Method and system for digital generation of implant surgery guide

Abstract

The application provides a kind of implant surgery guide plate digital generation method and system, the method comprises: the spatial registration of multi-modal data of patient oral cavity is carried out, obtains three-dimensional virtual surgery environment;Key anatomical structure in three-dimensional virtual surgery environment is segmented, and according to the segmentation result, the clear degree of image different region quantifies segmentation confidence;The adaptive three-dimensional morphological inflation of segmentation confidence is realized through the mapping rule function of pre-set, obtains safe inflation radius, generates dynamic safety envelope according to each safe inflation radius;With the three-dimensional position and angle of implant as optimization variable, the optimization objective function of maximum stability and best prosthesis angle is established, with the hard constraint condition of not intersecting with dynamic safety envelope, the optimal implant three-dimensional pose is calculated;According to the optimal implant three-dimensional pose, generate surgery guide plate.The application can realize the optimal balance on two core targets of surgery safety and bone mass utilization rate.

Description

Technical Field

[0001] This application belongs to the field of digital medical auxiliary technology, and in particular relates to a method and system for digitally generating surgical guides for implantation. Background Technology

[0002] One of the key prerequisites for successful oral implant surgery is the precise planning of the implant's three-dimensional positioning to avoid critical anatomical structures such as the mandibular canal (inferior alveolar nerve canal) and the maxillary sinus floor, preventing serious complications such as permanent nerve damage or maxillary sinus perforation. Currently, digital implant planning in clinical practice mainly relies on manual operation by doctors in specialized software, which has two major technical limitations:

[0003] Existing technology 1 (fully manual mode): Doctors need to manually outline the contours of structures such as the neural canal layer by layer on the patient's cone-beam computed tomography (CBCT) images. This process is time-consuming and highly dependent on the doctor's experience. The outline results have significant subjective differences, making it difficult to guarantee the consistency and objective accuracy of the planning.

[0004] Existing technology two (automatic segmentation combined with fixed safety distance mode): Some advanced systems use convolutional neural networks (CNNs) to automatically segment key anatomical structures. However, such methods have a fundamental flaw: they only output a deterministic segmentation boundary and cannot assess the reliability of the segmentation result in different regions of the image (i.e., there is actual uncertainty). Based on this, planning systems usually simply expand the segmentation boundary outward by a fixed distance (e.g., 2.0 mm) as a safety zone.

[0005] This "one-size-fits-all" strategy leads to the following: In areas with clear CBCT images and well-defined structural boundaries, an excessively large fixed safety distance may unnecessarily encroach on valuable available bone, potentially resulting in a misjudgment of "insufficient bone for implantation"; In areas where images are blurred due to metal artifacts, low contrast, or delicate anatomical structures, the same fixed distance may not be sufficient to compensate for segmentation errors, and there remains a high risk of nerve damage. Summary of the Invention

[0006] This application provides a method and system for digitally generating surgical guides for implantation, in order to solve at least one of the above-mentioned technical problems in the prior art.

[0007] The first aspect of this application provides a method for digitally generating a surgical guide for implantation, comprising:

[0008] Multimodal data of the patient's oral cavity were collected, and spatial registration of the multimodal data was performed to obtain a three-dimensional virtual surgical environment;

[0009] Based on a pre-built 3D segmentation network, key anatomical structures in a three-dimensional virtual surgical environment are segmented, and the segmentation confidence is quantified according to the clarity of the segmentation results in different regions of the image.

[0010] The segmentation confidence is used to achieve adaptive 3D morphological dilation through a preset mapping rule function to obtain the safe dilation radius, and a dynamic safe envelope is generated based on each safe dilation radius;

[0011] Using the three-dimensional position and angle of the implant as optimization variables, an optimization objective function for maximum stability and optimal prosthesis angle is established. With non-intersection with the dynamic safety envelope as a hard constraint, the optimal three-dimensional pose of the implant is calculated through a preset algorithm.

[0012] Based on the optimal three-dimensional pose of the implant, a surgical guide is generated.

[0013] In some preferred embodiments, the acquisition of multimodal data from the patient's oral cavity and the spatial registration of the multimodal data to obtain a three-dimensional virtual surgical environment include:

[0014] Acquire optical scanning models of the patient's oral cavity and dental arch;

[0015] The tapered wire bundle is preprocessed and its grayscale is normalized to obtain the processed tapered wire bundle;

[0016] Based on common anatomical features, feature matching and iterative nearest point algorithms are used to align the processed conical wire beam coordinate system with the dental optical scanning model coordinate system to obtain a three-dimensional virtual surgical environment.

[0017] In some preferred embodiments, the segmentation of key anatomical structures in a three-dimensional virtual surgical environment based on a pre-built 3D segmentation network, and the quantification of segmentation confidence based on the clarity of the segmentation results in different regions of the image, includes:

[0018] An improved Bayesian 3D V-Net convolutional neural network is constructed, which includes a random deactivation layer, a segmentation prediction channel, and a confidence quantization channel, wherein:

[0019] A random deactivation layer is used to perform T forward propagation cycles in a 3D virtual surgical environment to predict the arithmetic mean of the Softmax probability map. If the average probability of a voxel is greater than a preset threshold, it is marked as a target structure, a binary mask is generated, and a voxel-level semantic segmentation mask is output through the segmentation prediction channel. At the same time, the standard deviation of the probability distribution of the voxel belonging to the target category in the T prediction cycles is calculated, the confidence level is calculated based on the standard deviation, and the confidence level is output through the confidence quantification channel.

[0020] In some preferred embodiments, the segmentation confidence is adaptively expanded in three dimensions using a preset mapping rule function to obtain a safe expansion radius, and a dynamic safe envelope is generated based on each safe expansion radius. The process of setting the mapping rule function is as follows:

[0021] Set basic anatomical safety thresholds based on the clinical safety standards of key anatomical structures;

[0022] Risk penalty coefficients are set according to the severity of the consequences of damage to key anatomical structures;

[0023] The mapping rule function is set based on the basic anatomical safety threshold, risk penalty coefficient, and confidence level.

[0024] In some preferred embodiments, the mapping rule function is:

[0025]

[0026] In the above formula, Indicates the safe expansion radius; Indicates the basic anatomical safety threshold; α represents the risk penalty coefficient; P represents the confidence level;

[0027] In some preferred embodiments, the step of adaptively dilating the segmentation confidence using a preset mapping rule function to obtain a safe dilation radius, and generating a dynamic safe envelope based on each safe dilation radius, includes:

[0028] Calculate the confidence level for each point on the segmentation mask surface, and calculate the secure expansion radius based on the confidence level of that point and the mapping rule function. Use the secure expansion radius as the local expansion kernel radius.

[0029] A dynamic safety envelope is generated based on the radius of each local expansion core. The distances between each point on the surface of the dynamic safety envelope and each point on the surface of the key anatomical structure are not equal.

[0030] In some preferred embodiments, the optimization objective function for maximum stability and optimal prosthesis angle is established using the three-dimensional position and angle of the implant as optimization variables, with the constraint of not intersecting with the dynamic safety envelope as a hard constraint. The optimal three-dimensional pose of the implant is calculated using a preset algorithm, wherein:

[0031] Using the three-dimensional position and angle of the implant as optimization variables, the objective function F for maximizing stability and optimizing the prosthesis angle is expressed by the following formula:

[0032]

[0033] In the above formula, Indicates the initial stability factor; This indicates the repair guidance factor; w1 and w2 represent different preset parameters.

[0034] In some preferred embodiments, generating the surgical guide based on the optimal three-dimensional pose of the implant includes:

[0035] Based on the optimal three-dimensional pose of the implant, a guide plate matrix that matches the optical scanning model of the dental arch is generated;

[0036] Guide holes are generated on the guide plate substrate along the planned axis of the implant using Boolean subtraction operations;

[0037] Add auxiliary structures along the direction of force applied to the guide plate;

[0038] The output is a 3D model file of an implantation surgical guide that can be directly used for 3D printing.

[0039] In some preferred embodiments, generating a guide plate matrix that matches the optical scanning model of the dental arch based on the optimal three-dimensional pose of the implant further includes:

[0040] Shell offsetting is performed on the guide plate substrate.

[0041] As can be seen from the above, compared with the prior art, this application has at least the following beneficial effects:

[0042] This application quantifies segmentation confidence by assessing the clarity of segmentation results in different regions of an image. Based on this confidence, a dynamic safety envelope is constructed, and the optimal 3D pose of the implant is calculated with the constraint that the implant does not intersect with the dynamic safety envelope. In other words, the uncertainty of the segmentation results is directly incorporated into the safety planning decision. Therefore, it can intelligently increase the safety margin in risky areas of blurred images. Specifically, the dynamic safety envelope adaptively compensates for segmentation uncertainty areas, thereby reducing the risk of nerve damage caused by segmentation errors. Compared to fixed safety distance methods, this approach offers greater safety advantages, fundamentally reducing the planning safety blind spot caused by image quality limitations and achieving optimized safety performance.

[0043] Furthermore, in areas with clear imaging, a more compact safety distance can be used, avoiding unnecessary waste of bone mass. This allows for the preservation of more usable bone mass for patients under the same safety standards, or provides feasible implantation options for complex cases that would otherwise be considered contraindicated due to insufficient bone mass, thereby optimizing bone mass utilization.

[0044] Meanwhile, by using the three-dimensional position and angle of the implant as optimization variables, an optimization objective function for maximum stability and optimal prosthesis angle is established. With the constraint of not intersecting with the dynamic safety envelope as a hard constraint, the optimal three-dimensional pose of the implant is calculated through a preset algorithm. Based on the optimal three-dimensional pose of the implant, a surgical guide is generated. Obviously, it can realize the full automation of the process from image analysis to guide design, significantly improve efficiency, completely eliminate the subjective differences caused by different doctors' operations, and ensure the high standard and repeatability of the planning scheme.

[0045] In summary, the embodiments of this application achieve an optimal balance between the two core objectives of "surgical safety" and "bone utilization rate" by intelligently sensing the uncertainty of image quality and dynamically adjusting the safety strategy accordingly.

[0046] A second aspect of this application provides a digital generation system for implantation surgical guides, comprising:

[0047] The acquisition module is used to acquire multimodal data of the patient's oral cavity and perform spatial registration on the multimodal data to obtain a three-dimensional virtual surgical environment;

[0048] The quantization module is used to segment key anatomical structures in a three-dimensional virtual surgical environment based on a pre-built 3D segmentation network, and to quantify the segmentation confidence based on the clarity of the segmentation results in different regions of the image.

[0049] The generation module is used to adaptively dilate the segmentation confidence through a preset mapping rule function to obtain the safe dilation radius, and generate a dynamic safe envelope based on each safe dilation radius;

[0050] The planning module is used to establish an optimization objective function with the three-dimensional position and angle of the implant as optimization variables, to achieve maximum stability and optimal prosthesis angle, with the constraint of not intersecting with the dynamic safety envelope as a hard condition, and to calculate the optimal three-dimensional pose of the implant through a preset algorithm.

[0051] The output module is used to generate a surgical guide based on the optimal three-dimensional pose of the implant.

[0052] A third aspect of this application provides a terminal including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described in the first aspect.

[0053] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method described in the first aspect.

[0054] The fifth aspect of this application provides a computer program product that, when run on a terminal, causes the terminal to perform the steps of the method described in the first aspect.

[0055] It should be understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 This is a schematic diagram of the main process of the digital generation method for implantation surgical guides provided in the embodiments of this application; Figure 2 This is a schematic diagram of the dynamic security envelope principle provided in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of the digital generation system for implantation surgical guides provided in the embodiments of this application; Figure 4 This is a structural diagram of the terminal provided in the embodiments of this application. Detailed Implementation

[0058] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0059] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0060] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0061] It should be understood that the term "and / or" as used in this application specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0062] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrases "if determined" or "if [described condition or event] is detected" may be interpreted, depending on the context, as "once determined," "in response to determination," "once [described condition or event] is detected," or "in response to detection of [described condition or event]."

[0063] It should be understood that the sequence number of each step in this embodiment does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of this application embodiment.

[0064] To illustrate the technical solution described in this application, specific embodiments are provided below.

[0065] See Figure 1-2 , Figure 1 This is a schematic diagram of the main process of the digital generation method of implantation surgical guide provided in the embodiments of this application. The diagram clearly shows the complete five steps from "multimodal data input" to "digital generation of guide".

[0066] Figure 2 This is a schematic diagram of the dynamic safety envelope principle provided in the embodiments of this application. In this diagram, the central circle represents a key anatomical structure (such as the inferior alveolar nerve canal), which is the core target for safe avoidance. The gray dashed line represents the fixed safety distance R in the prior art. base This embodies a "one-size-fits-all" avoidance strategy; the outer solid line represents the "dynamic safety envelope" provided in the embodiments of this application, which is driven by confidence level P to achieve non-uniform expansion; the left region (clear image area): confidence level P≈0.9 (high confidence), the dynamic envelope is close to the basic safety distance, maximizing the preservation of usable bone volume; the right region (blurred image area): confidence level P≈0.4 (low confidence), the dynamic envelope automatically expands to form a reinforced safety barrier.

[0067] like Figure 1-2 As shown in the embodiments of this application, the method for digitally generating implantation surgical guides may include the following:

[0068] Step 1: Collect multimodal data of the patient's oral cavity and perform spatial registration on the multimodal data to obtain a three-dimensional virtual surgical environment;

[0069] Step 2: Based on the pre-built 3D segmentation network, the key anatomical structures in the three-dimensional virtual surgical environment are segmented, and the segmentation confidence is quantified according to the clarity of the segmentation results in different regions of the image.

[0070] Step 3: The segmentation confidence is adaptively expanded in three dimensions using a preset mapping rule function to obtain the safe expansion radius, and a dynamic safe envelope is generated based on each safe expansion radius;

[0071] Step 4: Using the three-dimensional position and angle of the implant as optimization variables, establish the optimization objective function for maximum stability and optimal prosthesis angle, with the condition of not intersecting with the dynamic safety envelope as a hard constraint, and calculate the optimal three-dimensional pose of the implant through a preset algorithm.

[0072] Step 5: Generate a surgical guide based on the optimal three-dimensional pose of the implant.

[0073] Clearly, this embodiment quantifies the segmentation confidence by assessing the clarity of the segmentation results in different regions of the image. Based on this confidence, a dynamic safety envelope is constructed, and the optimal three-dimensional pose of the implant is calculated with the constraint that the implant does not intersect with the dynamic safety envelope. In other words, the uncertainty of the segmentation results is directly incorporated into the safety planning decision. Therefore, it can intelligently increase the safety margin in risky areas of blurred images. Specifically, the dynamic safety envelope adaptively compensates for uncertain segmentation areas, thereby reducing the risk of nerve damage caused by segmentation errors. Compared to the fixed safety distance method, this approach offers greater safety advantages, fundamentally reducing the planning safety blind spot caused by image quality limitations and achieving optimized safety performance.

[0074] Furthermore, in areas with clear imaging, a more compact safety distance can be used, avoiding unnecessary waste of bone mass. This allows for the preservation of more usable bone mass for patients under the same safety standards, or provides feasible implantation options for complex cases that would otherwise be considered contraindicated due to insufficient bone mass, thereby optimizing bone mass utilization.

[0075] Meanwhile, by using the three-dimensional position and angle of the implant as optimization variables, an optimization objective function for maximum stability and optimal prosthesis angle is established. With the constraint of not intersecting with the dynamic safety envelope as a hard constraint, the optimal three-dimensional pose of the implant is calculated through a preset algorithm. Based on the optimal three-dimensional pose of the implant, a surgical guide is generated. Obviously, it can realize the full automation of the process from image analysis to guide design, significantly improve efficiency, completely eliminate the subjective differences caused by different doctors' operations, and ensure the high standard and repeatability of the planning scheme.

[0076] In summary, the embodiments of this application achieve an optimal balance between the two core objectives of "surgical safety" and "bone utilization rate" by intelligently sensing the uncertainty of image quality and dynamically adjusting the safety strategy accordingly.

[0077] In some embodiments, step 1, acquiring multimodal data of the patient's oral cavity and spatially registering the multimodal data to obtain a three-dimensional virtual surgical environment, may include the following:

[0078] Step 11: Acquire the patient's oral cavity cone beam and dental arch optical scan model, wherein the cone beam is in DICOM format and the dental arch optical scan model is in STL / PLY format;

[0079] Step 12: Preprocess the tapered wire harness to reduce noise and metal artifacts, and perform grayscale normalization to obtain the processed tapered wire harness.

[0080] Step 13: Based on common anatomical features, the feature matching and iterative nearest point (ICP) algorithm is used to precisely align the processed conical beam coordinate system with the dental optical scanning model coordinate system to obtain a unified three-dimensional virtual surgical environment. Common anatomical features can include tooth cusps, alveolar ridge apex, etc. Common anatomical feature points can be automatically detected by the SIFT algorithm.

[0081] In this embodiment, the iteration termination condition of the ICP algorithm is to stop when the iteration error is less than 0.01 mm.

[0082] Specifically, step 12 involves preprocessing the tapered wire harness to reduce noise and metal artifacts, and performing grayscale normalization to obtain the processed tapered wire harness, which may include the following:

[0083] Noise is reduced by a combination of median filtering (window size 3×3×3) and nonlocal mean filtering (NLM), and metal artifact suppression (MAR) algorithm is used to eliminate metal interference from implants / restorations.

[0084] The Z-score normalization method is used for grayscale normalization, and its formula can be expressed as follows:

[0085] X norm =(X-μ) / σ

[0086] In the above formula, X norm σ represents the normalized grayscale value of the tapered wire harness; X represents the original grayscale value of the tapered wire harness; μ represents the mean grayscale value of the tapered wire harness; σ represents the standard deviation of the grayscale value of the tapered wire harness.

[0087] In some embodiments, step 2, which involves segmenting key anatomical structures in a three-dimensional virtual surgical environment based on a pre-built 3D segmentation network and quantifying the segmentation confidence based on the clarity of the segmentation results in different regions of the image, may include the following:

[0088] An improved Bayesian 3D V-Net convolutional neural network is constructed, which includes a random deactivation layer, a segmentation prediction channel, and a confidence quantization channel, wherein:

[0089] A random deactivation layer is enabled for T forward propagation cycles in a 3D virtual surgical environment to predict the arithmetic mean of the Softmax probability map. If the average probability of a voxel is greater than a preset threshold, it is marked as a target structure, a binary mask is generated, and a voxel-level semantic segmentation mask is output through the segmentation prediction channel (i.e., the segmentation result is composed of several segmentation masks). At the same time, the standard deviation of the probability distribution of the voxel belonging to the target category in T prediction cycles is calculated, the confidence level is calculated based on the standard deviation, and the confidence level is output through the confidence quantification channel.

[0090] Obviously, the embodiments of this application employ a random deactivated layer in an improved Bayesian 3D V-Net convolutional neural network to perform T forward propagation cycles, predicting the arithmetic mean of the Softmax probability map. If the average probability of a voxel is greater than a preset threshold, it is marked as a target structure, generating a binary mask. This allows for the automatic extraction of key anatomical structures such as the inferior alveolar nerve canal and maxillary sinus from a 3D virtual surgical environment. Furthermore, by calculating the standard deviation of the probability distribution of the voxel belonging to the target category in T prediction cycles, the credibility of the segmentation results in different regions is quantified. This achieves the integration of anatomical structure segmentation and uncertainty quantification, overcoming the shortcomings of existing technologies in assessing segmentation credibility.

[0091] More specifically, step 2, based on a pre-built 3D segmentation network, segments key anatomical structures in the 3D virtual surgical environment and quantifies the segmentation confidence based on the clarity of the segmentation results in different regions of the image. This may include the following:

[0092] Construct an improved Bayesian 3DV-Net convolutional neural network. The structure of this convolutional neural network is improved and optimized as follows:

[0093] It includes a 5-layer encoder and a 5-layer decoder. The encoder uses a 3×3×3 convolution kernel (stride 2) for downsampling, and the decoder uses a transposed convolution kernel (3×3×3, stride 2) for upsampling.

[0094] Dropout layers are inserted in layers 2 and 4 of the encoder and layers 2 and 4 of the decoder, with the dropout probability set to 0.2. During the model inference stage, the dropout (random deactivation) layers in the network are enabled and T forward propagations are performed (i.e., the Monte Carlo Dropout method). In this embodiment, based on the commonly used confidence calculation accuracy requirements in the field, T=50 is set. However, in practical applications, it is not limited to this and can be set as needed.

[0095] Calculate the arithmetic mean of the Softmax probability maps predicted T times. If the average probability of a voxel is greater than a preset threshold (e.g., greater than 0.5), it is marked as the target structure, a binary mask is generated, and finally a voxel-level semantic segmentation mask is output to identify structures such as the inferior alveolar nerve canal and the maxillary sinus. The segmentation result is output through the segmentation prediction channel.

[0096] Calculate the standard deviation σ1 of the probability distribution of the voxel belonging to the target category in T predictions, calculate the confidence level based on the standard deviation, and output the confidence level through the confidence quantification channel;

[0097] The confidence level can be calculated using the following formula:

[0098] P = 1 - Normalize(σ1)

[0099] In the above formula, P represents the confidence level; Normalize represents the normalization function, specifically σ1 / σ max , where σ max σ1 represents the maximum value of the standard deviation of all voxels in this prediction, thus ensuring that P∈[0,1]; σ1 represents the standard deviation of the probability distribution belonging to a certain target category.

[0100] Clearly, this embodiment quantifies the model's predictions for high-confidence regions with clear structures and low-confidence regions with blurred images due to artifact interference. The clearer the image, the more consistent the prediction results, the smaller the standard deviation σ1, and the higher the confidence level P.

[0101] In some embodiments, step 3, which involves adaptively dilating the segmentation confidence using a preset mapping rule function to obtain a safe dilation radius, and generating a dynamic safe envelope based on each safe dilation radius, may include the following:

[0102] Step 31: Set a basic anatomical safety threshold based on the clinical safety standards for key anatomical structures, and set a risk penalty coefficient based on the severity of the consequences of damage to key anatomical structures; set a mapping rule function based on the basic anatomical safety threshold, risk penalty coefficient, and confidence level. This mapping rule function can be expressed by the following formula:

[0103]

[0104] In the above formula, Indicates the safe expansion radius;

[0105] This represents the basic anatomical safety threshold, and its value is determined based on the clinical safety standards of anatomical structures, such as the inferior alveolar nerve canal. maxillary sinus floor ;

[0106] α represents the risk penalty coefficient, which is set based on the severity of the consequences of anatomical damage. For example, Level 1 risk (inferior alveolar nerve canal, damage leading to permanent numbness) α=1.5, Level 2 risk (maxillary sinus floor, damage leading to perforation) α=0.8, and Level 3 risk (adjacent tooth root, damage leading to tooth sensitivity) α=0.5.

[0107] P represents the confidence level. The threshold of P can be set to 0.5. When P < 0.5, it is judged as a low confidence area and the maximum safe distance limit is automatically triggered (in actual applications, the maximum safe distance does not exceed 4.0mm).

[0108] Clearly, the above scheme can establish a spatially adaptive safe distance function, which maps the confidence value P of each voxel point to a mapping rule function of the required safe expansion radius R at that point.

[0109] Step 32: Calculate the confidence level for each point on the segmentation mask surface, and calculate the secure expansion radius based on the confidence level of the point and the mapping rule function. Use the secure expansion radius as the local expansion kernel radius.

[0110] Step 33: Generate a dynamic safety envelope based on the radius of each locally expanded nucleus. The distances between points on the surface of the dynamic safety envelope and points on the surface of key anatomical structures are not equal, such as... Figure 2 As shown, in the high-resolution region with confidence level P=0.9, the radius of dilation is only slightly larger than R. base Preserving more bone mass; in the artifact interference region with confidence level P=0.4, the expansion radius automatically increases to R. base +0.6α, forming an expanded "defensive" security barrier.

[0111] Clearly, the above scheme does not simply expand the segmented key anatomical structures by a fixed distance (such as 2mm), but rather performs adaptive three-dimensional morphological operations based on confidence level to generate a larger safety distance in low-confidence (blurred boundary) regions. This constructs a non-uniform, anatomically consistent, and risk-aware "dynamic safety envelope," achieving a key transformation from "mathematical uncertainty" to "physical avoidance distance."

[0112] In some embodiments, step 4 involves establishing an optimization objective function for maximum stability and optimal prosthesis angle, using the three-dimensional position and angle of the implant as optimization variables, with the constraint of not intersecting with the dynamic safety envelope as a hard constraint. The optimal three-dimensional pose of the implant is then calculated using a preset algorithm, wherein:

[0113] Using the three-dimensional position and angle of the implant as optimization variables, the objective function F for maximizing stability and optimizing the prosthesis angle can be expressed by the following formula:

[0114]

[0115] In the above formula, Indicates the initial stability factor; This indicates the repair guidance factor; w1 and w2 represent different preset parameters, where w1=0.6 and w2=0.4 can be set.

[0116] The initial stability factor can be calculated in the following way:

[0117] Calculate the Boolean intersection volume or bone-implant contact area between the implant 3D mesh and the high-density bone cell set (CT value > 800 Henle units), and normalize it;

[0118] The repair guidance factor can be calculated in the following way:

[0119] Calculate the cosine of the angle between the long axis of the implant and the "ideal restoration axis" (obtained by linear fitting of the occlusal center point of the opposing tooth and the long axis of the adjacent tooth using the least squares method).

[0120] The hard constraint of not intersecting with the dynamic safety envelope can be set as follows:

[0121] The Boolean intersection of the implant's 3D mesh model and the dynamic safety envelope must be empty.

[0122] The preset algorithm can be:

[0123] The non-dominated sorting genetic algorithm with elitist strategy (NSGA-II) is adopted, with the following parameters: population size 50, maximum number of iterations 100, crossover probability 0.8, and mutation probability 0.05.

[0124] Clearly, this embodiment models the implantation planning as a multi-objective constrained optimization problem, with "not intersecting with the dynamic safety envelope" as a hard constraint, and the initial stability of the implant, the coronal repair space and the feasibility of the surgical approach as optimization objectives. It can automatically solve for the optimal or Pareto optimal implant position, depth and angle.

[0125] In summary, this application addresses the shortcomings of existing technologies in assessing segmentation reliability by firstly integrating anatomical structure segmentation and uncertainty quantification using Bayesian 3D V-Net; secondly, it constructs a confidence-driven dynamic safety envelope to achieve spatial adaptive adjustment of the safety distance; and thirdly, it combines a multi-objective optimization algorithm to balance implant stability and repair effect. These three aspects work together to form a complete technical solution, achieving optimal implant path planning under multi-objective constraints, resulting in a safe and efficient implantation solution.

[0126] In some embodiments, step 5, generating a surgical guide based on the optimal three-dimensional pose of the implant, may include the following:

[0127] Step 51: Based on the optimal three-dimensional pose of the implant, generate a guide plate matrix that matches the optical scanning model of the dental arch, and perform shell offset on the guide plate matrix. For example, automatically perform undercut filling and small gap compensation (such as 0.05mm) on the inner surface of the guide plate to ensure smooth placement.

[0128] Step 52: Along the planned axis of the implant, guide holes are precisely generated on the guide plate substrate using Boolean subtraction operations;

[0129] Step 53: Add auxiliary structures along the force direction of the guide plate to ensure the stability and strength of the guide plate. For example, add 2-3 reinforcing ribs (2mm wide and 1.5mm thick) along the force direction of the guide plate, and set 2 retention windows with a diameter of 3mm (located in the non-engaging area of ​​the guide plate).

[0130] Step 54: Output a 3D model file of the implantation surgical guide that can be directly used for 3D printing, such as an STL format surgical guide 3D model file.

[0131] Obviously, based on the final determined implantation planning parameters, the embodiments of this application automatically complete the three-dimensional geometric design of the implantation guide plate through the algorithm, including the guide hole, fixation device and tissue surface morphology, and directly generate standardized files (such as STL models) that can be used for 3D printing, realizing full-process automation from image analysis to guide plate design.

[0132] Furthermore, based on existing mainstream GPU hardware (graphics processing units with CUDA acceleration capabilities), the fully automated process can control the planning time for a single tooth implantation within 5 minutes, significantly improving efficiency compared to manual planning (around 30 minutes). At the same time, it completely eliminates subjective differences caused by different doctors' operations, ensuring high standards and repeatability of the planning scheme.

[0133] Based on the same inventive concept, embodiments of this application also provide a digital generation system for implantation surgical guides, such as... Figure 3 As shown, it includes:

[0134] Acquisition module 1 is used to acquire multimodal data of the patient's oral cavity and perform spatial registration on the multimodal data to obtain a three-dimensional virtual surgical environment;

[0135] Quantization module 2 is used to segment key anatomical structures in a three-dimensional virtual surgical environment based on a pre-built 3D segmentation network, and to quantify the segmentation confidence based on the clarity of the segmentation results in different regions of the image.

[0136] The generation module 3 is used to achieve adaptive three-dimensional morphological dilation of the segmentation confidence through a preset mapping rule function to obtain the safe dilation radius, and generate a dynamic safe envelope based on each safe dilation radius;

[0137] Planning module 4 is used to establish an optimization objective function for maximum stability and optimal prosthesis angle with the three-dimensional position and angle of the implant as optimization variables, and to calculate the optimal three-dimensional pose of the implant through a preset algorithm with the hard constraint of not intersecting with the dynamic safety envelope.

[0138] Output module 5 is used to generate a surgical guide based on the optimal three-dimensional pose of the implant.

[0139] Clearly, this embodiment quantifies the segmentation confidence by assessing the clarity of the segmentation results in different regions of the image. Based on this confidence, a dynamic safety envelope is constructed, and the optimal three-dimensional pose of the implant is calculated with the constraint that the implant does not intersect with the dynamic safety envelope. In other words, the uncertainty of the segmentation results is directly incorporated into the safety planning decision. Therefore, it can intelligently increase the safety margin in risky areas of blurred images. Specifically, the dynamic safety envelope adaptively compensates for uncertain segmentation areas, thereby reducing the risk of nerve damage caused by segmentation errors. Compared to the fixed safety distance method, this approach offers greater safety advantages, fundamentally reducing the planning safety blind spot caused by image quality limitations and achieving optimized safety performance.

[0140] Furthermore, in areas with clear imaging, a more compact safety distance can be used, avoiding unnecessary waste of bone mass. This allows for the preservation of more usable bone mass for patients under the same safety standards, or provides feasible implantation options for complex cases that would otherwise be considered contraindicated due to insufficient bone mass, thereby optimizing bone mass utilization.

[0141] Meanwhile, by using the three-dimensional position and angle of the implant as optimization variables, an optimization objective function for maximum stability and optimal prosthesis angle is established. With the constraint of not intersecting with the dynamic safety envelope as a hard constraint, the optimal three-dimensional pose of the implant is calculated through a preset algorithm. Based on the optimal three-dimensional pose of the implant, a surgical guide is generated. Obviously, it can realize the full automation of the process from image analysis to guide design, significantly improve efficiency, completely eliminate the subjective differences caused by different doctors' operations, and ensure the high standard and repeatability of the planning scheme.

[0142] In summary, the embodiments of this application achieve an optimal balance between the two core objectives of "surgical safety" and "bone utilization rate" by intelligently sensing the uncertainty of image quality and dynamically adjusting the safety strategy accordingly.

[0143] Figure 4 This is a structural diagram of a terminal provided in an embodiment of this application. As shown in the figure, the terminal 4 of this embodiment includes: at least one processor 40 ( Figure 4(Only one is shown in the diagram), memory 41, and computer program 42 stored in memory 41 and executable on at least one processor 40, which, when executed by processor 40, performs the steps in any of the above-described method embodiments.

[0144] Terminal 4 can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. Terminal 4 may include, but is not limited to, a processor 40 and a memory 41. Those skilled in the art will understand that... Figure 4 This is merely an example of terminal 4 and does not constitute a limitation on terminal 4. It may include more or fewer components than shown, or combine certain components, or different components. For example, the terminal may also include input / output devices, network access devices, buses, etc.

[0145] The processor 40 can be a central processing unit, or it can be other general-purpose processors, digital signal processors, application-specific integrated circuits, field-programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0146] The memory 41 can be an internal storage unit of the terminal 4, such as a hard disk or RAM of the terminal 4. The memory 41 can also be an external storage device of the terminal 4, such as a plug-in hard disk, smart memory card, secure digital card, flash memory card, etc., equipped on the terminal 4. Furthermore, the memory 41 can include both internal storage units and external storage devices of the terminal 4. The memory 41 is used to store computer programs and other programs and data required by the terminal. The memory 41 can also be used to temporarily store data that has been output or will be output.

[0147] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0148] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0149] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0150] In the embodiments provided in this application, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0151] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0152] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0153] If an integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory, random access memory, electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0154] The methods described in this application can be implemented in whole or in part by a computer program product. When the computer program product is run on a terminal, the terminal executes the steps in the various method embodiments described above.

[0155] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for digitally generating a surgical guide for implantation, characterized in that, include: Multimodal data of the patient's oral cavity were collected, and spatial registration of the multimodal data was performed to obtain a three-dimensional virtual surgical environment; Based on a pre-built 3D segmentation network, key anatomical structures in a three-dimensional virtual surgical environment are segmented, and the segmentation confidence is quantified according to the clarity of the segmentation results in different regions of the image. The segmentation confidence is used to achieve adaptive 3D morphological dilation through a preset mapping rule function to obtain the safe dilation radius, and a dynamic safe envelope is generated based on each safe dilation radius; Using the three-dimensional position and angle of the implant as optimization variables, an optimization objective function for maximum stability and optimal prosthesis angle is established. With non-intersection with the dynamic safety envelope as a hard constraint, the optimal three-dimensional pose of the implant is calculated through a preset algorithm. The preset algorithm adopts a non-dominated sorting genetic algorithm with an elite strategy. Based on the optimal three-dimensional pose of the implant, a surgical guide is generated.

2. The method for digitally generating surgical guides for implantation according to claim 1, characterized in that, The process of collecting multimodal data from the patient's oral cavity and spatially registering the multimodal data to obtain a three-dimensional virtual surgical environment includes: Acquire optical scanning models of the patient's oral cavity and dental arch; The tapered wire bundle is preprocessed and its grayscale is normalized to obtain the processed tapered wire bundle; Based on common anatomical features, feature matching and iterative nearest point algorithms are used to align the processed conical wire beam coordinate system with the dental optical scanning model coordinate system to obtain a three-dimensional virtual surgical environment.

3. The method for digitally generating surgical guides for implantation according to claim 1, characterized in that, The pre-built 3D segmentation network segments key anatomical structures in a three-dimensional virtual surgical environment and quantifies segmentation confidence based on the clarity of the segmentation results in different regions of the image, including: An improved Bayesian 3D V-Net convolutional neural network is constructed, which includes a random deactivation layer, a segmentation prediction channel, and a confidence quantization channel, wherein: A random deactivation layer is used to perform T forward propagation cycles in a 3D virtual surgical environment to predict the arithmetic mean of the Softmax probability map. If the average probability of a voxel is greater than a preset threshold, it is marked as a target structure, a binary mask is generated, and a voxel-level semantic segmentation mask is output through the segmentation prediction channel. At the same time, the standard deviation of the probability distribution of the voxel belonging to the target category in the T prediction cycles is calculated, the confidence level is calculated based on the standard deviation, and the confidence level is output through the confidence quantification channel.

4. The method for digitally generating surgical guides for implantation according to claim 1, characterized in that, The segmentation confidence is adaptively expanded in three dimensions using a preset mapping rule function to obtain a safe expansion radius. A dynamic safe envelope is then generated based on each safe expansion radius. The process of setting the mapping rule function is as follows: Set basic anatomical safety thresholds based on the clinical safety standards of key anatomical structures; Risk penalty coefficients are set according to the severity of the consequences of damage to key anatomical structures; The mapping rule function is set based on the basic anatomical safety threshold, risk penalty coefficient, and confidence level.

5. The method for digitally generating implantation surgical guides according to claim 4, characterized in that, The mapping rule function is: ； In the above formula, Indicates the safe expansion radius; α represents the basic anatomical safety threshold; P represents the risk penalty coefficient; and P represents the confidence level.

6. The method for digitally generating implantation surgical guides according to claim 3, characterized in that, The process of adaptively dilating the segmentation confidence using a preset mapping rule function to obtain a safe dilation radius, and generating a dynamic safe envelope based on each safe dilation radius, includes: Calculate the confidence level for each point on the segmentation mask surface, and calculate the secure expansion radius based on the confidence level of that point and the mapping rule function. Use the secure expansion radius as the local expansion kernel radius. A dynamic safety envelope is generated based on the radius of each local expansion core. The distances between each point on the surface of the dynamic safety envelope and each point on the surface of the key anatomical structure are not equal.

7. The method for digitally generating surgical guides for implantation according to claim 1, characterized in that, The optimization objective function, which uses the three-dimensional position and angle of the implant as optimization variables, establishes the maximum stability and the optimal prosthesis angle. With the constraint of not intersecting the dynamic safety envelope as a hard condition, the optimal three-dimensional pose of the implant is calculated using a preset algorithm, wherein: Using the three-dimensional position and angle of the implant as optimization variables, the objective function F for maximizing stability and optimizing the prosthesis angle is expressed by the following formula: ； In the above formula, Indicates the initial stability factor; This indicates the repair guidance factor; w1 and w2 represent different preset parameters.

8. The method for digitally generating implantation surgical guides according to claim 1, characterized in that, The process of generating a surgical guide based on the optimal three-dimensional pose of the implant includes: Based on the optimal three-dimensional pose of the implant, a guide plate matrix that matches the optical scanning model of the dental arch is generated; Guide holes are generated on the guide plate substrate along the planned axis of the implant using Boolean subtraction operations; Add auxiliary structures along the direction of force applied to the guide plate; The output is a 3D model file of an implantation surgical guide that can be directly used for 3D printing.

9. The method for digitally generating implantation surgical guides according to claim 8, characterized in that, The process of generating a guide plate matrix that matches the optical scanning model of the dental arch based on the optimal three-dimensional pose of the implant also includes: Shell offsetting is performed on the guide plate substrate.

10. A digital generation system for implantation surgical guides, characterized in that, include: The acquisition module is used to acquire multimodal data of the patient's oral cavity and perform spatial registration on the multimodal data to obtain a three-dimensional virtual surgical environment; The quantization module is used to segment key anatomical structures in a three-dimensional virtual surgical environment based on a pre-built 3D segmentation network, and to quantify the segmentation confidence based on the clarity of the segmentation results in different regions of the image. The generation module is used to adaptively dilate the segmentation confidence through a preset mapping rule function to obtain the safe dilation radius, and generate a dynamic safe envelope based on each safe dilation radius; The planning module is used to establish an optimization objective function with the three-dimensional position and angle of the implant as optimization variables, and to calculate the optimal three-dimensional pose of the implant by means of a hard constraint that it does not intersect with the dynamic safety envelope. The preset algorithm adopts a non-dominated sorting genetic algorithm with an elitist strategy. The output module is used to generate a surgical guide based on the optimal three-dimensional pose of the implant.

Images