoracic cavity as described in claim 1, characterized in that, The deployment of virtual sensing components that dynamically deform with the thoracic cavity in the digital twin thoracic cavity model to form a virtual sensing array includes: Based on the spatial morphological features of the thoracic cavity in the dynamic point cloud data, the spatial distribution position of the virtual sensing components is determined according to a preset distribution rule. The dynamic point cloud of the thoracic cavity at the corresponding time phase is cut by cross-section, and the surface contour information of the thoracic cavity at each spatial distribution location is extracted. Based on the chest cavity surface contour information, the deployment pose of the virtual sensing component is determined; Based on the deployment pose of the virtual sensing component, a geometric model of the virtual sensing component that fits the surface of the chest cavity is constructed to form a single-frame virtual sensing array. After the dynamic point cloud of all phases within the respiratory cycle has generated a corresponding single-frame virtual sensor array, the single-frame virtual sensor arrays are combined according to the respiratory sequence into a virtual sensor array that dynamically changes synchronously with the movement of the chest cavity.

6. The point cloud-driven dynamic reconstruction and respiratory monitoring method for all organs of the thoracic cavity as described in claim 5, characterized in that, Determining the deployment pose of the virtual sensing component based on the chest cavity surface contour information includes: On the contour line corresponding to the contour information of the thoracic cavity surface, uniform sampling is performed from the starting point along a preset direction; A sampling point is selected at every preset arc length as the reference position of the virtual sensing component; Calculate the normal direction of the contour line corresponding to the reference position of each virtual sensing component, and determine the angle corresponding to the normal direction as the deployment angle of the virtual sensing component; The virtual sensing component is moved a preset distance away from the center of the chest cavity along the normal direction to obtain the final deployment position.

7. The point cloud-driven dynamic reconstruction and respiratory monitoring method for all organs of the thoracic cavity as described in any one of claims 1-6, characterized in that, The simulation monitoring of the dynamic respiratory process of the target object based on the virtual sensor array, generating virtual physiological signals, includes: Electrical characteristic parameters are imported into a digital twin thoracic cavity model including the virtual sensor array to construct a virtual monitoring scenario with electromagnetic response properties. According to the preset traversal sequence, different virtual sensing components in the virtual sensing array are selected as excitation sources in sequence. Based on the preset excitation rules, an excitation signal is applied to the excitation source, and the voltage signal is acquired using the remaining virtual sensing components; After the traversal is completed, all the collected voltage signals are used as virtual physiological signals covering the respiratory cycle. The virtual physiological signals include dynamic change components related to thoracic respiratory movements, and the number of virtual physiological signals is adapted to the configuration parameters of the virtual sensing components.

8. The point cloud-driven dynamic reconstruction and respiratory monitoring method for all organs of the thoracic cavity as described in claim 7, characterized in that, The step of processing the virtual physiological signals to obtain a dynamic reconstructed image reflecting respiratory movements includes: Based on the voltage signal, the voltage change is calculated and normalized. Based on the normalized voltage change, a sensitive field matrix is ​​constructed to complete the forward problem modeling. The sensitive field matrix is ​​used to reflect the physical mapping relationship between conductivity distribution and boundary voltage. The distribution of intrathoracic conductivity variation was obtained by solving the inverse problem using a regularization algorithm. The conductivity variation distribution is post-processed and optimized to generate the dynamic reconstructed image reflecting respiratory motion.

9. A point cloud-driven dynamic reconstruction and respiratory monitoring system for all organs of the thoracic cavity, characterized in that, The system includes: The static point cloud data generation unit is used to construct a three-dimensional geometric model corresponding to each target structure in the thoracic cavity based on the chest medical imaging data of the target object, and convert the three-dimensional geometric model into static point cloud data. The dynamic point cloud data generation unit is used to drive the static point cloud data to perform dynamic deformation simulation within the respiratory cycle based on the respiratory motion law of the target object. This includes: determining the corresponding dynamic driving parameters based on the specific respiratory motion characteristics of each target structure in the thoracic cavity within the respiratory cycle; driving the static point cloud data of the whole organ in the thoracic cavity to perform dynamic deformation simulation within the respiratory cycle through the dynamic driving parameters; dynamically adjusting the static point cloud data at each phase of the respiratory cycle and integrating it according to the respiratory time sequence to obtain dynamic point cloud data synchronized with the respiratory phase. The digital twin thoracic cavity model construction unit is used to spatiotemporally associate the electrical characteristic parameters of each target structure with the respiratory phase, organ position and morphological information of dynamic point cloud data, so that each point cloud unit in the thoracic cavity under a specific respiratory phase is matched with the corresponding temporal electrical characteristic parameters of the corresponding tissue, thus constructing a digital twin thoracic cavity model. A virtual sensor array generation unit is used to deploy virtual sensor components that dynamically deform with the chest cavity in the digital twin chest cavity model to form a virtual sensor array. The virtual physiological signal generation unit is used to simulate and monitor the dynamic respiratory process of the target object based on the virtual sensor array, and generate virtual physiological signals. The dynamic image reconstruction unit is used to obtain a dynamic reconstructed image reflecting respiratory movements by processing the virtual physiological signals.

Images