3D printing ankle joint ABS bone model

Abstract

The invention discloses a 3D printing ABS bone model for ankle joint operation plan reproduction and postoperative form prediction, and aims to solve the problem that an existing three-dimensional bone model can only display a preoperative anatomical structure and cannot visually display an operation plan and a postoperative bone form. According to the method, bone structure reconstruction is carried out based on preoperative three-dimensional imaging data of a patient, after a doctor selects an operation scheme, a three-dimensional model of an ankle joint postoperative bone structure is generated through parameterized simulation of osteotomy, rotation and correction processes, and the three-dimensional model is materialized with an ABS photosensitive resin material by adopting a photocuring 3D printing technology. Postoperative bone forms and related operation adjustment information can be displayed in the model and are used for assisting preoperative diagnosis, operation planning, doctor-patient communication and intraoperative reference. The preparation process is clear, the cost is controllable, and the preparation method has good clinical application prospect and popularization value.

Description

Technical Field

[0001] This invention relates to the fields of medical devices, digital orthopedics and surgical assistance technologies, specifically to a personalized 3D printed ankle joint ABS bone model system that integrates multimodal image fusion, biomechanical sensing and intelligent analysis functions. Background Technology

[0002] In the treatment of ankle joint diseases, such as ankle osteoarthritis, ankle instability, and malunion of fractures, precise preoperative planning and surgical simulation are crucial. Traditional preoperative planning mainly relies on two-dimensional medical images (such as X-rays and CT scans), requiring doctors to construct a three-dimensional anatomical structure in their minds, which suffers from poor intuitiveness and lack of personalization. Although CT-based three-dimensional reconstruction printing models have been applied clinically, their function is limited, only providing a static anatomical morphology display, unable to simulate the surgical procedure, and lacking quantitative assessment of postoperative biomechanical effects.

[0003] In current technologies, traditional 3D-printed orthopedic models have limited functionality. Some models attempt to integrate simple hinges to simulate joint movement, but lack sophisticated biomechanical feedback mechanisms. While some research-oriented biomechanical models can measure pressure or strain, their sensing systems are often external, with complex wiring that interferes with normal anatomical morphology and is not closely integrated with individualized patient imaging data and surgical plans. Furthermore, the ability to learn and optimize surgical plans from extensive clinical practice is lacking in current model systems. Therefore, there is an urgent need for a comprehensive ankle joint surgery assistance system that can deeply integrate multimodal patient imaging data, dynamically simulate the surgical process, verify biomechanical effects in real time, and continuously optimize plans through intelligent learning. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a comprehensive and intelligent 3D-printed ABS bone model for the ankle joint. By integrating multimodal image fusion, operable dynamic simulation, embedded biomechanical verification, and intelligent iterative optimization, it provides a closed-loop solution for ankle joint surgery that covers the entire process of preoperative planning, intraoperative guidance, and postoperative evaluation, with quantitative decision support.

[0005] To address the aforementioned issues, this invention provides a 3D-printed ABS bone model of the ankle joint, comprising: a multimodal image data fusion and preprocessing module, which acquires CT images, MRI images, and weight-bearing X-ray films of the patient's ankle joint, fuses the multimodal data using an image registration algorithm, and extracts bone tissue regions using an adaptive threshold segmentation algorithm. During the registration process, key anatomical landmarks of the ankle joint are introduced: the medial malleolus tip, the lateral malleolus tip, the center point of the talus articular surface, and the calcaneal talus process. The fusion process also integrates a deep learning-based semantic segmentation network for automatically identifying and correcting soft tissue interference in the images; a three-dimensional reconstruction and surgical plan reproduction module with a dynamic simulation mechanism, which generates a preoperative three-dimensional bone model of the ankle joint based on the preprocessed bone tissue data, and integrates a detachable dynamic simulation mechanism into the preoperative three-dimensional bone model. The dynamic simulation mechanism includes an elastic hinge and a scaled locking component; and a biomechanical verification module, which includes: a miniature pressure sensor array embedded in the joint contact surface and a laser force line indicator placed on the outside of the model. A fiber optic strain sensing network, in collaboration with a miniature pressure sensor array, is used to collect real-time data on the microscopic strain distribution of the bone model under simulated loads. The strain sensors are embedded along the trabecular bone of the ankle joint, and the strain signals are converted into stress cloud maps of the bone structure via a fiber optic demodulator. Simultaneously, this sensing network is linked with a laser force line indicator to generate a thermal map of joint surface pressure distribution, and additionally outputs stress concentration areas, maximum strain values, and strain gradient curves at the trabecular bone level. A personalized adaptation and intraoperative guidance component includes a 3D-printed personalized positioning base based on the patient's ankle skin contour data and raised guide markers printed with biodegradable PLA material, with surgical operation prompts engraved on the surface. An intelligent iterative optimization module is used to establish a "surgical plan-postoperative effect" database. This module analyzes the correlation between preoperative parameters, surgical plan parameters, and postoperative effects using a random forest algorithm, automatically recommending the optimal range of surgical parameters for new patients.

[0006] Furthermore, the image registration algorithm includes: firstly, extracting the edge feature points of ankle bone tissue from multimodal images using the SIFT algorithm, and then selecting effective feature points by combining key anatomical landmarks such as the medial malleolus tip; secondly, generating a dynamic deformation field through finite element deformation simulation technology to fine-tune the local image; and finally, using RMSE≤0.05mm and SSIM≥0.95 after registration as the qualified standard.

[0007] Furthermore, the adaptive threshold segmentation algorithm includes: dividing bone tissue into three layers based on the HU value of CT images: cortical bone (1000-1500 HU), outer cancellous bone (600-1000 HU), and inner cancellous bone (200-600 HU), and modeling them; calculating the initial threshold using the Otsu algorithm, and iteratively optimizing the threshold using a region growing algorithm; finally, verifying the segmentation results based on the anatomical constraints of the ankle joint and removing false pixel blocks.

[0008] Furthermore, the deep learning-based semantic segmentation network includes: normalized CT, MRI, and heavy-duty X-ray films as three-channel inputs; embedding a cross-modal attention mechanism in the encoding stage to assign dynamic weights to different modalities; introducing an anatomical structure constraint loss function in the decoding stage, which is combined with Dice and cross-entropy loss functions for optimization; and correcting soft tissue interference through morphological processing and connected region analysis after output.

[0009] Furthermore, the 3D reconstruction and surgical plan reproduction module with dynamic simulation mechanism uses an improved MarchingCubes algorithm to generate a 3D bone model, specifically including: cross-modal grayscale compensation: applying weight coefficients of 0.8 / 1.2 to the CT cortex and X-ray force line areas respectively, eliminating modal artifacts through bilinear interpolation, and ensuring that the curvature error of the talus trochlear articular surface is ≤0.2mm; dynamic threshold iteration: based on the adjacent slice SSIM≥0.92, the threshold of HU for cancellous bone fluctuates by ±15HU from 200-800, reducing the serrated artifacts at the tip of the medial malleolus; anatomical constraint mesh optimization: densifying the triangular facets to 0.3mm in the stress area of ​​the talus neck, merging coplanar meshes in the calcaneal suspensory process area to reduce redundancy, and ensuring that the direction of the trabeculae is consistent with the anatomical atlas through CT value gradient interpolation; closed-loop error correction: when the Hausdorff distance between the reconstructed model and the intraoperative O-arm data is >0.4mm, dynamically adjusting the voxel step size to 0.6mm.

[0010] Furthermore, in the biomechanical verification module, a fiber optic strain sensing network is implanted in a biomimetic layout: multiple sets of sensing units are arranged at an angle of 15°±5° along the axial direction in the principal stress zone of the neck, with a trabecular density >80 trabecular bones / cm². Each set contains three cascaded gratings with a grating length of 2mm, and adjacent units are 0.5mm apart. A miniature pressure sensor array with a range of 0-1000N and an accuracy of 0.1%FS is fused with the fiber optic network in real time using the LSTM algorithm to establish a joint kinematics-strain-pressure mapping model and eliminate soft tissue deformation interference. When the standard deviation of strain data for five consecutive frames is >30με, the laser temperature compensation module integrated into the positioning base is triggered. The temperature drift is corrected by scanning with a 2μm wavelength laser. The verification module also includes dynamic safety criteria: in addition to the maximum strain in the key area being ≤1500με, a dual threshold of trabecular strain rate change slope <10με / s and cortical-cancellous bone strain gradient ≤50με / mm is added to achieve early micro-motion warning of postoperative bone integration through distributed sensing.

[0011] Furthermore, the biomechanical verification module judges whether the ankle joint is qualified after surgery as follows: the maximum strain in the key area is ≤1500με and the strain gradient change is ≤50με / mm.

[0012] Furthermore, the "surgical plan-postoperative outcome" database stores information including preoperative parameters such as patient bone structure size and lesion type, surgical plan parameters such as osteotomy angle and displacement distance, and postoperative outcome data such as postoperative CT images, joint surface pressure distribution data, and force line deviation data. The parameter recommendation accuracy of the random forest algorithm is ≥85%.

[0013] Furthermore, the process of reproducing the surgical plan using the 3D reconstruction and surgical plan reproduction module with dynamic simulation mechanism includes: S1: Input osteotomy position, direction, and rotation angle parameters; S2: Bone block rotation or displacement is achieved through a flexible hinge; S3: Secure the bone block using a graduated locking assembly; S4: Generates a postoperative predictive 3D bone model, and displays the deviation between the actual parameters and the set parameters in real time during the operation.

[0014] Furthermore, the intelligent iterative optimization module can also access historical similar case data in the "surgical plan - postoperative effect" database to provide doctors with experience in adjusting the plan; the bone model is used to provide the mechanical basis for preoperative plan evaluation and doctor-patient communication, intraoperative operation guidance, and postoperative rehabilitation plan development.

[0015] The technical solution of this invention achieves full-process coverage: integrating preoperative planning, scheme simulation, biomechanical verification, intraoperative guidance, and postoperative learning, providing a closed-loop solution for ankle surgery. Dynamic and quantifiable: Through dynamically operable models and embedded sensor networks, surgical schemes are elevated from "morphological simulation" to "biomechanical verification," with quantitative evaluation indicators (such as strain and strain gradient) providing objective basis for scheme optimization. Highly personalized and intelligent: From images to physical models and intraoperative guides, everything is based on individual patient data; the intelligent learning module continuously accumulates clinical experience, assisting doctors in making better decisions. Intuitive and easy to use: The physical model combined with visualized data output greatly improves doctor-patient communication and doctors' understanding of complex three-dimensional surgery. This system can not only reproduce the individualized ankle skeletal anatomy of patients with high fidelity, but also dynamically simulate surgical operations such as osteotomy and correction, and evaluate the impact of surgical schemes on the mechanical environment of bone structure in real time and at a microscopic level through embedded sensor networks. Finally, through intelligent algorithms, experience is accumulated to provide optimization suggestions for new cases. Attached Figure Description

[0016] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 A system for 3D printing an ABS bone model of the ankle joint is shown; Figure 2 A flowchart illustrating the reenactment of the surgical procedure is shown; The above figures include the following reference numerals: 10, Multimodal image data fusion and preprocessing module; 20, 3D reconstruction and surgical plan reproduction module with dynamic simulation mechanism; 30, Biomechanical verification module; 40, Personalized adaptation and intraoperative guidance component; 50, Intelligent iterative optimization module. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the present invention or its application or use. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0018] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments 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.

[0019] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention. It should also be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale. Any specific values ​​in all examples shown and discussed herein should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters in the following drawings denote similar items; therefore, once an item is defined in one drawing, it need not be further discussed in subsequent drawings.

[0020] like Figure 1-2As shown: The multimodal image data fusion and preprocessing module 10 acquires high-resolution CT images (0.625mm slice thickness), MRI T1 / T2 weighted images (1.0mm slice thickness), and standard full-length weight-bearing X-ray images of the lower limb (anteroposterior and lateral views) in the same position. After all DICOM format data are imported, isotropic resampling to 0.5mm³ voxel space is performed and grayscale values ​​are normalized. The key step is for a senior physician to confirm the precise coordinates of four key anatomical landmarks on the 3D CT view, namely the medial malleolus tip, lateral malleolus tip, the center of the talus articular surface, and the calcaneal suspensory process, as prompted by the system, to establish a patient-specific spatial reference system. Subsequently, the system uses the SIFT algorithm in parallel to extract cortical bone edge feature points from CT images and bone-bone interface feature points from MRI images, and combines a specialized edge detection algorithm to extract feature points from X-ray images. After coarse alignment using the initial affine transformation matrix calculated from the calibrated keypoints, the KD-Tree nearest neighbor matching algorithm is used to find high-confidence feature point pairs between multimodal images. Weighted filtering is then performed with the keypoints as the center and a radius of 5mm, eliminating outliers with a deviation greater than 0.3mm from the RANSAC-estimated master model, ultimately retaining at least 150 reliable matching point pairs. To correct for local deformation caused by differences in soft tissue condition, a simplified ankle joint finite element model is constructed, simulating the surrounding soft tissue as a hyperelastic material. A dynamic deformation field is derived using CT as a reference state. This deformation field is applied to resample and fine-tune MRI images using bicubic B-spline interpolation, while X-ray images are simulated for projection-direction compression deformation. Registration quality is validated with a root mean square error (RMSE) ≤ 0.05mm and a structural similarity index (SSIM) ≥ 0.95 as the closed-loop verification standard. If the standards are not met, parameters are automatically adjusted and the iteration is repeated. In the bone tissue segmentation stage, the system collaboratively executes adaptive threshold segmentation and deep learning semantic segmentation: The former divides bone tissue into cortical bone (1000-1500 HU), outer cancellous bone (600-1000 HU), and inner cancellous bone (200-600 HU) based on the HU value of CT. For each layer, the initial threshold is calculated using the Otsu algorithm, and then a three-dimensional region growth algorithm is started for iterative optimization until the change in the number of growth voxels is less than 0.1% for three consecutive times. Then, artifacts are removed by three-dimensional morphological opening operation and verification using pre-stored anatomical atlas templates. The latter uses the registered three-modal images as a three-channel input to an improved U-Net network. A cross-modal attention gating module is embedded in the network encoder to dynamically allocate the weights of each modality. In the decoding stage, a composite loss function combining standard Dice loss, cross-entropy loss, and anatomical structure constraint loss based on Hausdorff distance is introduced for optimization. After binarization and morphological processing, the logical intersection of the network output and the threshold segmentation result is taken as the final bone tissue region. Inconsistencies are arbitrated by the doctor.

[0021] After preprocessing, the 3D reconstruction and surgical plan reproduction module 20 with dynamic simulation mechanism performs 3D reconstruction and surgical plan reproduction. The specific steps are as follows: An improved Marching Cubes algorithm is used to perform 3D reconstruction on volumetric data containing accurate HU values. To eliminate artifacts, the system applies a weighting factor of 1.2 to voxels in the tibial force line zone region shown on X-ray images to highlight the reconstruction, while applying a weighting factor of 0.8 to cortical bone voxels with HU>1000 in CT scans to smooth the stepped edges. Bilinear grayscale interpolation is used at the junctions to ensure that the curvature error of the talus trochlear articular surface is ≤0.2mm. To optimize surface smoothness, the algorithm adopts a layered slicing strategy. When the SSIM of adjacent slices is ≥0.92, the isosurface extraction threshold for the cancellous bone region is set between 200-800. The HU benchmark is adjusted by ±15HU to find local optima, thereby reducing serrated artifacts in areas such as the medial malleolus. The generated initial triangular mesh is further optimized by anatomical constraints. Specifically, in stress concentration areas such as the talus neck and the distal tibial weight-bearing area, the mesh is refined to a side length of approximately 0.3 mm through Loop subdivision. In non-critical areas such as the calcaneal suspensory process, coplanar meshes are merged using an edge-folding algorithm to reduce redundancy. At the same time, the gradient direction of the HU value in the original volume data is used to guide the mesh orientation to implicitly contain information about the trabecular bone structure. Finally, closed-loop error correction is performed, and the reconstructed model is compared with the "gold standard" model in three dimensions. If the maximum Hausdorff distance is >0.4 mm, the algorithm step size is dynamically adjusted from 1 voxel to 0.6 voxels for oversampling reconstruction.

[0022] Based on this high-precision 3D model, a detachable dynamic simulation mechanism is integrated during the design phase: According to the preoperative osteotomy plan (such as a medial distal tibial wedge osteotomy), polyurethane elastic hinges are designed on both sides of the predetermined osteotomy line, with their rotation center precisely set at the planned orthopedic center. The elastic modulus is calibrated to simulate the physiological constraints of the ligaments. A locking screw mechanism with precision threads and a dial is designed on the opposite side of the hinges to connect the proximal and distal bone blocks, enabling displacement accuracy of 0.1 mm and rotation reading and fixation of 0.5°. The surgical procedure replication process is as follows: S1: The doctor first inputs preset parameters such as osteotomy position, direction, correction angle and translation distance into the software interface; S2: Then, the printed ABS model is physically cut along the pre-set micro-etched lines, and the elastic hinge is operated to rotate or translate the distal bone block around the axis to feel the physiological resistance feedback it provides. S3: After moving to the desired position, tighten the graduated locking screw to fix the bone block, and directly read the scale value as the actual simulation parameter; S4: The system obtains actual operation values ​​through built-in micro sensors or visual recognition, compares them with preset values ​​in real time, and displays the deviation with numbers and colors (green for matching, red for deviation). The model in this state is regarded as the postoperative prediction model.

[0023] The biomechanical verification module 30 performs quantitative mechanical evaluation on the simulated model. The construction and implantation of the sensor network are crucial: the fiber optic strain sensor network employs a biomimetic placement strategy. For example, based on the preoperative CT bone density map, sensor units are arranged along the principal stress path of the talus neck where the trabecular density is >80 bones / cm², at an angle of 15°±5° to the macroscopic direction of the trabecular bone. Each unit consists of three 2mm long gratings connected in series and encapsulated in a 0.3mm diameter flexible microtube. Multiple units are arranged in parallel at 0.5mm intervals to form a distributed array. Utilizing 3D printing pause printing technology, when printing reaches a predetermined depth, a robotic arm implants the sensor units into a predetermined trajectory and continues printing and encapsulation, achieving embedded measurement. A miniature pressure sensor array (such as FlexiForce, range 0-1000N, nonlinearity <0.1% FS) is also embedded into the corresponding positions of the tibiotalar and subtalar articular surfaces using pause printing technology. A laser force line indicator is installed on the proximal tibial base of the model, its direction adjusted according to the force line determined by preoperative X-ray. During data acquisition, a simulated load (e.g., 500N) is applied to the model at a rate of 10N / s, and a fiber optic demodulator calculates micro-strain at a sampling rate of 1000Hz, with pressure data acquired simultaneously. All raw data are input into a pre-trained LSTM fusion model, which learns the spatiotemporal mapping relationship between joint pressure and bone strain under composite loads, and inversely subtracts non-physiological interferences such as ABS material creep and temperature drift, outputting purified mechanical data. Based on this, the system calculates and displays in real time the thermal map of joint surface contact pressure, the stress / strain cloud map inside the bone, and the laser force line projection position, and performs an evaluation based on dynamic safety criteria: the main criterion requires the maximum strain in the critical area to be ≤1500με; the newly added microscopic criteria include requiring the slope of the trabecular strain rate change during the loading and holding phase to be <10με / s, and the strain gradient at the cortical-cancellous bone interface to be ≤50με / mm. The system also has a self-calibration function, continuously monitoring the standard deviation of the fiber optic signal. If the standard deviation of five consecutive frames of data is >30με while the load remains unchanged, it is determined that there is interference. The system immediately triggers the laser temperature compensation module on the positioning base to emit a 2μm wavelength laser for scanning compensation and issues an alert in the software interface.

[0024] To bridge the gap between preoperative planning and intraoperative procedures, the system provides personalized adaptation and intraoperative guidance components 40. Using the same set of 3D data to reconstruct the contour of the lateral skin surface of the patient's ankle joint, a negative-type matched flexible resin positioning base is designed and 3D printed. During surgery, placing the patient's foot ensures that the posture matches the model. Simultaneously, a guide plate with specific protrusions, holes, and laser engraving (such as "Osteotomy line: 15mm from the joint surface") is designed according to the surgical plan. Printed using biodegradable PLA material, it adheres to the bone surface during surgery, providing precise guidance for osteotomy, screw placement, and other procedures. Furthermore, the material gradually degrades within the body.

[0025] Finally, all case data are aggregated into the intelligent iterative optimization module 50. The system automatically generates a structured record for each patient, including preoperative parameters such as age, bone density, and distal tibial articular surface angle; surgical plan parameters such as osteotomy type and correction angle; and postoperative outcome data such as CT scans at 6 months and 1 year postoperatively, articular surface pressure distribution, and force line deviation. This data is then anonymized and stored in the "Surgical Plan - Postoperative Outcome" database. When new patient data is input, the system calls a random forest regression model, using preoperative parameters and surgical type as features and historical high-quality cases as targets for training. This model outputs the optimal range of surgical parameters for new patients (e.g., correction angle 7°-9°, confidence ≥85%) and provides details of 3-5 most similar successful cases for reference. The system continuously retrains using new case data to continuously improve the accuracy of the recommendations.

[0026] In the description of this invention, it should be understood that the orientation or positional relationship indicated by directional terms such as "front, back, up, down, left, right", "horizontal, vertical, horizontal" and "top, bottom" is generally based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing this invention and simplifying the description. Unless otherwise stated, these directional terms do not indicate or imply that the device or element referred to must have a specific orientation or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on the scope of protection of this invention; the directional terms "inner" and "outer" refer to the inner and outer contours relative to the outline of each component itself.

[0027] For ease of description, spatial relative terms such as "above," "on top of," "on the upper surface of," "above," etc., are used herein to describe the spatial positional relationship of a device or feature as shown in the figures to other devices or features. It should be understood that spatial relative terms are intended to encompass different orientations in use or operation beyond the orientation of the device as described in the figures. For example, if the device in the figures were inverted, a device described as "above" or "on top of" other devices or structures would subsequently be positioned as "below" or "under" other devices or structures. Thus, the exemplary term "above" can include both "above" and "below." The device may also be positioned in other different ways (rotated 90 degrees or in other orientations), and the spatial relative descriptions used herein will be interpreted accordingly.

[0028] Furthermore, it should be noted that the use of terms such as "first" and "second" to define components is merely for the purpose of distinguishing the corresponding components. Unless otherwise stated, the above terms have no special meaning and therefore should not be construed as limiting the scope of protection of this invention.

[0029] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A 3D-printed ABS bone model of an ankle joint, characterized in that, include: The multimodal image data fusion and preprocessing module is used to acquire CT images, MRI images and weight-bearing X-ray films of the patient's ankle joint. The multimodal data is fused through an image registration algorithm, and the bone tissue region is extracted using an adaptive threshold segmentation algorithm. Key anatomical landmarks of the ankle joint are introduced during the registration process: medial malleolus tip, lateral malleolus tip, center point of the articular surface of the talus, and calcaneal talus process. The fusion process also integrates a semantic segmentation network based on deep learning to automatically identify and correct soft tissue interference in the images. A 3D reconstruction and surgical plan reproduction module with dynamic simulation mechanism. This module generates a preoperative 3D bone model of the ankle joint based on preprocessed bone tissue data, and integrates a detachable dynamic simulation mechanism in the preoperative 3D bone model. The dynamic simulation mechanism includes an elastic hinge and a scaled locking component. The biomechanical verification module includes: a miniature pressure sensor array embedded in the joint contact surface; a laser force line indicator placed on the outside of the model; and a fiber optic strain sensing network that works in conjunction with the miniature pressure sensor array. The strain sensors are embedded along the trabecular bone of the ankle joint to collect real-time microscopic strain distribution data of the bone model under simulated loads. The strain signals are converted into a stress cloud map of the bone structure using a fiber optic demodulator. Simultaneously, the sensing network and the laser force line indicator work in tandem to generate a joint surface pressure distribution heatmap, and additionally output the stress concentration area, maximum strain value, and strain gradient change curve at the trabecular bone level. Personalized fit and intraoperative guidance components include a personalized positioning base 3D printed based on the patient's skin contour data around the ankle joint, and raised guide markers printed using biodegradable PLA material with surgical operation prompts engraved on the surface. The intelligent iterative optimization module is used to establish a "surgical plan-postoperative effect" database. It analyzes the correlation between preoperative parameters, surgical plan parameters and postoperative effects through random forest algorithm, and automatically recommends the optimal range of surgical parameters for new patients.

2. The 3D-printed ankle joint ABS bone model according to claim 1, characterized in that, The image registration algorithm includes: firstly, extracting the edge feature points of ankle bone tissue from multimodal images using the SIFT algorithm, and then selecting effective feature points by combining key anatomical landmarks such as the medial malleolus tip; secondly, generating a dynamic deformation field using finite element deformation simulation technology to fine-tune the local image; and finally, using RMSE≤0.05mm and SSIM≥0.95 after registration as the qualified standard.

3. The 3D-printed ankle joint ABS bone model according to claim 1, characterized in that, The adaptive threshold segmentation algorithm includes: dividing bone tissue into three layers based on the HU value of CT images: cortical bone (1000-1500 HU), outer cancellous bone (600-1000 HU), and inner cancellous bone (200-600 HU), and modeling them; calculating the initial threshold using the Otsu algorithm, and iteratively optimizing the threshold using a region growing algorithm; finally, verifying the segmentation results based on the anatomical constraints of the ankle joint and removing false pixel blocks.

4. The 3D-printed ankle joint ABS bone model according to claim 1, characterized in that, The deep learning-based semantic segmentation network includes: normalized CT, MRI, and heavy-duty X-ray films as three-channel inputs; embedding a cross-modal attention mechanism in the encoding stage to assign dynamic weights to different modalities; introducing an anatomical structure constraint loss function in the decoding stage, which is combined with Dice and cross-entropy loss functions for optimization; and correcting soft tissue interference through morphological processing and connected region analysis after output.

5. The 3D-printed ankle joint ABS bone model according to claim 1, characterized in that, The 3D reconstruction and surgical plan reproduction module with dynamic simulation mechanism uses an improved Marching Cubes algorithm to generate a 3D bone model, specifically including: Cross-modal grayscale compensation: Weighting coefficients of 0.8 / 1.2 are applied to the CT cortex and X-ray force line areas respectively, and modal artifacts are eliminated by bilinear interpolation to make the curvature error of the talus trochlear articular surface ≤0.2mm; Dynamic threshold iteration: Based on SSIM≥0.92 of adjacent slices, the threshold of HU for cancellous bone fluctuates by ±15HU from 200 to 800 to reduce serrated artifacts at the tip of the medial malleolus. Anatomical constraint mesh optimization: The triangular facets in the stress zone of the talus neck are densified to 0.3 mm, and the coplanar meshes in the calcaneal load-bearing protuberance zone are merged to reduce redundancy. At the same time, CT value gradient interpolation is used to ensure that the direction of the trabecular bone is consistent with the anatomical atlas. Closed-loop error correction: When the Hausdorff distance between the reconstructed model and the intraoperative O-arm data is >0.4mm, the voxel step size is dynamically adjusted to 0.6mm.

6. The 3D-printed ankle joint ABS bone model according to claim 1, characterized in that, In the biomechanical verification module, a fiber optic strain sensing network is implanted in a biomimetic layout: multiple sets of sensing units are arranged at an angle of 15°±5° along the axial direction in the principal stress zone of the neck, with a trabecular density >80 trabecular bones / cm². Each set contains three cascaded gratings with a grating length of 2mm, and adjacent units are 0.5mm apart. A miniature pressure sensor array with a range of 0-1000N and an accuracy of 0.1%FS is fused with the fiber optic network in real time using the LSTM algorithm to establish a joint kinematics-strain-pressure mapping model and eliminate soft tissue deformation interference. When the standard deviation of strain data for five consecutive frames is >30με, the laser temperature compensation module integrated into the positioning base is triggered, and body temperature drift is corrected by scanning with a 2μm wavelength laser. The verification module also includes dynamic safety criteria: in addition to the maximum strain in the key area being ≤1500με, a dual threshold of trabecular strain rate change slope <10με / s and cortical-cancellous bone strain gradient ≤50με / mm is added to achieve early micro-motion warning of postoperative bone integration through distributed sensing.

7. The 3D-printed ankle joint ABS bone model according to claim 6, characterized in that, The biomechanical verification module judges whether the ankle joint is qualified after surgery according to the following criteria: maximum strain in the key area ≤1500με, strain gradient change ≤50με / mm.

8. The 3D-printed ankle joint ABS bone model according to claim 1, characterized in that, The "surgical plan-postoperative effect" database stores information including preoperative parameters such as patient bone structure size and lesion type, surgical plan parameters such as osteotomy angle and displacement distance, and postoperative effect data such as postoperative CT images, joint surface pressure distribution data, and force line deviation data. The parameter recommendation accuracy of the random forest algorithm is ≥85%.

9. The 3D-printed ankle joint ABS bone model according to claim 1, characterized in that, The process of reproducing the surgical plan by the 3D reconstruction and surgical plan reproduction module with dynamic simulation mechanism includes: S1: Input osteotomy position, direction, and rotation angle parameters; S2: Bone block rotation or displacement is achieved through a flexible hinge; S3: Secure the bone block using a graduated locking assembly; S4: Generates a postoperative predictive 3D bone model, and displays the deviation between the actual parameters and the set parameters in real time during the operation.

10. The 3D-printed ankle joint ABS bone model according to claim 1, characterized in that, The intelligent iterative optimization module can also call up historical similar case data in the "surgical plan-postoperative effect" database to provide doctors with experience in adjusting the plan; the bone model is used to provide the mechanical basis for preoperative plan evaluation and doctor-patient communication, intraoperative operation guidance, and postoperative rehabilitation plan formulation.

Images