Ankle joint fracture reduction assisting method under guidance of digital image
The digital image-guided ankle fracture reduction method solves the problems of process disconnect and inconsistent reduction in existing technologies, achieving efficient, safe and reliable reduction of ankle fractures, and reducing intraoperative differences and radiation risks.
Patent Information
- Application Number
- CN202511250370.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-03
- Publication Date
- 2025-12-12
AI Technical Summary
Existing technologies lack a unified image-guided procedure for ankle fracture reduction, leading to a disconnect between preoperative planning and intraoperative execution. The reduction quality is difficult to quantify objectively, and there is a lack of three-dimensional measurement and real-time navigation. The lack of integrated procedures results in significant differences in procedures between different surgeons and hospitals, leading to inconsistent reduction outcomes.
Digital image-guided methods, including setting CT scan parameters, performing 3D reconstruction, virtual repositioning, planning surgical approach and internal fixation device positions, combined with real-time image registration and fusion, generate visualized guiding images, forming an end-to-end replicable process. Adaptive CT parameters, weighted iterative nearest-point algorithm, and optical-electromagnetic dual-modal tracking system are used to ensure the stability and accuracy of intraoperative navigation.
It significantly reduces process differences between different surgeons and hospitals, achieves consistency in repositioning and fixation targets, improves the objectivity and reproducibility of repositioning quality, reduces the risk of vascular and nerve injury and radiation exposure, and ensures the safety and traceability of the surgery.
Smart Images

Figure CN121101753A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of medical technology, in particular to a digital image-guided ankle fracture reduction auxiliary method. BACKGROUND
[0002] The ankle joint is composed of the distal end of the tibia, the distal end of the fibula and the talus, forming a "mortise" structure, bearing complex axial weight and rotational shear force. Trauma-induced ankle fractures (including single / double / three ankle type and intra-articular fractures) are clinically common, and the quality of reduction and internal fixation stability are directly related to the joint surface smoothness, force line recovery and long-term function. The current clinical pathway mainly relies on closed reduction or open reduction guided by two-dimensional fluoroscopy, combined with preoperative CT evaluation; but there are still the following common problems in the aspect of preoperative and intraoperative information transmission and standardized process:
[0003] (1) Image parameters—reconstruction quality lacks front-end constraints: clinical settings of CT layer thickness, pitch, tube voltage / current, etc. rely on experience, which can easily cause bone cortex boundary blur or high noise, making the subsequent three-dimensional reconstruction and measurement accuracy uncontrollable, and the reduction baseline unstable.
[0004] (2) Three-dimensional measurement and reduction lack objective quantification chain: most systems only provide interactive alignment tools, lack of standardized process of "measurement first, then reduction", and quantitative indicators such as point distance and angle are not included in the reduction criteria, making it difficult to objectively reproduce "whether the reduction meets the standard".
[0005] (3) Preoperative planning and intraoperative execution are broken: the model and planning established preoperatively are difficult to be stably and intuitively mapped to real-time images intraoperatively, and the navigation screen updates lag or is unstable, making it difficult to form continuous guidance to instrument operation.
[0006] (4) Intraoperative real-time guidance is insufficient: it is difficult to grasp the spatial relationship relying only on two-dimensional fluoroscopy; even if intraoperative images are obtained, there is a lack of standard steps to register and fuse them with preoperative three-dimensional models, so that the virtual reduction result cannot be intuitively converted into intraoperative navigation.
[0007] (5) The process lacks integration of "preoperative-intraoperative" closed loop: in clinical work, image acquisition, reconstruction / reduction, planning, and intraoperative guidance are often performed separately, lacking a process that is connected in a fixed order, making it difficult to ensure repeatability across teams and cases.
[0008] In view of the above problems, a standardized method is needed with "three-dimensional reconstruction—virtual reduction—planning—intraoperative registration—real-time navigation" as the main line: parameter setting requirements are included in the image acquisition link; quantitative measurement is performed on the three-dimensional model and virtual reduction is performed accordingly; the model after reduction is used as a reference for surgical planning; real-time images are collected intraoperatively and registered and fused with the model after reduction, and finally continuous guidance is provided on the navigation interface. SUMMARY
[0009] In order to overcome the above technical defects, the purpose of the present application is to provide a digital image guided ankle fracture reduction auxiliary method, through the fixed sequence of steps S100 to S500, compared with the prior art, the links are scattered, and there is no unified main line; the preoperative planning is often disconnected with the intraoperative execution, the navigation screen and the operator operation cannot form stable mapping, in the scheme of the present application, a replicable end-to-end process is formed, which significantly reduces the process difference between different operators / hospitals, and the target, basis and result of reduction and fixation remain consistent before and during operation.
[0010] The application discloses a digital image guided ankle fracture reduction auxiliary method, characterized in that comprising the following steps:
[0011] Step S100: acquiring CT image data of the ankle fracture site of a patient, wherein the acquisition process comprises setting CT scanning parameters;
[0012] Step S200: importing the CT image data into a medical image processing software for three-dimensional reconstruction to obtain an original three-dimensional fracture model;
[0013] Step S300: measuring and virtually reducing the fracture block on the original three-dimensional fracture model to obtain a three-dimensional fracture model after reduction;
[0014] Step S400: planning a surgical approach and the placement position and angle of an internal fixator based on the three-dimensional fracture model after reduction;
[0015] Step S500: acquiring real-time image data during the operation, and registering and fusing the real-time image data with the three-dimensional fracture model after reduction to generate a visual guidance image, thereby providing real-time navigation for the operation.
[0016] Preferably, the CT scanning parameters are adaptively adjusted to meet the following conditions:
[0017]
[0018]
[0019]
[0020] wherein, is the layer thickness, is the scanning speed, is the pitch, U is the tube voltage, I is the tube current, and W is the body weight, is the floor function, and min() is the minimum value operator.
[0021] Preferably, the distance between the two corresponding feature points and is calculated in step S300 when measuring the original three-dimensional fracture model ; the included angle between the two corresponding vectors , is calculated in step S300; ;
[0022] wherein, is the distance between two points (mm), is the included angle (°).
[0023] Preferably, the surgical approach is determined based on the weighted score S in step S400:
[0024]
[0025] wherein A is the age score, B is the BMI score, C is the comminution score, D is the soft tissue injury grading score, and E is the underlying disease score, all of which are scored from 1 to 5; when S≤3, a percutaneous minimally invasive approach is selected, 3
[0026] Preferably, the internal fixation scheme is obtained by minimizing the objective function in step S400.
[0027] wherein, is the maximum equivalent force of the internal fixator (MPa), is the residual displacement of the fracture fragment after reduction (mm), is the bone-screw contact area, is the reference contact area; and satisfies , the angle between the screw axis and the fracture line , and the screw length , wherein is the thickness of the fracture fragment.
[0028] Preferably, the intraoperative imaging in step S500 satisfies: , , the spatial resolution is ≤0.15 mm, cm; wherein, is the low-dose single-frame dose, is the acquisition interval, is the diameter of the imaging field.
[0029] Preferably, the registration fusion in step S500 uses a weighted iterative closest point algorithm to solve the rotation matrix R and the translation vector :
[0030]
[0031] in, These are the feature points of the three-dimensional fracture model after reduction. For real-time image feature points, =0.8 (cortical bone) or 0.2 (cancellous bone), N is the number of matching points; As a termination condition, and when the residual at a certain point... Those that are removed should be discarded; the error in the coarse registration stage should be ≤2mm.
[0032] Preferably, in step S500, the instrument employs optical-electromagnetic dual-modal tracking and is fused with the image in time synchronization to meet the requirements. The distance from the tip of the instrument to the fracture surface is ;in, Let S be the coordinates of the instrument tip, and S be the set of fracture surface points on the three-dimensional fracture model after reduction.
[0033] Preferably, a dataset D is created postoperatively. Dataset D includes DICOM images, a three-dimensional fracture model after reduction, and surgical parameter logs. Its integrity is verified as follows:
[0034]
[0035] Where H is the hash value; the data has disaster recovery redundancy and a retention period of ≥15 years.
[0036] Preferably, the database adopts a consortium blockchain architecture, and data modifications require consensus from at least three nodes; cross-institutional analysis employs federated learning, with the parameters of the k-th node set as follows: The weight is Global aggregation parameters ;
[0037] Where K is the number of participating nodes; the validity criterion for the discovered association rules is statistical significance P≤0.01.
[0038] Compared with existing technologies, the above technical solution has the following advantages:
[0039] 1. This invention employs a fixed sequence of steps S100 to S500: setting CT parameters → obtaining an original three-dimensional fracture model through three-dimensional reconstruction → obtaining a three-dimensional fracture model after virtual reduction following quantitative measurement → planning the approach and internal fixation based on this model → acquiring real-time images during surgery and registering and fusing them with the reduced model, providing continuous guidance on the navigation interface. Compared to existing technologies, which are fragmented and lack a unified main thread, preoperative planning and intraoperative execution are often disconnected, and the navigation screen and surgeon's operation cannot form a stable mapping, this invention forms a replicable end-to-end process, significantly reducing process differences between different surgeons / hospitals, and ensuring that the goals, basis, and results of reduction and fixation remain consistent before and during surgery.
[0040] 2. This invention uses a formulaic parameter mapping (layer thickness) pitch The relationship between voltage U, current I, and body weight / bone mass is used to determine the minimum image quality requirements for reconstruction and navigation before acquisition (e.g., ...). =0.5 / 0.8mm to distinguish cortical / cancellous tissue); Compared with the existing technology that relies on technician experience to set parameters, resulting in blurred cortical boundaries and high noise, which brings uncertainty to subsequent segmentation and registration; therefore, in this solution, the "reconstruction quality" is moved to the controllable acquisition stage, reducing baseline fluctuations caused by parameter differences, and providing stable input for subsequent 3D reconstruction and registration;
[0041] 3. In this invention, the distance between points is first measured on the original three-dimensional fracture model. Vector angle After measuring the range, the residual displacement is used as the basis. As a standard indicator, existing technologies mostly rely on interactive "feeling" alignment, and the repositioning standard lacks objective thresholds and a consistent measurement chain; therefore, adopting this solution makes the repositioning criteria objective and standardized, which is convenient for in-hospital quality control and surgeon training, and the repositioning quality is "comparable" and "reproducible".
[0042] 4. This invention introduces an approach score S (weighted by five factors: age, BMI, degree of comminutedness, soft tissue grade, and underlying disease), and superimposes a three-dimensional projection of the vascular nerve bundle onto the repositioned model, setting a safety boundary (e.g., 3-4 mm) to prohibit entry into the pathway. In existing technologies, judgment is mainly based on experience according to classification / soft tissue, making it difficult to quantify multiple factors; it is also difficult to avoid vascular nerves in three dimensions under two-dimensional fluoroscopy. Therefore, this solution achieves "quantitative diversion + three-dimensional avoidance" to reduce the risk of vascular nerve injury, and the approach selection is interpretable and traceable.
[0043] 5. This invention is based on With the goal of uniting stress / Residual shift / Contact area gap Optimize; and apply geometric and biomechanical constraints (such as...) , , , In existing technologies, screw length / angle and plate type are often determined based on experience, and single indicators (such as "long enough / hard enough") are often the main focus, lacking multi-objective coordination. Therefore, by adopting this solution, stability and alignment are taken into account while meeting safety / geometric constraints, reducing the risk of complications caused by postoperative imbalance or excessive stiffness.
[0044] 6. This invention employs a two-stage registration process: coarse registration of anatomical points / principal axes (error ≤ 2mm) → weighted ICP (cortical). Loose texture =0.2), with RMSE≤0.3mm as the termination condition, and simultaneously using "3 "Rule-based outlier removal; In existing technologies, ICP is commonly used, which is sensitive to occlusion, artifacts and outliers, has inconsistent termination conditions, and insufficient registration stability; Therefore, this scheme results in faster convergence, less sensitivity to noise, more stable intraoperative navigation coordinates, and reliable mapping of resetting and planning to the image."
[0045] 7. In this invention, optics and electromagnetics are combined, and time synchronization is performed. ≤1ms, can be downweighted / switched in case of obstruction or interference, continuously outputting the actual distance between the instrument tip and the fracture surface. In existing technologies, most methods rely on single-mode tracking or simple redundancy, lacking quality control linked to synchronization / registration errors. Therefore, this solution significantly reduces the impact of "track loss / drift" on navigation. Indicators can directly guide safe procedures and improve the safety margin of surgery;
[0046] 8. In the procedure of this invention, low-dose sequences are used for localization (e.g., , ), , In existing technologies, repeated fluoroscopy or 3D acquisition lacks quantitative triggering and field-of-view control, which can easily lead to unnecessary dose accumulation. Therefore, by adopting this solution while maintaining registration / navigation requirements, it can significantly reduce ineffective radiation exposure, which is in line with the ALARA principle.
[0047] 9. In this invention, the distance from the instrument tip to the fracture surface is displayed in real time on the navigation interface. The data is classified according to thresholds (e.g., ≥1.0mm green, 0.5-1.0mm yellow, <0.5mm red), and pause / withdrawal suggestions or forced verification acquisition are triggered when necessary. In the existing technology, there is a lack of real-time quantification of three-dimensional distance under two-dimensional perspective, and more reliance is placed on tactile and experience judgment. Therefore, after adopting this solution, the approach process of key structures can be quantified and can be warned, reducing the probability of iatrogenic injury.
[0048] 10. In this invention, a complete preoperative / intraoperative / postoperative dataset D is constructed for unified calculation. As a completeness verification, it is combined with ≥15 years of redundant processes and audit logs; in the existing technology, PACS mostly only contains images; structured data such as models, registration errors, device trajectories, and optimization logs are stored in a scattered manner, making it difficult to obtain evidence and reproduce; therefore, after adopting the solution, evidence-level consistency and traceability are achieved, which facilitates medical quality control, dispute resolution and scientific research review;
[0049] 11. In this invention, data writing requires consensus from at least 3 nodes to ensure on-chain trustworthiness; multi-center approach... The data was aggregated, and P≤0.01 was used as the statistical validity criterion. In existing technologies, offline identification and sharing are often used, which is not reliable and efficient. Multicenter joint analysis lacks unified statistical access. Therefore, after adopting this approach, continuous learning and feedback to clinical pathways under the premise of compliance will lead to faster evidence formation and higher reliability. Attached Figure Description
[0050] Figure 1 This is a schematic diagram illustrating the steps of a digital image-guided method for assisting in the reduction of ankle fractures according to the present invention. Detailed Implementation
[0051] The advantages of the present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments.
[0052] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.
[0053] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The singular forms “a,” “the,” and “the” as used in this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.
[0054] It should be understood that although the terms first, second, third, etc., may be used in this disclosure to describe various information, such information should not be limited to these terms. These terms are used only to distinguish information of the same type from one another. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."
[0055] In the description of this invention, it should be understood that the terms "longitudinal", "lateral", "up", "down", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and 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. Therefore, they should not be construed as limitations on this invention.
[0056] In the description of this invention, unless otherwise specified and limited, it should be noted that the terms "installation", "connection" and "linking" should be interpreted broadly. For example, they can refer to mechanical or electrical connections, or internal connections between two components. They can be direct connections or indirect connections through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms according to the specific circumstances.
[0057] In the following description, suffixes such as "module," "part," or "unit" used to denote elements are used only for the convenience of the description of the invention and have no specific meaning in themselves. Therefore, "module" and "part" can be used interchangeably.
[0058] In this embodiment, refer to Figure 1 As shown, a digital image-guided method for assisting in the reduction of ankle fractures will be described in detail below.
[0059] Regarding the terms, coordinates, and symbols mentioned in this embodiment:
[0060] Three-dimensional coordinate system: using a fixed coordinate system of the patient's body. (Unit: mm), Image voxel coordinate system With navigation world coordinate system Unless otherwise specified, the default setting is... The Chinese expression.
[0061] Coordinates and Vectors: In three-dimensional coordinates (mm); , It is a three-dimensional vector; It is a Euclidean norm; " is the dot product.
[0062] Geometric quantities: Point-to-point distance (mm); δ is the angle between vectors (°); δ is the shortest distance between a point and a surface (mm).
[0063] CT parameters: Layer thickness (mm); The scanning speed is (mm / s). U is the pitch, U is the tube voltage (kV), I is the tube current (mA), and W is the weight (kg). To round down, min( ) is the minimum value operator.
[0064] Scoring and Optimization: Approach Scoring ; The maximum equivalent force of the internal fixation device (MPa); The residual displacement of the fracture fragments after reduction (mm). The bone-nail contact area. For reference contact area; .
[0065] Registration: R is the rotation matrix. It is a translation vector; { } model point set, { } represents the image point set; The weight is N; the number of matching points is N. For residuals; The mean; Standard deviation; This is the root mean square error.
[0066] Navigation and Synchronization: To track timestamps, For image timestamps, This is the synchronization error; S represents the coordinates of the instrument tip; S represents the set of fracture surface points.
[0067] Imaging dose: For low-dose single-frame dose ( ); The acquisition interval is (s / frame). The diameter of the imaging field of view (cm).
[0068] Data and Learning: D represents the dataset; Hash; K is the number of nodes; For local parameters; As weight, Global aggregation parameter; P is the statistical p-value.
[0069] Regarding hardware components:
[0070] Medical multi-slice spiral CT or dual-source C-arm CT (supports switching between low-dose and high-resolution modes, imaging field of view) =12-15cm.
[0071] Surgical navigation workstation (CPU ≥ 8 cores, memory ≥ 32GB, GPU ≥ 8GB video memory) is used for 3D reconstruction, registration and visualization.
[0072] Optical-electromagnetic dual-mode tracking system (optical camera ≥120fps, spatial accuracy ≤0.1mm; electromagnetic transmitter with metallic environment compensation, error ≤0.3mm).
[0073] Surgical instruments are equipped with reflective balls or electromagnetic sensors, and instrument identification codes.
[0074] 3D visualization display terminal (optional holographic projection or medical 4K display).
[0075] Safe power supply and network isolation modules (medical-grade UPS, isolation switches).
[0076] Hospital information system interface (HIS / EMR / PACS gateway, supporting HL7 or FHIR).
[0077] Software components:
[0078] Image processing and 3D reconstruction module (supports DICOM import, segmentation, denoising, surface reconstruction, and topology repair).
[0079] Reset and Planning Module (Measurement, Constraint Modeling, Approach Scoring, Internal Fixed Multi-Objective Optimization).
[0080] Intraoperative registration and instrument fusion module (coarse registration + weighted ICP, outlier removal, hand-eye calibration, time synchronization).
[0081] Visualization and Interaction Module (Distance) (Real-time display, path prediction, error heatmap, dosage indication)
[0082] Data and security module (D construction, hash H, permission auditing, consortium blockchain and federated learning adaptation layer).
[0083] Data acquisition and adaptive CT parameters
[0084] Patient positioning within the scanning area:
[0085] The patient should be placed in a supine or prone position with the affected limb immobilized, and the foot and ankle kept in a neutral position to avoid external or internal rotation. The scanning range should cover the distal tibia and fibula to the talus and calcaneus, extending upwards by 5-8 cm to facilitate internal fixation planning.
[0086] Parameter adaptive setting:
[0087] Layer thickness, pitch, tube voltage, and tube current are set automatically / semi-automatically according to the following formula:
[0088]
[0089]
[0090]
[0091] in, Where W is the scanning speed and W is the body weight. The system assigns regions based on the pre-scan density map, or the technician determines the region before starting the actual scan.
[0092] Dosage and quality control:
[0093] Adopting the ALARA principle, automatic recording and If metal implants or plaster casts are detected causing artifacts, the system will prompt you to enable the metal artifact suppression sequence and reduce its size. .
[0094] 3D Reconstruction and Fracture Model Generation
[0095] Preprocessing:
[0096] Denoising: Anisotropic diffusion and Gaussian filtering are combined for edge-preserving denoising.
[0097] Voxel calibration: Switch to And maintain the voxel size and orientation matrix.
[0098] Metal artifact suppression: projection domain interpolation + iterative reconstruction or frequency domain filtering to remove stripes.
[0099] Segmentation and Topology Repair:
[0100] Threshold segmentation: The cortical bone and cancellous bone are initially segmented based on the HU value range.
[0101] Regional growth / graphic cutting: refinement along the bone cortex boundary.
[0102] Hole filling and self-intersection removal: based on mesh Boolean and Laplacian smoothing, preserving edge and corner constraints.
[0103] Surface reconstruction:
[0104] Marching Cubes are preferred to generate isosurfaces; to ensure stable registration later, sharpness information is preserved at key edges (normal variation threshold of 30°). The original 3D fracture model (STL / PLY) is output, and candidate point sets for the proximal / distal fracture lines are automatically extracted.
[0105] Measurement, virtual repositioning and anatomical alignment
[0106] Geometric vectors:
[0107] The distance between two feature points and the angle between their vectors are respectively: , ;in and lie in ; , It is the feature tangential or principal direction vector.
[0108] Reset operation and constraints:
[0109] Transformation representation: Define a rigid body transformation for each fracture fragment ( ,J).
[0110] Constraint type:
[0111] Joint surface "matching" constraint: concave-convex matching error ≤ 1-2mm;
[0112] Joint line smoothness constraint: Curvature abrupt change threshold ≤ 0.2 ;
[0113] Anatomical boundary constraints: The overlap between the fibular notch and the outer edge of the tibia is ≥85%.
[0114] Interactive tools:
[0115] Translate / rotate the handle and snap it to the target surface; provide a "mirror opposite side" as a reference (if available).
[0116] Initial assessment of reset quality:
[0117] Before proceeding with surgical planning, the displaced bone fragments were initially compressed. .
[0118] Surgical approach scoring and internal fixation optimization
[0119] Approach scoring and path avoidance:
[0120] Approach scoring formula The criteria for assessment included A (age), BBB (BMI), CCC (comminutedness), DDD (soft tissue grade), and EEE (underlying disease), all scored from 1 to 5. The system superimposed the three-dimensional projection of the vascular bundles from CTA / MRA onto the post-reduction three-dimensional fracture model, imposing "no-entry zones" and a 2–4 mm safety boundary on the planned path.
[0121] Screw / plate parameterization
[0122] Candidate screws should have a diameter of 3.5-4.5 mm and a length of... .
[0123] Angle of incidence along the axis For the fracture line.
[0124] The steel plate is selected via the outer / rear outer path; the hole position sequence and screw tilt angle are discretized to a 5° granularity to accelerate the search.
[0125] Multi-objective optimization problem:
[0126] Objectives and constraints:
[0127] ;
[0128] ;
[0129] in, It can be estimated by simplified finite element approximation or empirical stress model (bone-nail contact with Hertz approximation to correct the contact area). The standard contact area is defined for the same bone density. The solver includes, but is not limited to, genetic algorithms (population 60-120, iterations 200-500 generations), and can also be compared with parallel particle swarm optimization / Bayesian optimization, retaining Pareto front alternatives.
[0130] Results verification and visualization:
[0131] The system generates real-time graphs of residual displacement, contact area isosurfaces, and stress cloud diagrams, and displays "red / yellow / green" indicators in the UI to indicate whether constraints are met. It also outputs printable reports (including parameter tables and 3D mini-plots).
[0132] Intraoperative imaging, registration and instrument fusion
[0133] Imaging strategy and triggering:
[0134] Reset Positioning: Low-Dose Mode, Dosage and Frequency ( , =3s / frame).
[0135] Reset verification: High resolution mode (spatial resolution ≤ 0.15mm), cm, enabling metal artifact suppression).
[0136] Triggering logic: When the instrument tip A high-resolution acquisition will be automatically triggered when the deviation is less than 2mm or the positional deviation exceeds the threshold.
[0137] Coarse registration and weighted ICP:
[0138] Coarse registration employs anatomical landmark / principal axis fitting or rapid matching based on surface normals, ensuring an error ≤2mm before proceeding to fine registration. Fine registration solution:
[0139]
[0140] in, =0.8 (cortical bone) or 0.2 (cancellous bone). Abnormal point removal: Remove at the specified time. Termination condition: RMSE ≤ 0.3mm or iteration difference < 0.01mm. (Numerical solution: SVD analytical method or quaternion least squares (Arun / Umeyama) is used, implemented in parallel on GPU to ensure sub-second response.)
[0141] Hand-eye calibration and time synchronization:
[0142] Hand-eye calibration: The coordinates of the optical and electromagnetic coordinate systems are calculated separately. The external parameters were calibrated using the Tsai-Lenz or Park method, with residuals ≤0.2mm.
[0143] Time synchronization:
[0144]
[0145] The system synchronizes its time with PTP (IEE 1588) and inserts a unified timestamp at the data access layer.
[0146] Device-model fusion and safety thresholds:
[0147] Distance calculation:
[0148] ;
[0149] It uses KD-Tree / Octree acceleration and has a refresh rate of 50-100Hz.
[0150] Safety threshold: Enter "alert" state. Enter a "dangerous" state and penalize the suspension / withdrawal of recommendations.
[0151] UI prompts: Path prediction, fracture line proximity bar, minimum spacing of vascular and nerve bundles (automatically add 2=4mm safety boundary).
[0152] Sources of error, budget and quality control
[0153] Sources of error: imaging noise and voxel discretization error; segmentation and meshing approximation error; initial deviation of coarse registration; tracking system drift and metal interference; local deformation caused by soft tissue traction.
[0154] Error budget (example): Reconstruction error ≤ 0.1-0.2 mm; coarse registration residual ≤ 2.0 mm; fine registration RMSE ≤ 0.3 mm; tracking system overall ≤ 0.1-0.3 mm; UI display and sampling delay < 50 ms. Error superposition uses the L2 norm to approximate the upper bound of the error, and key indicators are monitored and recorded in real time during the operation.
[0155] Quality control process:
[0156] Preoperative: Geometric phantom calibration (rod / rod / stepped body); Intraoperative: Snapshot verification (alignment difference of key anatomical points <0.5–0.8 mm); Postoperative: Image review and statistical summary to form the department's monthly KPI (RMSE distribution, Distribution, reset success rate, etc.
[0157] Data closed loop, storage and compliance analysis
[0158] Data sets and hashes:
[0159] Dataset D was constructed, comprising preoperative / intraoperative / postoperative images (DICOM), a 3D fracture model after reduction (STL / PLY), instrument trajectory, and optimization parameters. Curves, RMSE convergence curves, etc.; integrity verification And write it to the audit log.
[0160] Security and Retention: Tiered access control and multi-factor authentication; off-site disaster recovery and WORM (Write Once Read Many) storage; retention period ≥ 15 years; HL7 / FHIR interface and RACS / HIS integration.
[0161] Consortium blockchains and federated learning:
[0162] Consortium blockchain: Data writing or modification is performed after consensus is reached by no fewer than 3 nodes;
[0163] Federated Learning Aggregation: ;
[0164] Statistics: After pooling multiple centers, the rule is considered valid if P ≤ 0.01, and multiple comparison corrections (such as the BH method) should be performed if necessary.
[0165] Safety and Abnormal Handling
[0166] Mismatch warning: When RMSE > 0.3mm or >1ms for 3 frames, the UI turns red and suggests reverting to the perspective guide.
[0167] Tracking obstructed: When electromagnetic distortion is caused by metallic interference, the electromagnetic data is automatically reduced in weight, and optical navigation is temporarily used only; when the obstruction is severe, the user is prompted to adjust the viewing angle.
[0168] Dosage limit: When the preset cumulative dose reaches 80%, an alternative verification method (ultrasound / tactile feedback / mechanical gauge) will be prompted.
[0169] Data anomaly: If the hash H is inconsistent or the on-chain consensus fails, the discharge archiving process will be blocked and the information department will be notified.
[0170] Specific implementation methods:
[0171] Example 1 (Percutaneous minimally invasive approach, small trauma scenario)
[0172] Patient preoperative assessment:
[0173] Basic information: Male, 35 years old, height 175cm, weight 60kg (W=60), BMI (Body Mass Index) ≈19.6; no underlying diseases; closed trauma; mild comminuted injury, grade I soft tissue injury.
[0174] Preoperative imaging: Routine X-ray and CT scans initially indicated a fracture of the lateral malleolus with slight displacement, but no obvious articular surface collapse.
[0175] Preoperative goals: Minimize invasiveness and reduce dosage; minimize residual displacement after repositioning. ≤0.3mm; Minimum safe distance between instrument and fracture surface Internal fixation stress ≤800MPa; bone-nail contact area ratio .
[0176] Image acquisition:
[0177] Scan speed setting =100mm / s; according to ;
[0178] Weight-voltage mapping: ;
[0179] Partition thickness / current: , ;
[0180] Visual field: Approximately 70 mm proximally from the center of the ankle joint; To cover the medial and lateral malleoli and the talus dome.
[0181] Quality control: Metal artifact suppression was not triggered; pre-scan showed mild soft tissue swelling, which did not affect bone boundary identification.
[0182] 3D Reconstruction:
[0183] Preprocessing: Anisotropic diffusion (8 iterations) + median filtering (kernel 3); Voxel calibration to world coordinate system ; HU threshold initial segmentation of cortical bone [-250, 3000].
[0184] Segmentation and Repair: Region growing + graph cutting; Hole filling (maximum hole diameter 1.0mm); Mesh reconstruction using MarchingCubes (voxel step size 0.5mm); Normal smoothing ( (10 iterations).
[0185] Output: Original 3D fracture model (STL, face count ≈ 220k); Automatic extraction of feature point sets (36) at both ends of the fracture line.
[0186] Measurement and Virtual Reset:
[0187] Initial geometric error: Select the corresponding point and , ;
[0188] Corresponding bone block main direction , included angle ;in Point-to-point distance (mm) The angle between the vectors is (°).
[0189] Interactive reduction: applying rigid body transformation to the distal fibular fragment. ), applied to the small bone fragments of the medial malleolus The constraint joint surface mating error is ≤2mm, and the abnormal curvature of the joint line flatness is <0.2. .
[0190] Initial assessment of repositioning: From 1.2mm to 0.82mm (generated from the three-dimensional fracture model after reduction).
[0191] Surgical planning:
[0192] Approach score selection → Select a percutaneous minimally invasive approach; use CTA to generate a vascular bundle projection and apply a 3mm safety boundary no-entry zone.
[0193] Internal fixation parameterization and optimization:
[0194] Screw candidate: Diameter 4.0mm; Length The thickness of the bone fragment was measured. =25mm→ (2.5mm from the walking distance).
[0195] angle: (Step length 2.5°).
[0196] Objectives and constraints:
[0197]
[0198]
[0199] Solver: Genetic algorithm (population 80, 300 generations, crossover rate 0.8, mutation rate 0.1); Parallel particle swarm optimization as control (population 50, 200 iterations).
[0200] Convergence results (convergence in the 214th generation): =85°, =40mm (two pieces), prediction =620MPa, =0.84, after the second registration iteration =0.24.
[0201] Visual verification: The residual displacement heat map shows a "green zone" with continuous isosurfaces of the contact area; the blood vessel restricted zone is not crossed.
[0202] Intraoperative navigation:
[0203] Imaging:
[0204] Positioning phase: low dose , A total of 18 frames were captured (≈54s).
[0205] Validation phase: High-resolution reconstruction (spatial resolution ≤ 0.15mm), =12cm), collected twice.
[0206] Registration:
[0207] Coarse registration error: 1.7mm (anatomical landmarks + principal axis fitting).
[0208] Weighted ICP:
[0209] .
[0210] Outlier removal rules: ;
[0211] RMSE iteration: 0.62→0.41→0.33→0.26mm (terminated).
[0212] Device fusion and safety thresholds:
[0213] Time synchronization: ≤0.7ms;
[0214] Tip distance Real-time 50-80Hz refresh rate, throughout. =0.9mm; =0.24mm.
[0215] Postoperative verification and data closed loop:
[0216] Immediate postoperative CT scan showed good alignment; internal fixation position was consistent with the plan; and there were no steps on the articular surfaces.
[0217] Construct dataset D (DICOM, STL, instrument trajectory, optimization log, ... (curve, RMSE convergence plot); .
[0218] On-chain consortium blockchain: Local weights Case sample size; federal aggregation will be conducted after the multi-center study is completed.
[0219] Example 2 (Standard open access, heavy crushing scenario)
[0220] Patient preoperative assessment:
[0221] Male, 65 years old, height 170cm, weight 75kg (W=75), BMI≈25; type 2 diabetes (controlled with oral medication); severe comminuted injury, grade III soft tissue injury.
[0222] Objective: To ensure stable fixation to the joint surface under severe crushing conditions; with similar constraints. ≤0.3mm, MPa .
[0223] Image Acquisition (Upper limit);
[0224] Weight - Voltage: ;
[0225] The layer thickness / current of each partition is the same as in Example 1; Metal artifact suppression preset enabled (diabetic patients may have a history of early metal treatment, although the metal was not detected in the end, but the strategy is retained).
[0226] 3D Reconstruction: Noise control enhancement (diffusion iterations 12 times); graph cut and cortical bone edge constraints strengthened; topology repair performed after mesh triangulation (gap closure threshold 1.5mm). Output: Original 3D fracture model (face count ≈ 310k); automatic identification of feature lines of tripolecular complex fractures, extraction of 58 feature points.
[0227] Measurement and Virtual Reset:
[0228] Initial misalignment range: between different bone fragments =10.2~22.7mm; =18°~41°.
[0229] Reduction strategy: First, align the talofibular syndesmosis, then the medial malleolus, and finally adjust the contact surface of the posterior malleolus.
[0230] Preliminary assessment: The three-dimensional fracture model achieved satisfactory results after reduction. ≈0.9mm, joint surface step <1.0mm (still needs further optimization).
[0231] Surgical planning:
[0232] Entry score >4.5→Standard open approach, vascular bundle near posterolateral, with a safety boundary of 4mm.
[0233] Internal fixation optimization:
[0234] Steel plate: Externally anatomically shaped steel plate (7 holes, discrete hole angles within ±15° range);
[0235] Screws: 4.0 / 4.5mm diameter optional; for each key bone segment. =22-28mm→ ∈[33,56]mm.
[0236] Optimization: Genetic algorithm (population 120, generations 400), parallel NSGA-II to generate Pareto solution set for the operator to choose from.
[0237] Constraint monitoring: If >800MPa or This means reverting to the previous generation's scheme; adding a "disabled" mark to the pores adjacent to the medial blood vessels.
[0238] Convergent solution: outer steel; plate + multiple screws; key screw 1: , Key screw 2: , ,predict =710MPa; =0.86; after two fine-tuning mm.
[0239] Intraoperative navigation:
[0240] Imaging, low-dose localization: Acquire 24 frames (≈72s); High-resolution verification: 3 times (each time) =12cm, metal artifact suppression enabled).
[0241] Registration and robustness: Registration error: 2.0mm (acceptable under heavy crushing); Weighted ICP: RMSE trajectory 0.74→0.51→0.37→0.29mm, outlier rejection rate ≈6%; Hand-eye calibration residual: 0.21mm; ≤0.9ms.
[0242] Device fusion:
[0243] Real-time monitoring: The "alert" is triggered twice, each time for less than 3 seconds; the path prediction detours around the outer restricted area, avoiding the vasovagal bundle.
[0244] Dosage and quality control: Cumulative dose ≈ 78% of the path threshold; the system automatically prompts to reduce one high-resolution acquisition and switch to low dose + tactile confirmation;
[0245] Scope KPI: mm; =710MPa; Contact area ratio 0.86; RMSE=0.29mm; .
[0246] Postoperative verification and data closed loop:
[0247] Postoperative CT scan: The joint surface is smooth and without steps; the plate fits well.
[0248] Dataset D is archived and generated ;
[0249] Consortium blockchain ledger: parameters of this center With weight Reporting to higher authorities, awaiting multi-center approval. Aggregation; statistical test p ≤ 0.01 was performed before inclusion in the clinical evidence-based database.
[0250] Example 3 (Trimalleolar fracture)
[0251] Patient preoperative assessment:
[0252] Female, 48 years old, height 162 cm, weight 66 kg (W=66), BMI≈25.1; hypertension (well controlled with medication); trimalleolar fracture, severe comminuted fracture, grade II soft tissue injury.
[0253] Image acquisition:
[0254] set up ;
[0255] Weight - Voltage: ;
[0256] The layer thickness / current of each partition is the same as described above; The pre-scan indicated a partial soft tissue hematoma; therefore, automatic exposure compensation was set.
[0257] 3D Reconstruction:
[0258] Denoising (diffusion 10 times); HU threshold adaptive refinement (±15HU); joint surface refinement using graph cut + active contour; Marching Cubes (step size 0.5mm) + Laplacian smoothing. .
[0259] The original three-dimensional fracture model (approximately 260k faces) was obtained; 47 feature points of the fracture line and a feature vector field were extracted.
[0260] Measurement and Virtual Reset:
[0261] Initial test: Key misalignment =12.1mm; =24.6°.
[0262] Reduction strategy: First correct the rotational misalignment of the lateral and posterior malleoli, then address the medial malleolus displacement;
[0263] Initial assessment of repositioning: ≈0.65mm; joint surface mating error ≤1.5mm.
[0264] Surgical planning:
[0265] Approach scoring (critical selection) Approaching the 3-point threshold, it is recommended to choose between percutaneous minimally invasive surgery or small incision-assisted surgery, with dynamic decision-making based on the soft tissue condition during the operation; the vascular nerve bundle projection shows that the medial side has a lower risk, while the posterolateral side requires a 3mm safety boundary.
[0266] Internal fixation optimization (median parameters):
[0267] Screw: 4.0mm in diameter; 24mm→ ∈[36,48]mm;
[0268] angle ∈[70°,90°];
[0269] The objectives and constraints are the same as before.
[0270] Convergence (Generation 168): =80°, =1.5, =36mm; Predicted =650MPa; =0.88; after fine-tuning mm.
[0271] Intraoperative navigation:
[0272] Imaging:
[0273] Coarse registration error: 1.5mm;
[0274] Weighted ICP: RMSE 0.56→0.38→0.23 (terminated); outlier removal rate ≈4%; ≤0.8ms; Hand-eye calibration residual 0.18mm.
[0275] Device fusion:
[0276] Trajectory: Average 1.6mm, =1.1mm; no continuous "alert" status; path prediction is stable.
[0277] Dosage and quality control:
[0278] Cumulative dose ≈ 58% of the pathway threshold;
[0279] Snapshot verification: Alignment difference 0.35-0.55mm;
[0280] Intraoperative KPIs: ; =650MPa; =0.88; RMSE=0.23; =1.1mm.
[0281] Postoperative verification and data closed loop:
[0282] Immediate CT: Excellent alignment; fixed-position anastomosis planning;
[0283] Dataset D archiving and H computation; consortium linking; federated learning awaits multi-center aggregation .
[0284] Comparison Table 1 for Examples 1-3:
[0285] Indicators Example 1 Example 2 Example 3 Approach score S 1.2 (minimally invasive) 4.8 (open) 2.55 (minimally invasive / small incision) Reduction midstate dr 0.24 0.28 0.22 Internal fixation stress s 620 710 650 Contact area ratio Acon / Aref 0.84 0.86 0.88 Registration RMSE (mm) 0.26 0.29 0.23 Deltamin 0.9 0.8 1.1 High resolution acquisition parameters 2 3 2 Cumulative dose (relative threshold) 65% 78% 58%
[0286] It should be noted that the embodiments of the present invention have better implementability and are not intended to limit the present invention in any way. Any person skilled in the art may use the above-disclosed technical content to change or modify it into equivalent effective embodiments. However, any modifications or equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of the technical solution of the present invention.
Claims
1. A digital image-guided method for assisting in the reduction of ankle fractures, characterized in that, Including the following steps: Step S100: Acquire CT image data of the patient's ankle fracture site, wherein the acquisition process includes setting CT scan parameters; Step S200: Import the CT image data into medical image processing software for three-dimensional reconstruction to obtain the original three-dimensional fracture model; Step S300: Measure and virtually reduce the fracture fragments on the original three-dimensional fracture model to obtain a reduced three-dimensional fracture model; Step S400: Based on the three-dimensional fracture model after reduction, plan the surgical approach and the placement and angle of the internal fixation device; Step S500: During the operation, real-time image data is acquired and registered and fused with the three-dimensional fracture model after reduction to generate a visual guide image, providing real-time navigation for the surgical operation.
2. The digital image-guided method for assisting in the reduction of ankle fractures according to claim 1, characterized in that, The CT scan parameters are adaptively adjusted to satisfy: in, For layer thickness, For scanning speed, U is the pitch, U is the tube voltage, I is the tube current, and W is the weight. To round down, min( () is the minimum value operator.
3. The digital image-guided method for assisting in the reduction of ankle fractures according to claim 1, characterized in that, When measuring the original three-dimensional fracture model in step S300, two corresponding feature points and Distance according to Calculate; two corresponding vectors , The included angle calculate; in, The distance between the two points is in mm. The included angle is (°).
4. The digital image-guided method for assisting ankle fracture reduction according to claim 1, characterized in that, In step S400, the surgical approach is determined based on a weighted score S: Among them, A is the age score, B is the BMI score, C is the degree of comminutedness score, D is the soft tissue injury grade score, and E is the basic disease score, all of which are taken as 1-5 points; when S≤3, the percutaneous minimally invasive approach is selected, when 3<S≤4.5, the small incision auxiliary approach is selected, and when S>4.5, the standard open approach is selected.
5. The digital image-guided method for assisting in the reduction of ankle fractures according to claim 1, characterized in that, In step S400, the internal fixation scheme is achieved by minimizing the objective function. get; in, The maximum equivalent force (MPa) of the internal fixation device. The residual displacement of the fracture fragments after reduction (mm). The bone-nail contact area. For reference contact area; and satisfying And the angle between the screw axis and the fracture line Screw length ,in This refers to the thickness of the fracture fragment.
6. The digital image-guided method for assisting ankle fracture reduction according to claim 1, characterized in that, Intraoperative imaging in step S500 satisfies: , Spatial resolution ≤ 0.15mm cm; among which, For low-dose single-frame dose, For the collection interval, The diameter of the imaging field of view.
7. The digital image-guided method for assisting ankle fracture reduction according to claim 1, characterized in that, In step S500, the registration and fusion employs a weighted iterative nearest-point algorithm to solve for the rotation matrix R and the translation vector. : in, These are the feature points of the three-dimensional fracture model after reduction. For real-time image feature points, =0.8 (cortical bone) or 0.2 (cancellous bone), N is the number of matching points; As a termination condition, and when the residual at a certain point... Those that are removed should be discarded; the error in the coarse registration stage should be ≤2mm.
8. The digital image-guided method for assisting ankle fracture reduction according to claim 1, characterized in that, In step S500, the instrument employs optical-electromagnetic dual-modal tracking and time-synchronized fusion with the image to meet the requirements. The distance from the tip of the instrument to the fracture surface is ;in, Let S be the coordinates of the instrument tip, and S be the set of fracture surface points on the three-dimensional fracture model after reduction.
9. The digital image-guided method for assisting ankle fracture reduction according to claim 1, characterized in that, A dataset D was created postoperatively, containing DICOM images, a post-reduction 3D fracture model, and surgical parameter logs. Its integrity was verified as follows: Wherein, H is the hash value; the data has disaster recovery redundancy and a retention period of ≥15 years.
10. The digital image-guided method for assisting in the reduction of ankle fractures according to claim 9, characterized in that, The database employs a consortium blockchain architecture, requiring consensus among at least three nodes for data modifications; cross-institutional analysis utilizes federated learning, with the parameters of the k-th node set as follows: The weight is Global aggregation parameters ; Where K is the number of participating nodes; the validity criterion for the discovered association rules is statistical significance P≤0.01.
