A tumor thermal ablation multi-modal closed-loop boundary evaluation method and system

Abstract

The application provides a tumor thermal ablation multi-modal closed-loop boundary evaluation method and system. The method generates a planned boundary by finite element prediction of a temperature field and a damage field, completes spatial registration and puncture guidance through augmented reality, reconstructs a temperature field by using a baseline / ablation CT attenuation difference and extracts a thermal boundary, obtains a biological boundary by using a general morphology and NADH vitality staining, quantifies errors of the boundaries and feeds back a model, temperature measurement, navigation and threshold parameters. The scheme can realize multi-modal interpretation, calibration and closed-loop optimization of a thermal ablation boundary.

Description

Technical Field

[0001] This invention relates to the field of tumor thermal ablation navigation and efficacy evaluation technology, and in particular to a multimodal closed-loop boundary evaluation method and system for tumor thermal ablation. Background Technology

[0002] Thermal ablation techniques such as microwave ablation, radiofrequency ablation, laser ablation, and high-intensity focused ultrasound (HIFU) have been widely used in the minimally invasive treatment of solid tumors. The adequacy of treatment hinges on whether the ablation zone completely covers the target area while maintaining the necessary safety margin, and simultaneously avoiding excessive thermal damage to blood vessels, bile ducts, nerves, or adjacent organs. Therefore, reliably identifying and interpreting the ablation boundary is a core issue in treatment planning, intraoperative navigation, and postoperative efficacy evaluation.

[0003] Existing ablation boundaries are typically derived from gross morphological observation, intraoperative or postoperative imaging, temperature thresholds, or numerical model predictions. Gross morphology can directly display areas of tissue whitening, charring, or discoloration, but these areas often exhibit color gradients at their periphery, and the boundaries are significantly influenced by the observer and the tissue treatment method. Finite element models can simulate electromagnetic energy deposition, heat conduction, and thermal damage accumulation, but the model results depend on tissue thermophysical properties, electromagnetic parameters, boundary conditions, needle placement, and damage kinetic assumptions.

[0004] CT thermometry can estimate the temperature field by utilizing changes in CT attenuation values ​​before and after ablation, and further extract isotherms or thermal boundaries. It has advantages such as compatibility with CT-guided interventional procedures, high spatial resolution, and ease of 3D reconstruction. However, CT temperature estimation is easily affected by image registration, metal artifacts, bubbles, noise, tissue dehydration, and temperature-attenuation calibration relationships; a fixed temperature threshold is not necessarily equivalent to the boundary of irreversible necrosis.

[0005] NADH viability staining can reflect the activity of tissue metabolic enzymes. Areas with preserved activity usually show obvious staining, while thermally damaged or inactive areas show no staining or weak staining. Therefore, it can provide a biological reference for tissue viability after ablation. However, NADH verification usually occurs after tissue sampling or sectioning and cannot directly replace intraoperative imaging and model assessment. It also requires establishing a correspondence between the image coordinates, model coordinates, and actual operation coordinates.

[0006] In existing technologies, finite element simulation, AR navigation, CT temperature measurement, and NADH validation are often used as independent steps. When there are systematic deviations between model prediction boundaries, image temperature boundaries, gross morphology boundaries, and biological boundaries, the lack of a unified coordinate management, error quantification, and feedback update mechanism makes it difficult to determine the source of the deviation and also makes it difficult to use the biological validation results in reverse to correct the model, navigation, or temperature measurement parameters.

[0007] Therefore, there is an urgent need for a multimodal closed-loop assessment technology for tumor thermal ablation, which can enable preoperative / intraoperative finite element prediction, augmented reality navigation, CT temperature field reconstruction, and NADH biological validation to work together within the same workflow. This technology can output the ablation range under different boundary definitions and correct the model, temperature measurement, navigation, and threshold parameters based on boundary differences, thereby improving the consistency and traceability of ablation boundary interpretation. Summary of the Invention

[0008] In view of this, the purpose of the present invention is to provide a multimodal closed-loop boundary assessment method and system for tumor thermal ablation, which aims to form a complete closed loop from planning and prediction, spatial navigation, image measurement, biological verification to parameter feedback and updating, so as to improve the accuracy, interpretability and reproducibility of ablation boundary assessment.

[0009] To achieve the above objectives, the present invention adopts the following technical solution: a multimodal closed-loop boundary assessment method for tumor thermal ablation, comprising the following steps: Step 1: Obtain preoperative or intraoperative CT images of the subject to thermal ablation, information on the target area or area to be ablated, geometric parameters of the ablation needle or antenna, energy output parameters, and tissue thermal and electromagnetic properties, and establish a unified data management coordinate system; Step 2: Based on CT images, target area information and ablation device parameters, establish a finite element ablation model, calculate electromagnetic energy deposition, transient temperature field and thermal damage field, and obtain the finite element ablation model's predicted ablation boundary and recommended puncture or needle placement scheme; Step 3: Map the ablation boundary, puncture path, ablation needle posture and target safety margin predicted by the finite element ablation model to the augmented reality navigation space. Align the virtual plan with the actual intervention environment and output navigation prompts through markers, camera or image coordinate registration. Step 4: Acquire baseline CT images and CT images after ablation or during ablation, spatially register the two and calculate the CT attenuation difference, and reconstruct the three-dimensional CT temperature field using the temperature-attenuation calibration model. Step 5: Extract the image source thermal boundary from the three-dimensional CT temperature field. The thermal boundary includes isothermal boundary, thermal dose boundary, or ablation boundary derived from the temperature field, and transform it into a coordinate system that is the same as or mutually convertible with the finite element model, augmented reality navigation, and tissue slices. Step 6: Perform gross morphological observation and / or NADH activity staining on the ablated tissue to obtain the gross visible boundary and the biological boundary reflecting the loss of tissue metabolic activity; Step 7: Spatial comparison of the model prediction boundary, augmented reality navigation position, CT temperature measurement boundary, gross morphology boundary and NADH biological boundary, and calculation of orientation correlation error, boundary distance, overlap, major axis and minor axis size difference and credibility index. Step 8: Based on the spatial comparison results, form a closed-loop feedback and update the finite element model parameters, CT temperature calibration parameters, boundary thresholds, navigation registration parameters, or the next round of ablation plan; Step 9: Output the multimodal ablation boundary assessment results, closed-loop feedback parameters, visualization report, and correction suggestions for subsequent experimental or clinical decisions.

[0010] In a preferred embodiment: the finite element ablation model in step 2 is at least coupled with electromagnetic field solution, specific absorptivity calculation, Pennes' biological heat transfer equation, and Arrhenius thermal damage model; wherein the electromagnetic field solution is used to calculate the energy deposition around the ablation antenna, the Pennes' biological heat transfer equation is used to calculate the change of tissue temperature over time, and the Arrhenius thermal damage model is used to obtain the irreversible damage index or damage fraction.

[0011] In a preferred embodiment: the finite element ablation model further incorporates temperature-related thermal conductivity, electrical conductivity, dielectric constant, specific heat capacity or blood perfusion parameters, and uses one or more of Ω=1, damage fraction threshold, isothermal threshold or thermal dose threshold as the basis for the finite element ablation model to predict the ablation boundary.

[0012] In a preferred embodiment: the augmented reality navigation in step 3 uses dense sampling markers, QR code markers, optical markers, spatial positioning sensors or a combination thereof to establish a rigid transformation between physical space and camera space, and superimposes virtual needle tracks, target areas and safety boundaries onto the actual field of view through camera intrinsic parameter matrix, marker coordinate transformation and model coordinate transformation.

[0013] In a preferred embodiment: the CT temperature field reconstruction in step 4 includes CT image registration, HU difference map construction, outlier removal, temperature-attenuation mapping and three-dimensional temperature matrix generation; the temperature-attenuation mapping can be linear regression, piecewise linear model, nonlinear model or data-driven model obtained by experimental calibration.

[0014] In a preferred embodiment, the outlier removal employs interquartile range, robust regression, metal artifact identification, bubble region masking, spatial filtering, or a combination thereof, to reduce the influence of metal needles, gas, noise, and local non-uniform tissue on the CT temperature measurement boundary.

[0015] In a preferred embodiment: NADH viability staining in step 6 is used to distinguish between tissues that retain metabolic activity and tissues that have been deactivated after thermal damage, wherein the stained area serves as an activity reference and the unstained or weakly stained area serves as an inactive reference, and the NADH biological boundary is used to provide a biological interpretation of the CT temperature measurement boundary, the model prediction boundary, and the gross morphological boundary.

[0016] In a preferred embodiment: the spatial comparison in step 7 includes at least one or more of the following: major axis diameter, minor axis diameter, equivalent diameter, volume, area, Dice coefficient, Hausdorff distance, average surface distance, boundary normal deviation, and directional errors along the ablation needle axis and radial direction.

[0017] In a preferred embodiment: the closed-loop feedback in step 8 includes: correcting the thermophysical parameters, boundary conditions, damage kinetic parameters, or energy deposition parameters when there is a systematic deviation between the model prediction boundary and the NADH biological boundary or the CT thermometry boundary; correcting the temperature-attenuation calibration parameters or thermal boundary threshold when there is a deviation between the CT thermometry boundary and the NADH biological boundary; and correcting the registration matrix, marker positioning, or puncture path when there is a deviation between the navigation position and the planned path. The closed-loop feedback is used for preoperative planning optimization, in vitro validation experiment optimization, postoperative efficacy evaluation, model individualized calibration, ablation boundary deviation analysis, or generation of the next round of ablation scheme, and the closed-loop feedback is not limited to real-time control of ablation energy output.

[0018] This invention also provides a multimodal closed-loop boundary assessment system for tumor thermal ablation, comprising: a data acquisition module, a finite element prediction module, an augmented reality navigation module, a CT temperature measurement module, a boundary extraction module, a biological validation module, a cross-modal comparison module, a closed-loop feedback update module, and a result output module; wherein, the finite element prediction module is used to generate the model-predicted temperature field, thermal damage field, and planned boundary; the augmented reality navigation module is used for spatial registration and path overlay; the CT temperature measurement module is used to reconstruct the temperature field from CT attenuation changes; the biological validation module is used to receive or analyze NADH activity staining results; and the closed-loop feedback update module is used to update the model, temperature measurement, navigation, or planning parameters based on cross-modal deviations. The system further includes a closed-loop database and a visualization interaction module; the closed-loop database is used to store CT images, finite element meshes, AR registration matrices, temperature fields, damage fields, boundary curves, NADH images and error indicators, and the visualization interaction module is used to simultaneously display ablation boundaries from different sources in two-dimensional slices, three-dimensional models or augmented reality views.

[0019] Compared with the prior art, the present invention has the following beneficial effects: 1) Forming a closed loop of finite element method - AR navigation - CT temperature measurement - NADH validation. This approach does not output model or image boundaries in isolation, but connects planning prediction, space navigation, image temperature measurement and biological validation into a unified workflow, allowing for the tracking of the source and meaning of each boundary.

[0020] 2) Improve the reliability of ablation boundary interpretation. Comparing finite element boundaries, CT thermal boundaries, gross morphology boundaries, and NADH biological boundaries within a unified coordinate framework can distinguish model deviations, image thermometry deviations, navigation registration deviations, and differences in biological endpoints, avoiding the simple interchange of boundaries between different modalities.

[0021] 3) Feedback calibration of the model and temperature measurement parameters is possible. By using NADH viability staining and gross morphology results, thermophysical parameters, damage threshold, temperature-attenuation calibration coefficient, and boundary extraction threshold are corrected in reverse, making subsequent ablation plans or experimental verifications closer to the actual tissue response.

[0022] 4) Enhance spatial consistency between navigation and verification. By overlaying virtual plans, needle positions, and predicted boundaries onto the intervention scenario through augmented reality registration, and incorporating navigation errors into the closed-loop evaluation, boundary judgment errors caused by needle deviations or coordinate inconsistencies can be reduced.

[0023] 5) Applicable to in vitro validation, animal experiments, and clinical auxiliary evaluation. This invention can be used for the research and validation of different thermal ablation methods such as microwave ablation and radiofrequency ablation, as well as for postoperative evaluation, individualized model calibration, and multimodal data accumulation in image-guided ablation. Attached Figure Description

[0024] Figure 1 A schematic diagram of the physical scenario setup for the software and hardware system.

[0025] Figure 2 This is a schematic diagram of the preoperative simulated temperature field and tissue damage field in an embodiment of the present invention.

[0026] Figure 3 This is a schematic diagram comparing the ablation example and the visualization results of the CT temperature measurement model in the embodiments of the present invention.

[0027] Figure 4 This is a schematic diagram comparing the ablation example and NADH staining results in an embodiment of the present invention. Detailed Implementation

[0028] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0029] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0030] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations according to this application; as used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise; furthermore, it should be understood that when the terms “comprising” and / or “including” are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0031] A multimodal closed-loop boundary assessment method for tumor thermal ablation, referencing Figure 1-4 This includes the following steps: Step 1: Establishing a Closed-Loop Data Structure. Taking microwave ablation of liver tumors or ex vivo liver tissue as an example, the system first acquires CT images, ablation antenna models, target areas, planned needle paths, output power, treatment time, and tissue thermophysical and electromagnetic properties. The system records CT voxel coordinates, finite element mesh node coordinates, AR marker coordinates, camera coordinates, and post-ablation slice image coordinates in the closed-loop database, and saves traceable timestamps, calibration parameters, and version numbers for different data sources.

[0032] Step 2: Finite Element Prediction. A three-dimensional finite element mesh is established based on the ablation antenna and tissue model. The electromagnetic field is solved to obtain the energy deposition distribution in the tissue, and this energy deposition is used as an external heat source to calculate the transient temperature field. Thermal damage can be calculated using the Arrhenius model, which can be expressed as Ω=A∫exp(-Ea / RT(τ))dτ, and the damage fraction can be expressed as θd=1-exp(-Ω). When Ω reaches a preset threshold or the damage fraction reaches a preset value, the system extracts the corresponding isosurface as the ablation boundary predicted by the model.

[0033] Step 3: Augmented Reality Navigation. The system sets up dense sampling markers or other spatial markers near the intervention area. After the industrial camera or display device acquires images of the markers, the transformation from the marker coordinate system to the camera coordinate system is obtained through an attitude estimation algorithm. The planned needle path point Pvirtual in the virtual model is projected onto the image plane through the transformation from model to marker, the transformation from marker to camera, and the camera intrinsic parameter matrix K. Thus, the puncture path, planned boundary, and needle tip position are simultaneously displayed in the augmented reality display.

[0034] Step 4: CT Temperature Measurement Boundary. After ablation, CT images are acquired and registered with the baseline CT to obtain the voxel-by-voxel HU difference ΔHU(x,y,z). In one embodiment, the temperature change satisfies ΔT(x,y,z)=α·ΔHU(x,y,z)+β, where α and β are obtained through pre-calibration or experimental fitting. The system uses the interquartile range method to remove abnormal CT values ​​and performs spatial smoothing or artifact masking on the temperature field to obtain the three-dimensional temperature matrix Mtemp. Subsequently, the 60℃ isotherm or other threshold boundaries are extracted as the thermal boundary from which the CT image originates.

[0035] Step 5: NADH Biological Validation. After ablation, the tissue is cut along the long axis of the ablation needle, and gross morphological images are taken. The long and short axis dimensions of clear and blurred boundaries are measured. Representative sections are then stained with NADH activity. Active tissue, due to the retention of metabolic enzyme activity, appears blue-purple or other designated stains, while thermally necrotic or inactive areas are unstained or weakly stained. The system can obtain NADH-defined inactive boundaries through manual outlining, threshold segmentation, color evidence extraction, or annotation by pathologists.

[0036] Step Six: Cross-modal Comparison. On the corresponding cross-sections, the system obtains finite element boundaries, CT temperature measurement boundaries, generally clear boundaries, generally blurred boundaries, and NADH boundaries, respectively. The system calculates the dimensional differences in the L and S directions, where the L direction is the major axis along the ablation needle axis, and the S direction is the minor axis perpendicular to the ablation needle axis. The system can also calculate the average surface distance, maximum distance, Dice overlap coefficient, and deviation sign between the boundaries. These indicators are used to determine whether a boundary is conservative, expansive, or directionally offset relative to another boundary.

[0037] Step 7: Feedback Update Rules. If the finite element boundary is close to the grossly clear boundary but significantly larger than the NADH boundary, the system indicates that the model boundary is closer to visible thermal damage than metabolic inactivation boundary, and the damage boundary threshold weight can be reduced or an NADH calibration factor can be introduced. If the CT thermometry boundary is smaller than the gross morphology boundary but close to the NADH boundary, the system marks the CT boundary as a relatively conservative biologically relevant thermal boundary. If the long axis error is significantly greater than the short axis error, the system indicates that there is a directional deviation in energy deposition, heat conduction, or endpoint identification along the ablation needle direction, and prioritizes updating the long axis-related model or measurement parameters.

[0038] Step 8: System Implementation. The system can be implemented on a medical imaging workstation or jointly implemented by server-side computing and front-end display devices. The data acquisition module receives CT images, camera images, pathological images, and ablation device parameters; the finite element prediction module calls the simulation program to generate temperature and damage fields; the augmented reality navigation module performs marker recognition, coordinate transformation, and image overlay; the CT temperature measurement module performs image registration, HU difference, temperature reconstruction, and thermal boundary extraction; the biological validation module receives NADH stained images and generates biological boundaries; the closed-loop feedback update module updates the model, threshold, registration, or planning parameters based on cross-modal errors; and the results output module generates visualized images and structured reports.

[0039] In a specific experimental scenario, isolated porcine liver can be ablated using microwaves at a fixed power, with different treatment times creating ablation zones of varying sizes. The system compares the model-predicted boundaries, CT-measured boundaries, gross morphological boundaries, and NADH boundaries along their major and minor axes. If the finite element prediction is close to the well-defined gross boundary, while the CT-measured boundary is smaller than the gross boundary and close to the NADH inactive boundary, the system records that different boundaries correspond to different tissue endpoints and feeds this result back into the optimization of subsequent model thresholds and CT thermal boundary thresholds.

[0040] This invention does not require all embodiments to simultaneously possess clinical real-time energy control. For research or preclinical validation scenarios, closed-loop control can manifest as parameter updates and strategy optimization between experimental rounds; for systems with real-time data links, closed-loop control can be further extended to intraoperative plan adjustments, additional ablation recommendations, or risk alerts.

Claims

1. A multimodal closed-loop boundary assessment method for tumor thermal ablation, characterized in that, Includes the following steps: Step 1: Obtain preoperative or intraoperative CT images of the subject to thermal ablation, information on the target area or area to be ablated, geometric parameters of the ablation needle or antenna, energy output parameters, and tissue thermal and electromagnetic properties, and establish a unified data management coordinate system; Step 2: Based on CT images, target area information and ablation device parameters, establish a finite element ablation model, calculate electromagnetic energy deposition, transient temperature field and thermal damage field, and obtain the finite element ablation model's predicted ablation boundary and recommended puncture or needle placement scheme; Step 3: Map the ablation boundary, puncture path, ablation needle posture and target safety margin predicted by the finite element ablation model to the augmented reality navigation space. Align the virtual plan with the actual intervention environment and output navigation prompts through markers, camera or image coordinate registration. Step 4: Acquire baseline CT images and CT images after ablation or during ablation, spatially register the two and calculate the CT attenuation difference, and reconstruct the three-dimensional CT temperature field using the temperature-attenuation calibration model. Step 5: Extract the image source thermal boundary from the three-dimensional CT temperature field. The thermal boundary includes isothermal boundary, thermal dose boundary, or ablation boundary derived from the temperature field, and transform it into a coordinate system that is the same as or mutually convertible with the finite element model, augmented reality navigation, and tissue slices. Step 6: Perform gross morphological observation and / or NADH activity staining on the ablated tissue to obtain the gross visible boundary and the biological boundary reflecting the loss of tissue metabolic activity; Step 7: Spatial comparison of the model prediction boundary, augmented reality navigation position, CT temperature measurement boundary, gross morphology boundary and NADH biological boundary, and calculation of orientation correlation error, boundary distance, overlap, major axis and minor axis size difference and credibility index. Step 8: Based on the spatial comparison results, form a closed-loop feedback and update the finite element model parameters, CT temperature calibration parameters, boundary thresholds, navigation registration parameters, or the next round of ablation plan; Step 9: Output the multimodal ablation boundary assessment results, closed-loop feedback parameters, visualization report, and correction suggestions for subsequent experimental or clinical decisions.

2. The method for evaluating the multimodal closed-loop boundary of tumor thermal ablation according to claim 1, characterized in that: The finite element ablation model in step 2 is coupled with at least the electromagnetic field solution, specific absorptivity calculation, Pennes' biological heat transfer equation, and Arrhenius thermal damage model. The electromagnetic field solution is used to calculate the energy deposition around the ablation antenna, the Pennes' biological heat transfer equation is used to calculate the change of tissue temperature over time, and the Arrhenius thermal damage model is used to obtain the irreversible damage index or damage fraction.

3. The method for evaluating the multimodal closed-loop boundary of tumor thermal ablation according to claim 2, characterized in that: The finite element ablation model further incorporates temperature-related thermal conductivity, electrical conductivity, dielectric constant, specific heat capacity, or blood perfusion parameters, and uses one or more of Ω=1, damage fraction threshold, isothermal threshold, or thermal dose threshold as the basis for extracting the ablation boundary predicted by the finite element ablation model.

4. The method for evaluating the multimodal closed-loop boundary of tumor thermal ablation according to claim 1, characterized in that: In step 3, augmented reality navigation uses dense sampling markers, QR code markers, optical markers, spatial positioning sensors, or a combination thereof to establish a rigid transformation between physical space and camera space, and overlays virtual needle paths, target areas, and safety boundaries onto the actual field of view through camera intrinsic parameter matrices, marker coordinate transformation, and model coordinate transformation.

5. The method for evaluating the multimodal closed-loop boundary of tumor thermal ablation according to claim 1, characterized in that: Step 4, CT temperature field reconstruction, includes CT image registration, HU difference map construction, outlier removal, temperature-attenuation mapping, and three-dimensional temperature matrix generation. The temperature-attenuation mapping can be performed using linear regression, piecewise linear models, nonlinear models, or data-driven models obtained from experimental calibration.

6. The method for evaluating the multimodal closed-loop boundary of tumor thermal ablation according to claim 5, characterized in that: The outlier removal employs interquartile range, robust regression, metal artifact identification, bubble region masking, spatial filtering, or a combination thereof, to reduce the influence of metal needles, gas, noise, and local non-uniform tissue on the CT temperature measurement boundary.

7. The method for evaluating the multimodal closed-loop boundary of tumor thermal ablation according to claim 1, characterized in that: In step 6, NADH viability staining is used to distinguish between tissues that retain metabolic activity and tissues that have been deactivated by thermal damage. The stained areas serve as active references, while the unstained or weakly stained areas serve as inactive references. The NADH biological boundaries are used to provide a biological interpretation of CT temperature measurement boundaries, model prediction boundaries, and gross morphological boundaries.

8. The method for evaluating the multimodal closed-loop boundary of tumor thermal ablation according to claim 1, characterized in that: The spatial comparison in step 7 includes at least one or more of the following: major axis diameter, minor axis diameter, equivalent diameter, volume, area, Dice coefficient, Hausdorff distance, average surface distance, boundary normal deviation, and directional errors along the ablation needle axis and radial direction.

9. The method for evaluating the multimodal closed-loop boundary of tumor thermal ablation according to claim 1, characterized in that: The closed-loop feedback in step 8 includes: correcting the thermophysical parameters, boundary conditions, damage kinetic parameters, or energy deposition parameters when there is a systematic deviation between the model prediction boundary and the NADH biological boundary or the CT thermometry boundary; correcting the temperature-attenuation calibration parameters or thermal boundary threshold when there is a deviation between the CT thermometry boundary and the NADH biological boundary; and correcting the registration matrix, marker positioning, or puncture path when there is a deviation between the navigation position and the planned path. The closed-loop feedback is used for preoperative planning optimization, in vitro validation experiment optimization, postoperative efficacy evaluation, model individualized calibration, ablation boundary deviation analysis, or generation of the next round of ablation scheme, and the closed-loop feedback is not limited to real-time control of ablation energy output.

10. A multimodal closed-loop boundary assessment system for tumor thermal ablation, characterized in that, include: The system comprises a data acquisition module, a finite element prediction module, an augmented reality navigation module, a CT temperature measurement module, a boundary extraction module, a biological validation module, a cross-modal comparison module, a closed-loop feedback update module, and a result output module. The finite element prediction module generates the model's predicted temperature field, thermal damage field, and planned boundary. The augmented reality navigation module performs spatial registration and path overlay. The CT temperature measurement module reconstructs the temperature field from CT attenuation changes. The biological validation module receives or analyzes NADH viability staining results. The closed-loop feedback update module updates the model, temperature measurement, navigation, or planning parameters based on cross-modal deviations. The system further includes a closed-loop database and a visualization interaction module; the closed-loop database is used to store CT images, finite element meshes, AR registration matrices, temperature fields, damage fields, boundary curves, NADH images and error indicators, and the visualization interaction module is used to simultaneously display ablation boundaries from different sources in two-dimensional slices, three-dimensional models or augmented reality views.

Images