A three-dimensional reconstruction and quantitative evaluation system for inguinal hernia defect area

Abstract

This invention provides a three-dimensional reconstruction and quantitative assessment system for inguinal hernia defects, comprising a processor, an image data acquisition unit, and an output unit. The processor acquires medical image data of the inguinal region of the same object under different abdominal pressure states. Based on a reference anatomical region whose positional changes satisfy reference conditions, a reference coordinate system is established and spatial registration is completed. In the registered images, a target tissue layer characterizing the continuity of abdominal wall tissue is identified, and an abdominal wall reference surface is established. Defect candidate information is determined by combining the structural continuity interruption information of the target tissue layer with the normal displacement component across abdominal pressure states. A closed boundary is then formed through spatial correspondence and anatomical topological constraints. Based on this, a three-dimensional model of the defect region under different abdominal pressure states is constructed, and static morphological parameters and dynamic deformation parameters are determined. Furthermore, the reconstruction uncertainty is determined based on registration error, boundary reliability, and model correspondence consistency, and feedback correction is performed to improve the reliability of defect boundary determination and quantitative assessment.

Description

Technical Field

[0001] This invention relates to the field of computer-aided analysis technology, specifically to a three-dimensional reconstruction and quantitative evaluation system for inguinal hernia defect areas. Background Technology

[0002] Inguinal hernia is a condition in which abdominal tissues or organs protrude outward through a weak or defective section of the abdominal wall in the groin region. The groin region has multiple anatomical layers, with complex spatial relationships between abdominal wall tissues, inguinal canal-related structures, blood vessels, and surrounding soft tissues. Furthermore, the morphology of the defect area changes with respiration, body position, and intra-abdominal pressure. Objectively characterizing the spatial location, boundary morphology, and changes under intra-abdominal pressure of the defect area is of practical significance for preoperative understanding of local anatomy, recording defect morphology, and comparing defect changes under different conditions. Current techniques typically employ medical imaging such as ultrasound, computed tomography (CT), or magnetic resonance imaging (MRI) to observe the groin region. Clinical or imaging analysis often relies on two-dimensional sections under a single intra-abdominal pressure state, manually selecting sections and measuring the major and minor axes or area of ​​the defect. Some methods can segment the hernia sac, hernia contents, or the groin region and generate a three-dimensional display model based on the segmentation results. However, this process still has some shortcomings, specifically: First, inguinal hernia defects are essentially local interruptions in the continuity of the target tissue layer of the abdominal wall. The outer contour of the hernia sac or hernia contents is not necessarily equivalent to the true boundary of the abdominal wall defect opening. Determining the defect area solely based on a single cross-section, a single grayscale threshold, or the contour of the hernia contents is easily affected by factors such as the direction of the cross-section, local low contrast, acoustic shadowing, and tissue adhesions. Furthermore, the soft tissues of the groin region have significant deformability. Defects that are not obvious at rest may exhibit significant local bulging or changes in opening when abdominal pressure increases. Conversely, local tissue displacement may also originate from the overall movement of the subject, respiratory movements, changes in the position of the imaging probe, or spatial deviations between different image data. If a unified spatial benchmark is not established beforehand for different abdominal pressure states, and tissue positions or model dimensions in images from different states are directly compared, it becomes difficult to distinguish between the actual changes caused by stress deformation of the defect area and the apparent changes caused by differences in acquisition position and overall pose.

[0003] Secondly, existing 3D reconstruction schemes typically focus on extracting target regions from single sets of medical images and generating surface models. Their boundary determination process mainly relies on local grayscale, texture, or segmentation probability. For areas where the abdominal wall tissue interface is discontinuous, the local tissue layer is thin, or there are strong artifacts, the segmentation results may show breaks, cross normal continuous tissue, self-intersection, or form non-closed boundaries. Simply smoothing the segmentation results or filling in holes can obtain a model with a continuous appearance, but this processing may mistake imaging defects for tissue defects or improperly fill in the real defect areas. Therefore, it is difficult to ensure that the obtained boundaries are consistent across anatomical topology and different abdominal pressure states.

[0004] Finally, existing technologies typically treat image registration, target region segmentation, 3D mesh generation, and parameter calculation as sequential, unidirectional processing steps. When there are local deviations in upstream registration or boundary determination, these deviations are passed to the 3D model and quantization results. However, the system often only outputs the current processing result and cannot return to the corresponding processing step for local correction based on the source of the error. It also cannot provide targeted supplementary acquisition prompts when the existing images are insufficient to support reliable reconstruction.

[0005] Therefore, how to comprehensively utilize the continuity information of abdominal wall tissue and the deformation information of transabdominal pressure under a unified spatial benchmark to determine the closure boundary of the defect, and how to quantify the reconstruction quality and provide feedback correction based on the quality evaluation results are the technical problems that need to be solved in the three-dimensional reconstruction and quantitative evaluation of the inguinal hernia defect area. Summary of the Invention

[0006] The purpose of this invention is to provide a three-dimensional reconstruction and quantitative evaluation system for inguinal hernia defects. To achieve the above objective, this invention employs the following technical solutions: A three-dimensional reconstruction and quantitative assessment system for inguinal hernia defect area includes a processor and an image data acquisition unit and an output unit communicatively connected to the processor. The processor is programmed to: acquire medical image data of the inguinal region of the same subject under at least two different abdominal pressure states; establish a common reference coordinate system based on the reference anatomical region whose positional changes in each image data meet the reference conditions; and register the medical image data under different abdominal pressure states to the common reference coordinate system. In the registered medical image data, the target tissue layer characterizing the continuity of the abdominal wall tissue is identified and an abdominal wall reference surface is established. Based on the structural continuity interruption information of the target tissue layer and the normal displacement components relative to the abdominal wall reference surface under different abdominal pressure states, defect candidate information is determined. The closed boundary of the defect area is determined by combining the cross-state space correspondence and anatomical topological constraints. Based on the closed boundary, construct a three-dimensional model of the defect area under different abdominal pressure states and establish the model correspondence, thereby determining the static morphological parameters and dynamic deformation parameters of the defect area. The reconstruction uncertainty is determined based on the registration error, boundary reliability, and model correspondence consistency. When the reconstruction uncertainty does not meet the preset reconstruction conditions, the processing parameters of the local correction area are updated and the three-dimensional model of the defective area is reconstructed until the preset reconstruction conditions are met. The three-dimensional model of the defective area, static morphological parameters, dynamic deformation parameters, and reconstruction uncertainty that meet the preset reconstruction conditions are output through the output unit.

[0007] Preferably, the medical imaging data includes three-dimensional volumetric data or an image sequence that can be converted into three-dimensional volumetric data according to the acquisition position relationship; the at least two different abdominal pressure states include a first abdominal pressure state and a second abdominal pressure state with abdominal pressure higher than the first abdominal pressure state; the establishment of the association between the medical imaging data and the corresponding abdominal pressure state includes: determining the abdominal pressure state identifier corresponding to each medical imaging data according to the state detection information synchronized with the medical imaging data, wherein the state detection information includes at least one of abdominal pressure detection information, respiratory phase information, body surface motion information, and intra-image tissue motion information.

[0008] Preferably, the processor is programmed to: select reference anatomical regions from the medical imaging data based on the positional changes and local deformations of anatomical structures under different abdominal pressure states; use the reference anatomical regions as spatial constraints to sequentially determine the overall spatial transformation relationship and local spatial transformation relationship between the medical imaging data under different abdominal pressure states; perform overall alignment of the medical imaging data under different abdominal pressure states based on the overall spatial transformation relationship, and correct local positional deviations caused by non-rigid tissue deformation in the groin region based on the local spatial transformation relationship.

[0009] Preferably, the processor is programmed to: determine multiple candidate locations of target tissue layers based on tissue interface features in medical image data; determine a target tissue layer from the multiple candidate locations of target tissue layers based on the continuity relationship of tissue interfaces between adjacent image locations, the spatial relationship between the candidate locations of target tissue layers and preset anatomical structures, and the correspondence between candidate locations of target tissue layers under different abdominal pressure states; perform surface fitting on the target tissue layer to obtain an abdominal wall reference surface, and determine the surface normal of each sampling location on the abdominal wall reference surface.

[0010] Preferably, the processor is programmed to: determine structural continuity interruption information based on the degree of interruption of the target tissue layer between adjacent image positions; determine the normal displacement component based on the displacement of the target tissue layer positions along the surface normal under different abdominal pressure states; determine candidate defect points based on the joint distribution of structural continuity interruption information and normal displacement components, and determine boundary anchor points that repeatedly appear between adjacent image positions and satisfy anatomical topological constraints from the candidate defect points; determine boundary connection segments between adjacent boundary anchor points that simultaneously satisfy image matching constraints, boundary continuity constraints, and cross-abdominal pressure state corresponding constraints, and form a closed boundary by the boundary anchor points and boundary connection segments.

[0011] Preferably, the processor is programmed to: perform meshing processing on the target tissue layer within the closed boundary to obtain a three-dimensional model of the defect region including mesh vertices and mesh patches; adjust the mesh sampling density at corresponding positions according to the local curvature of the target tissue layer and the spatial resolution of the medical image data; perform surface smoothing on the three-dimensional model of the defect region while keeping the spatial position and topological relationship of the closed boundary unchanged; and determine static morphological parameters based on the mesh vertices and mesh patches, wherein the static morphological parameters include at least two of the following: defect area, defect perimeter, maximum diameter, minor diameter intersecting the maximum diameter, equivalent diameter, surface curvature, shape irregularity, and spatial orientation.

[0012] Preferably, the processor is programmed to: establish a correspondence between mesh vertices under different abdominal pressure states based on the common reference coordinate system and boundary anchor points on the closed boundary; determine local displacement information based on the position changes of corresponding mesh vertices under different abdominal pressure states; and determine dynamic deformation parameters based on the changes in defect area, defect perimeter, and local displacement information under different abdominal pressure states. The dynamic deformation parameters include at least two of the following: area change rate, perimeter change rate, normal bulge, principal deformation direction, and abdominal pressure normalized deformation parameters.

[0013] Preferably, the processor is programmed to: determine the registration error component, boundary reliability component, and model correspondence consistency component within the local correction region; determine the feedback correction method based on the dominant error component in the reconstruction uncertainty; update the spatial transformation parameters of the local correction region when the registration error component is dominant; update the target tissue layer determination parameters or closed boundary determination parameters when the boundary reliability component is dominant; update the correspondence between the target tissue layer positions under different abdominal pressure states when the model correspondence consistency component is dominant; and generate acquisition prompt information for supplementing medical imaging data in the local correction region when the reconstruction uncertainty still does not meet the preset reconstruction conditions after a preset number of reconstructions based on the updated parameters. The acquisition prompt information includes at least one of supplementary acquisition position, supplementary acquisition direction, and supplementary abdominal pressure state.

[0014] Compared with the prior art, the present invention has the following advantages: 1. This invention establishes a common reference coordinate system for medical imaging data under different abdominal pressure states based on a reference anatomical region, and combines the structural continuity interruption information of the target tissue layer with the normal displacement component relative to the abdominal wall reference surface to determine candidate defect information; further, it combines the continuity relationship between adjacent image positions, the spatial correspondence across abdominal pressure states, and anatomical topological constraints to form a closed boundary, thereby reducing misjudgments caused by the movement of the examined object, imaging position deviation, local low contrast, or general soft tissue displacement, and improving the integrity and anatomical consistency of the boundary of irregular defect areas.

[0015] 2. This invention constructs a three-dimensional model of the defect region corresponding to different abdominal pressure states under closed boundary constraints, and establishes the correspondence between mesh positions under different abdominal pressure states. It can not only obtain static morphological parameters such as defect area, perimeter, maximum diameter, minor diameter and equivalent diameter, but also dynamic deformation parameters such as area change rate, perimeter change rate, normal bulge and main deformation direction. It realizes the quantitative extension from two-dimensional radial line measurement under a single state to the three-dimensional morphology of the defect region and the deformation response across abdominal pressure states.

[0016] 3. This invention determines the reconstruction uncertainty based on registration error, boundary reliability, and model correspondence consistency. It also feeds back local reconstruction deviations to the spatial registration, target tissue layer determination, closed boundary determination, or model correspondence establishment stages based on the dominant error component, so as to reprocess the local correction area. When the existing medical imaging data is insufficient to support reliable reconstruction, it can also generate supplementary acquisition location, acquisition direction, or abdominal pressure status prompts, thereby forming a closed-loop correction mechanism driven by reconstruction quality evaluation, improving the reliability and traceability of the three-dimensional model and quantitative results. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the three-dimensional reconstruction and quantitative assessment process for inguinal hernia defects.

[0018] Figure 2 This is a schematic diagram of the local feedback correction process based on reconstruction uncertainty. Detailed Implementation

[0019] The embodiments of the present invention will now be described in further detail with reference to the accompanying drawings.

[0020] In this embodiment, a three-dimensional reconstruction and quantitative assessment system for inguinal hernia defects includes an image data acquisition unit, a state detection unit, a data storage unit, a processor, and an output unit. The image data acquisition unit acquires medical image data of the inguinal region of the same subject under at least two different abdominal pressure states. The state detection unit acquires abdominal pressure state information corresponding to the medical image data. The data storage unit stores the medical image data, abdominal pressure state information, registration parameters, a three-dimensional model of the defect region, and quantitative results. The processor is communicatively connected to the above units and configured to perform spatial standardization, target tissue layer determination, defect boundary determination, three-dimensional reconstruction, quantitative calculation, and feedback correction. The output unit outputs the three-dimensional model, quantitative parameters, and uncertainty information characterizing the reconstruction quality.

[0021] In some embodiments, the image data acquisition unit can receive computed tomography volume data, magnetic resonance imaging volume data, three-dimensional ultrasound volume data, or volume data reconstructed from a two-dimensional ultrasound image sequence according to the probe pose and acquisition timing. Therefore, this invention does not depend on a specific image modality. When using a two-dimensional ultrasound image sequence, the spatial pose corresponding to each image frame can be obtained through a position detection device mounted on the ultrasound probe. Alternatively, the relative positional relationship between image frames can be estimated based on stable tissue characteristics in adjacent image frames, and the two-dimensional image sequence can be converted into three-dimensional volume data accordingly. The at least two different abdominal pressure states include a first abdominal pressure state and a second abdominal pressure state, where the abdominal pressure in the second abdominal pressure state is higher than that in the first abdominal pressure state. The first abdominal pressure state can be a resting state under natural breathing; the second abdominal pressure state can be a breath-holding pressure state, a coughing state, an active increase in abdominal pressure state, or an externally applied pressure state. In some embodiments, multiple transitional states between the first and second abdominal pressure states can also be acquired to form time-series image data corresponding to the abdominal pressure change process. The state detection unit can include at least one of an abdominal pressure sensor, a breathing belt, a body surface displacement sensor, a pressure detection component of a pressure application device, or a state recognition program for extracting tissue motion states from medical image data. The processor establishes a synchronization relationship based on the timestamp of the status detection information and the acquisition timestamp of the medical image data, and writes a corresponding abdominal pressure status identifier to each set of medical image data. For implementations where the abdominal pressure value can be directly obtained, the abdominal pressure status identifier includes the abdominal pressure value and the stage of abdominal pressure change; for implementations where the abdominal pressure value cannot be directly obtained, the abdominal pressure status identifier may include at least one of the following status categories: resting, rising, holding, and recovering.

[0022] In this embodiment, the processor first preprocesses the medical image data. Preprocessing includes at least one of grayscale normalization, noise suppression, artifact correction, spatial resolution unification, and region of interest (ROI) truncation. For medical image data from different acquisition states, it is resampled to a common spatial resolution according to its voxel spacing. It should be understood that the purpose of preprocessing is to reduce the impact of acquisition device parameters and imaging noise on subsequent spatial registration and tissue layer extraction, without altering the actual spatial relationships of the anatomical structures within the groin region.

[0023] The processor determines reference anatomical regions from preprocessed medical imaging data. These reference anatomical regions are anatomical structures whose positional changes satisfy reference conditions under different intra-abdominal pressure states; they may consist of bony structures, vascular branching structures, abdominal wall fixation areas, or combinations thereof. Specifically, the processor filters reference anatomical regions based on the amount of positional change and local deformation of candidate anatomical structures under different intra-abdominal pressure states. For example, for candidate anatomical regions… Its reference stability can be expressed as:

[0024] in, Indicates the location within the candidate anatomical region Initial displacement between two intra-abdominal pressure states This represents the average displacement of the candidate anatomical region. This indicates the number of sampling locations included in the candidate anatomical region. This represents the average displacement constraint coefficient. The smaller the value, the less the corresponding candidate anatomical region is affected by changes in abdominal pressure. The processor can determine the candidate anatomical region as the reference anatomical region if the reference stability meets the preset conditions and the spatial distribution can constrain the inguinal region.

[0025] After determining the reference anatomical region, the processor uses medical imaging data under one abdominal pressure state as baseline data to establish a common reference coordinate system. It then calculates the spatial transformation relationship between medical imaging data under other abdominal pressure states and the baseline data, and registers all medical imaging data to the common reference coordinate system. The spatial registration is performed in a global-to-local order. First, the reference anatomical region is used to determine the overall spatial transformation relationship to correct for overall pose differences between different data. Then, local spatial transformation relationships are used to correct local positional deviations caused by non-rigid tissue deformation within the groin region. The spatial transformation relationship corresponding to each abdominal pressure state can be determined by the following formula:

[0026] in, Indicates the first Spatial transformation relationship from individual abdominal pressure state to reference abdominal pressure state. This represents medical imaging data at baseline abdominal pressure. Indicates the first Medical imaging data under abdominal pressure conditions, This indicates differences in image matching within the reference anatomical region. This represents the overall spatial transformation constraint term. Represents the smoothing constraint term for local spatial transformation. and These represent the corresponding constraint weights. The overall spatial transformation relationship can be a rigid transformation or an affine transformation, while the local spatial transformation relationship can be a non-rigid transformation determined based on the correspondence between control points, deformation fields, or organizational features.

[0027] In some embodiments, to prevent the bulging area of ​​the hernia contents from causing traction on the overall registration, the matching weights of suspected defect areas and significantly deformed areas are reduced when calculating image matching differences, while the matching weights of the reference anatomical area are increased. After registration, the bony landmarks, vascular branches, and abdominal wall fixation areas corresponding to different abdominal pressure states are located at their respective positions in a common reference coordinate system, thereby providing a unified spatial benchmark for comparison across abdominal pressure states.

[0028] The processor identifies target tissue layers in the registered medical image data to characterize the continuity of the abdominal wall tissue in the groin region. The target tissue layer is not simply the surface of the hernia sac, but rather the actual anatomical layer or equivalent tissue layer that characterizes the continuity of the abdominal wall tissue and can be used to determine the location of the defect opening. In different image modalities, the target tissue layer can be composed of a single tissue interface or determined by the spatial relationships of multiple adjacent tissue interfaces. Specifically, the processor determines multiple candidate locations for the target tissue layer based on grayscale variations, texture variations, tissue interface gradients, the continuity of adjacent slices, and the spatial relationships between preset anatomical structures in the medical image data. For sampling locations… The probability that it belongs to the target organizational layer can be expressed as:

[0029] in, Indicates sampling position The probability of belonging to the target organizational layer. Indicates the first Imaging features or anatomical relationship features Indicates the corresponding feature weights. Indicates the bias parameter. This represents a mapping function that maps the calculation results to a preset probability range. It should be understood that the above expression is used to illustrate the formation relationship of the target organizational layer probability; in other embodiments, graph search, energy minimization, classification models, or segmentation models can also be used to determine the target organizational layer without affecting the implementation of the inventive concept.

[0030] The processor determines the target tissue layer from multiple candidate locations based on the continuity of tissue interfaces between adjacent image locations, the spatial relationship between candidate locations of the target tissue layer and preset anatomical structures, and the correspondence between candidate locations under different abdominal pressure states. For short-distance interface gaps caused by local low contrast, interpolation can be performed based on the interface orientation in adjacent image locations. For areas where the missing distance exceeds the allowable interpolation range or is inconsistent with the relationship of surrounding anatomical structures, they are not directly filled in but are retained as areas to be determined for subsequent use in determining defect boundaries and calculating uncertainty.

[0031] In this embodiment, the processor performs surface fitting on the target tissue layer to obtain the abdominal wall reference surface. Surface fitting, while preserving the overall spatial position of the target tissue layer, suppresses local abrupt changes caused by individual voxel noise. Its objective function can be expressed as:

[0032] in, This represents the abdominal wall reference surface obtained through fitting. Indicates the first level of the target organization. One sampling point, Indicates the number of sampling points. This represents the distance from the sampling point to the reference surface of the abdominal wall. This represents a robust cost function for suppressing the impact of outliers. This represents the surface smoothing constraint weight.

[0033] The processor further determines the unit surface normal at each sampling location on the abdominal wall reference surface. For the target tissue layer locations corresponding to different abdominal pressure states, the processor calculates the component of its displacement in the normal direction of the abdominal wall reference surface:

[0034] in, Indicates the first The normal displacement component of each abdominal pressure state relative to the reference abdominal pressure state. Indicates the first Location of the target tissue layer under abdominal pressure. This indicates the position corresponding to the baseline abdominal pressure. This represents the corresponding spatial transformation relationship. This indicates the surface normal at the corresponding position under the reference abdominal pressure state.

[0035] The processor determines candidate defect information based on the joint distribution of structural continuity interruption information and normal displacement components of the target tissue layer. Structural continuity interruption information characterizes whether the target tissue layer persists between adjacent image locations, whether its orientation is continuous, and whether the corresponding tissue interface experiences local discontinuity. For sampling locations on the abdominal wall reference surface... The defect response value can be expressed as:

[0036] in, Indicates the degree of disruption in the structural continuity of the target organizational layer. Indicates the normal displacement component, This represents the gradient of normal displacement along the reference surface of the abdominal wall. This indicates the degree to which the corresponding location deviates from the pre-defined anatomical topology of the abdominal wall. to This indicates the corresponding weight. The weight can be determined based on calibration or validation data, or it can be updated during the feedback correction process.

[0037] In this embodiment, the processor does not determine the defect boundary solely based on the defect response value of a single image location. Instead, it identifies high-confidence candidate defect points that repeatedly occur between adjacent image locations, have spatial correspondence under at least two intra-abdominal pressure states, and satisfy anatomical topological constraints. The processor then determines boundary anchor points from these high-confidence candidate points, establishes boundary connection segments between adjacent boundary anchor points, and forms a closed boundary using the boundary anchor points and boundary connection segments.

[0038] The boundary connection segment is simultaneously subject to image matching constraints, boundary continuity constraints, anatomical topology constraints, and trans-abdominal pressure state correspondence constraints. The closed boundary can be determined by the following formula:

[0039] in, Indicates a defined closed boundary. The image matching cost represents the relationship between the boundary and candidate information for defects. This represents the continuity cost corresponding to boundary fractures and abrupt curvature changes. This represents the topological cost when the boundary does not conform to the preset anatomical topological relationship. This represents the dynamic consistency cost when the boundaries do not satisfy the spatial correspondence under different intra-abdominal pressure states. , , and The corresponding weights are indicated. It should be understood that the anatomical topological constraints do not limit the defect boundaries to a fixed geometry, but rather exclude boundary connections that clearly cross continuous abdominal wall tissue, bypass reference anatomical structures, or form unreasonable self-intersections. Therefore, the system can adapt to circular, elliptical, and irregularly shaped defect regions.

[0040] After determining the closed boundary, the processor performs meshing on the target tissue layer within the closed boundary, obtaining a 3D model of the defect region composed of mesh vertices and mesh patches. The mesh sampling density is adjusted based on the local curvature of the target tissue layer and the spatial resolution of the medical imaging data: the mesh sampling density is increased where curvature changes significantly or near the boundary, and decreased in areas with smooth surfaces. While maintaining the spatial position and topological relationship of the closed boundary, the mesh surface is smoothed to suppress jagged abrupt changes caused by voxel discretization.

[0041] For a 3D model of a defect region composed of triangular facets, its surface area is determined by the following formula:

[0042] in, Indicates the area of ​​the missing surface. Indicates the number of mesh patches. , and Indicates the first The three mesh vertices of a triangular facet are used. After mapping the closed boundary and its internal surface to the reference surface of the abdominal wall, the projected area of ​​the defect can be determined. Therefore, the equivalent diameter is determined:

[0043] The maximum diameter can be determined by the maximum geodesic distance or projected distance between any two boundary points on the closed boundary; the minor diameter can be determined by the boundary intercept that intersects the maximum diameter and satisfies a preset directional relationship. The shape irregularity can be determined based on the relationship between the perimeter of the defect and the projected area. In some embodiments, the static morphological parameters include at least two of the following: defect area, defect perimeter, maximum diameter, minor diameter, equivalent diameter, surface curvature, shape irregularity, and spatial orientation.

[0044] The processor establishes the model correspondence between mesh vertices under different intra-abdominal pressure states based on a common reference coordinate system and boundary anchor points on the closed boundary. For corresponding mesh vertices, the processor calculates their positional changes to obtain local displacement information, and uses this to form a displacement distribution map of the defective region. The rate of change of area of ​​an intra-abdominal pressure state relative to the baseline intra-abdominal pressure state can be expressed as:

[0045] in, This represents the area of ​​the defect under baseline abdominal pressure. Indicates the first The defect area under abdominal pressure.

[0046] In embodiments where synchronized abdominal pressure values ​​can be obtained, the normalized abdominal pressure deformation parameter can be expressed as:

[0047] in, and Representing the baseline abdominal pressure state and the first The abdominal pressure value corresponding to each abdominal pressure state. These represent stable parameters used to avoid a denominator of zero. Dynamic deformation parameters may also include the rate of change of perimeter, normal bulge, and principal deformation direction. These dynamic deformation parameters characterize the geometric response of the defect area under changes in abdominal pressure and do not directly output disease diagnosis or treatment conclusions.

[0048] After completing the initial reconstruction, the processor performs a quality assessment on the reconstruction results. The processor determines the local registration error, boundary reliability, and consistency of the trans-ventral pressure state model. For the first... The local reconstruction uncertainty of a grid cell can be expressed as:

[0049] in, Indicates the first The registration error of each grid cell corresponding to the image region This indicates the degree of discreteness of the boundary candidate information or the reliability of the boundary connections. This represents the deviation between corresponding grid positions under different intra-abdominal pressure states. This represents the amount of abrupt change in local curvature on the mesh surface. to This indicates the corresponding weight. The overall reconstruction uncertainty can be obtained by weighting the local reconstruction uncertainties according to the grid area or the importance of the boundary.

[0050] When the overall reconstruction uncertainty meets the preset reconstruction conditions and there are no grid cells whose local reconstruction uncertainty meets the local correction conditions, the processor outputs the current 3D model and quantization parameters of the defect area. When the overall reconstruction uncertainty does not meet the preset reconstruction conditions, or when the local reconstruction uncertainty of at least one grid cell meets the local correction conditions, the processor maps the corresponding grid cell to a local image region in the medical image data and identifies that local image region as the local correction region.

[0051] The processor determines the feedback correction method based on the dominant error component within the local correction region. When the registration error component is dominant, the registration matching weight of the corresponding position in the reference anatomical region is increased, or the local spatial transformation parameters are adjusted; when the boundary reliability component is dominant, the candidate decision parameters of the target tissue layer, the boundary continuity constraint weight, or the anatomical topology constraint weight are adjusted; when the model correspondence consistency component is dominant, the correspondence between the target tissue layer position or the mesh vertices under different abdominal pressure states is redefined; when the local curvature abrupt change is dominant, the mesh sampling density at the corresponding position is increased and boundary smoothing is performed again. In some embodiments, the first... The parameter set after the feedback processing can be updated according to the following formula:

[0052] in, Indicates the first The parameter set used in the secondary feedback processing includes at least one of the following: registration parameters, target tissue layer determination parameters, boundary determination parameters, and model corresponding parameters. Indicates the update step size; This indicates the parameter update direction determined based on the influence direction of each error component on the reconstruction uncertainty. For implementations that do not employ gradient updates, regular updates can also be performed based on a preset correspondence between error components and parameters.

[0053] After parameter updates, the processor only reprocesses the local correction area and its adjacent areas, without repeating the processing of all medical image data. In this embodiment, reprocessing includes at least one of local registration, target tissue layer re-determination, defect candidate information recalculation, closed boundary reconnection, local mesh reconstruction, and model correspondence re-determination. After reprocessing, the reconstruction uncertainty is recalculated, and it is determined whether the preset reconstruction conditions are met. The feedback process stops when any of the following conditions are met: the overall reconstruction uncertainty is lower than the overall threshold and the local reconstruction uncertainty is lower than the local threshold; the change in uncertainty between two adjacent feedback processes is lower than the minimum effective improvement; the number of feedbacks reaches the maximum number of feedbacks. When the reconstruction uncertainty still does not meet the preset reconstruction conditions after reaching the maximum number of feedbacks, the processor does not directly mark the corresponding area as a high-confidence result, but generates acquisition prompt information. The acquisition prompt information includes at least one of supplementary acquisition location, supplementary acquisition direction, and supplementary abdominal pressure status, and marks the local area that needs supplementary data in the output 3D model.

[0054] In this embodiment, the output unit outputs at least the 3D model of the defect area, closed boundaries, static morphological parameters, dynamic deformation parameters, and reconstruction uncertainty. The output format can be a 3D visualization interface, a structured data file, or a report containing model screenshots and quantification results. For the 3D visualization interface, different display attributes can be used to distinguish the 3D model of the defect area under different intra-abdominal pressure states, and the normal displacement distribution and uncertainty distribution can be superimposed and displayed on the model surface. When a user selects a local area on the model, the system can display the corresponding original image location, boundary candidate information, and uncertainty source for that area.

[0055] In some embodiments, the processor can also save the parameters, local correction areas, and changes in reconstruction uncertainty for each feedback process, forming a reconstruction process record. This reconstruction process record is used to verify the model formation process and can be used to subsequently adjust the initial processing parameters for similar image data. It should be noted that the historical reconstruction process record is only used to improve the quality of medical image processing and 3D model reconstruction and does not replace the quality assessment performed on the current subject.

[0056] In other embodiments, candidate locations of target tissue layers can be determined by a pre-trained segmentation model. Training data includes labeled medical images of the groin region and corresponding target tissue layer annotations; the training loss can consist of region overlap loss, boundary loss, topological consistency loss, and trans-abdominal pressure state consistency loss. After training, the segmentation model outputs a probability field of the target tissue layer. The processor still constrains the probability field according to the aforementioned adjacent image continuity relationships, anatomical structure spatial relationships, and trans-abdominal pressure state correspondences, thereby avoiding directly using the single output of the segmentation model as the final defect boundary.

[0057] It should be understood that the above-described segmentation model is not a necessary condition for implementing the present invention. Those skilled in the art can also use graph search, active contouring, level sets, region growing, or multi-feature classification to determine the target tissue layer and defect candidate information, as long as they can combine the structural continuity interruption of the target tissue layer and the normal displacement of the trans-abdominal pressure state, and form a closed boundary through spatial correspondence and anatomical topological constraints, the core technical concept of the present invention can be realized.

[0058] It should be understood that the term "abdominal pressure state" in this invention is used to distinguish load states that can cause detectable changes in the morphology of the inguinal region, and does not require that absolute abdominal pressure be measured through invasive means; the term "reference anatomical region" in this invention refers to an anatomical region that can provide stable constraints for spatial registration between different abdominal pressure states; the term "target tissue layer" in this invention refers to the actual tissue layer or equivalent tissue layer used to characterize the continuity of the abdominal wall tissue in the inguinal region; the term "reconstruction uncertainty" in this invention is used to characterize the reconstruction quality of medical imaging data, closed boundaries, and three-dimensional models, and does not represent a clinical risk conclusion for the subject.

[0059] Combination Figure 1 and Figure 2 In this embodiment, the three-dimensional reconstruction and quantitative assessment of the inguinal hernia defect area includes: acquiring medical image data of the inguinal region of the same subject under at least two different abdominal pressure states, and establishing the correlation between the medical image data and the corresponding abdominal pressure states; determining the reference anatomical region whose positional changes between different abdominal pressure states satisfy the reference conditions, establishing a common reference coordinate system based on the reference anatomical region, and registering the medical image data under different abdominal pressure states to the common reference coordinate system; determining the target tissue layer and establishing an abdominal wall reference surface in the registered medical image data, determining defect candidate information based on the structural continuity interruption information of the target tissue layer and the normal displacement component across the abdominal pressure state, and forming a closed boundary of the defect area by combining the spatial correspondence between the defect candidate information and the anatomical topological constraints; constructing a three-dimensional model of the defect area under different abdominal pressure states based on the closed boundary, and determining the static morphological parameters and dynamic deformation parameters; further determining the reconstruction uncertainty based on the registration error, boundary reliability, and model correspondence consistency; when the reconstruction uncertainty meets the preset reconstruction conditions, outputting the three-dimensional model of the defect area, quantitative parameters, and reconstruction uncertainty; when the reconstruction uncertainty does not meet the preset reconstruction conditions, according to... Figure 2 The feedback correction process shown identifies the local correction region and its dominant error component. Based on the dominant error component, it updates the spatial transformation parameters, target tissue layer determination parameters, closed boundary determination parameters, or the correspondence between the trans-abdominal pressure state model and the local correction region and its adjacent regions. Registration, delimitation, and model reconstruction are then re-performed until the preset reconstruction conditions are met. If the preset reconstruction conditions are still not met after reaching the maximum number of feedback iterations, a prompt message indicating a supplementary acquisition location, supplementary acquisition direction, or supplementary abdominal pressure state is generated. It should be understood that... Figure 1 and Figure 2 The sequence of steps shown illustrates the data transfer and feedback relationships between each processing stage. Without changing the aforementioned processing logic, the independent processing steps can be executed in parallel or the execution order can be adjusted according to the actual image data type.

[0060] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A three-dimensional reconstruction and quantitative assessment system for inguinal hernia defect areas, comprising a processor and an image data acquisition unit and an output unit communicatively connected to the processor, characterized in that: The processor is programmed to: acquire medical image data of the groin region of the same subject under at least two different abdominal pressure states, establish a common reference coordinate system based on the reference anatomical region whose positional changes in each image data meet the reference conditions, and register the medical image data under different abdominal pressure states to the common reference coordinate system. In the registered medical image data, the target tissue layer characterizing the continuity of the abdominal wall tissue is identified and an abdominal wall reference surface is established. Based on the structural continuity interruption information of the target tissue layer and the normal displacement components relative to the abdominal wall reference surface under different abdominal pressure states, defect candidate information is determined. The closed boundary of the defect area is determined by combining the cross-state space correspondence and anatomical topological constraints. Based on the closed boundary, construct a three-dimensional model of the defect area under different abdominal pressure states and establish the model correspondence, thereby determining the static morphological parameters and dynamic deformation parameters of the defect area. The reconstruction uncertainty is determined based on the registration error, boundary reliability, and model correspondence consistency. When the reconstruction uncertainty does not meet the preset reconstruction conditions, the processing parameters of the local correction area are updated and the three-dimensional model of the defective area is reconstructed until the preset reconstruction conditions are met. The three-dimensional model of the defective area, static morphological parameters, dynamic deformation parameters, and reconstruction uncertainty that meet the preset reconstruction conditions are output through the output unit.

2. The three-dimensional reconstruction and quantitative evaluation system for inguinal hernia defect area according to claim 1, characterized in that: The medical imaging data includes three-dimensional volumetric data or image sequences that can be converted into three-dimensional volumetric data according to the acquisition position relationship; the at least two different abdominal pressure states include a first abdominal pressure state and a second abdominal pressure state with abdominal pressure higher than the first abdominal pressure state; the establishment of the association between the medical imaging data and the corresponding abdominal pressure state includes: determining the abdominal pressure state identifier corresponding to each medical imaging data according to the state detection information synchronized with the medical imaging data, wherein the state detection information includes at least one of abdominal pressure detection information, respiratory phase information, body surface motion information, and intra-image tissue motion information.

3. The three-dimensional reconstruction and quantitative evaluation system for inguinal hernia defect area according to claim 1, characterized in that: The processor is programmed to: select reference anatomical regions from the medical imaging data based on the positional changes and local deformations of anatomical structures under different abdominal pressure states; use the reference anatomical regions as spatial constraints to sequentially determine the overall spatial transformation relationship and local spatial transformation relationship between the medical imaging data under different abdominal pressure states; perform overall alignment of the medical imaging data under different abdominal pressure states based on the overall spatial transformation relationship, and correct local positional deviations caused by non-rigid tissue deformation in the groin region based on the local spatial transformation relationship.

4. The three-dimensional reconstruction and quantitative evaluation system for inguinal hernia defect area according to claim 1, characterized in that: The processor is programmed to: determine multiple candidate locations of target tissue layers based on tissue interface features in medical image data; and determine a target tissue layer from the multiple candidate locations based on the continuity relationship of tissue interfaces between adjacent image locations, the spatial relationship between the candidate locations of target tissue layers and preset anatomical structures, and the correspondence between candidate locations of target tissue layers under different abdominal pressure states. Surface fitting is performed on the target tissue layer to obtain the abdominal wall reference surface, and the surface normal at each sampling position on the abdominal wall reference surface is determined.

5. The three-dimensional reconstruction and quantitative evaluation system for inguinal hernia defect area according to claim 4, characterized in that: The processor is programmed to: determine structural continuity interruption information based on the degree of interruption of the target tissue layer between adjacent image positions; and determine the normal displacement component based on the displacement of the target tissue layer positions along the surface normal under different abdominal pressure states. Based on the joint distribution of structural continuity interruption information and normal displacement components, candidate defect points are determined, and boundary anchor points that repeatedly appear between adjacent image locations and satisfy anatomical topological constraints are identified from the candidate defect points. Boundary connection segments that simultaneously satisfy image matching constraints, boundary continuity constraints, and cross-abdominal pressure state corresponding constraints are identified between adjacent boundary anchor points, and closed boundaries are formed by boundary anchor points and boundary connection segments.

6. The three-dimensional reconstruction and quantitative evaluation system for inguinal hernia defect area according to claim 1, characterized in that: The processor is programmed to: mesh the target tissue layer within the closed boundary to obtain a three-dimensional model of the defect region including mesh vertices and mesh patches; adjust the mesh sampling density at corresponding positions according to the local curvature of the target tissue layer and the spatial resolution of the medical image data; perform surface smoothing on the three-dimensional model of the defect region while keeping the spatial position and topological relationship of the closed boundary unchanged; and determine static morphological parameters based on the mesh vertices and mesh patches, wherein the static morphological parameters include at least two of the following: defect area, defect perimeter, maximum diameter, minor diameter intersecting the maximum diameter, equivalent diameter, surface curvature, shape irregularity, and spatial orientation.

7. The three-dimensional reconstruction and quantitative evaluation system for inguinal hernia defect area according to claim 6, characterized in that: The processor is programmed to: establish the correspondence between mesh vertices under different intra-abdominal pressure states based on the common reference coordinate system and the boundary anchor points on the closed boundary; and determine local displacement information based on the position changes of the corresponding mesh vertices under different intra-abdominal pressure states. Dynamic deformation parameters are determined based on the changes in defect area, defect perimeter, and local displacement under different abdominal pressure conditions. The dynamic deformation parameters include at least two of the following: area change rate, perimeter change rate, normal bulge, main deformation direction, and abdominal pressure normalized deformation parameters.

8. The three-dimensional reconstruction and quantitative evaluation system for inguinal hernia defect area according to claim 1, characterized in that: The processor is programmed to: determine the registration error component, boundary reliability component, and model correspondence consistency component within the local correction region; determine the feedback correction method based on the dominant error component in the reconstruction uncertainty; update the spatial transformation parameters of the local correction region when the registration error component is dominant; update the target tissue layer determination parameters or closed boundary determination parameters when the boundary reliability component is dominant; update the correspondence between the target tissue layer positions under different abdominal pressure states when the model correspondence consistency component is dominant; and generate acquisition prompt information for supplementing medical imaging data in the local correction region when the reconstruction uncertainty still does not meet the preset reconstruction conditions after a preset number of reconstructions based on the updated parameters. The acquisition prompt information includes at least one of supplementary acquisition position, supplementary acquisition direction, and supplementary abdominal pressure state.

Images