A method for generating maxillofacial soft tissue models for the orthodontic process

Abstract

This invention belongs to the field of maxillofacial image analysis technology and provides a method for generating a maxillofacial soft tissue model for the orthodontic process. This invention matches the facial bone information of the current patient with historical model datasets, thereby constructing a current finite data block and a reference finite data block. It then analyzes the distribution of various soft tissues in the reference finite data block corresponding to the current finite data block, obtaining a set of distribution probabilities for various soft tissues in the current finite data block. Based on the distribution probability set, the current finite data block is segmented to obtain several sub-finite data blocks and sub-distribution probability sets. Finally, based on the sub-distribution probability sets, the consistency of soft tissue distribution between the sub-finite data blocks and all their neighboring sub-finite data blocks is analyzed, and the sub-finite data blocks are merged to obtain a more accurate soft tissue model structure for the current patient.

Description

Technical Field

[0001] This invention relates to the field of maxillofacial image analysis technology, and in particular to a method for generating a maxillofacial soft tissue model for the orthodontic process. Background Technology

[0002] In orthodontic treatment, changes in the maxillofacial soft tissues are a key factor affecting the patient's appearance and treatment outcomes. Traditional soft tissue prediction relies heavily on experience or static models, lacking precise modeling and dynamic prediction tailored to individual differences. Establishing accurate coupling modeling methods for maxillofacial soft tissues and bone tissues has become an important research direction in the interdisciplinary fields of dentistry, maxillofacial surgery, and biomechanics. During orthodontic and orthognathic treatment, doctors need to accurately predict the facial soft tissue responses caused by tooth movement and jawbone remodeling to assist in treatment planning, simulation, postoperative evaluation, and patient communication.

[0003] The clinical applications of CBCT (Cone Beam Computed Tomography) are mainly concentrated in dentistry and radiotherapy. In dentistry, CBCT is sometimes specifically referred to as oral CBCT, primarily used for three-dimensional visualization of the entire tooth structure and diseased tissues. CBCT, using cone-beam X-ray scanning, significantly improves X-ray utilization. It only requires a 360° rotation (one full circle) to acquire all the raw data needed for reconstruction, unlike traditional CT which uses axial or spiral scanning and multi-slice data acquisition. In the generation of maxillofacial soft tissue during orthodontic procedures, current technologies generally first acquire facial bone and surface data using CBCT and 3D scanning techniques. However, CBCT and 3D scanning cannot directly reconstruct the maxillofacial soft tissue model based on the patient's facial information and bone data. Therefore, MRI (Magnetic Resonance Imaging) is also needed to acquire the maxillofacial soft tissue model data. However, MRI is time-consuming and computationally intensive in acquiring soft tissue information, making it unsuitable for clinical applications and more suited for laboratory research.

[0004] Therefore, an efficient and accurate method is needed to generate three-dimensional models of the maxillofacial soft tissues during the orthodontic process, in order to achieve more accurate, intuitive, and personalized orthodontic treatment planning and orthodontic prediction results. Summary of the Invention

[0005] To address the above technical problems, this invention provides a method for generating a maxillofacial soft tissue model for the orthodontic process.

[0006] A method for generating a maxillofacial soft tissue model for orthodontic procedures, provided by the present invention, the method comprising:

[0007] Obtain the current patient's facial bone data and historical model dataset;

[0008] Based on the reference attachment points in the historical model dataset, the facial bone data is matched with the facial bone information in the historical model dataset to obtain the simulated attachment points of the facial bone data.

[0009] Based on the simulated attachment point and the reference attachment point, a current finite metadata block and a reference finite metadata block are constructed respectively, and the corresponding reference finite metadata block of the current finite metadata block is obtained;

[0010] Analyze the distribution of various soft tissues in the corresponding reference finite metadata block to obtain the distribution probability set of various soft tissues in the current finite metadata block;

[0011] Based on the set of probability distributions, the current finite metadata block is divided to obtain several sub-finite metadata blocks and sub-sets of probability distributions;

[0012] Based on the sub-distribution probability set, the consistency of soft tissue distribution between the sub-finite metadata block and all its neighboring sub-finite metadata blocks is analyzed, and the sub-finite metadata blocks are merged to obtain the soft tissue model structure of the current patient.

[0013] In some embodiments of the present invention, based on reference attachment points in the historical model dataset, the facial bone data is matched with the historical model dataset to obtain simulated attachment points of the facial bone data, including:

[0014] Based on the facial bone data, a current bone model of the current patient is constructed;

[0015] The current skeleton model is registered with the historical skeleton model in the historical model dataset, and the reference attachment points in the historical model dataset are mapped to the current skeleton model to obtain the inheritance mapping points of the current skeleton model.

[0016] Analyze the referenceability of the reference attachment point corresponding to the inherited mapping point;

[0017] The shape feature matching degree between the inherited mapping point and the reference attachment point is analyzed, and the reliability of the inherited mapping point is obtained by combining the referenceability degree.

[0018] A preset confidence threshold is used to obtain the simulated attachment points of the facial bone data based on the confidence level.

[0019] In some embodiments of the present invention, analyzing the referability of the reference attachment point corresponding to the inherited mapping point includes:

[0020] Based on the reference attachment points in the historical model dataset, the historical model data corresponding to different sampled specimens in the historical model dataset are matched to obtain the number of times the reference attachment points are matched for each inherited mapping point.

[0021] Based on the number of matching occurrences and the number of sampled specimens in the historical model dataset, the referenceability of the reference attachment point corresponding to the inherited mapping point is obtained.

[0022] In some embodiments of the present invention, analyzing the shape feature matching degree between the inherited mapping point and the reference attachment point includes:

[0023] By analyzing the degree of difference in bone curvature and bone normal vector between the inherited mapping point and the reference attachment point, the shape feature matching degree between the inherited mapping point and the reference attachment point is obtained.

[0024] In some embodiments of the present invention, based on the simulated attachment point and the reference attachment point, a current finite metadata block and a reference finite metadata block are constructed respectively, and the corresponding reference finite metadata block of the current finite metadata block is obtained, including:

[0025] Obtain the current patient's facial surface data;

[0026] Starting from the simulated attachment point, a current finite metadata block is constructed. The soft tissue domain between the facial bone data and the facial surface data is segmented using the current finite metadata block to obtain the number of the current finite metadata blocks.

[0027] Starting from the reference attachment point and using the number of current finite metadata blocks as the segmentation number, the soft tissue domain in the historical model dataset is segmented to obtain the reference finite metadata blocks;

[0028] The current finite metadata block and the reference finite metadata block are matched one by one to obtain the corresponding reference finite metadata block of the current finite metadata block.

[0029] In some embodiments of the present invention, the distribution of various soft tissues in the corresponding reference finite metadata block is analyzed to obtain a set of distribution probabilities for various soft tissues in the current finite metadata block, including:

[0030] Obtain the reference volume of the reference finite metadata block and the reference tissue volume of various soft tissues in the reference finite metadata block;

[0031] Based on the reference tissue volume and the reference volume, the volume distribution of various soft tissues in the reference finite metadata block corresponding to the current finite metadata block is analyzed to obtain the distribution probability set of various soft tissues in the current finite metadata block.

[0032] In some embodiments of the present invention, the current finite metadata block is segmented based on the distribution probability set to obtain several sub-finite metadata blocks and sub-distribution probability sets, including:

[0033] Obtain the maximum probability in the set of probability distributions;

[0034] Preset probability threshold;

[0035] Determine whether the maximum probability is less than the probability threshold;

[0036] If so, the current finite metadata block is divided into two.

[0037] The distribution probability set of various soft tissues in the segmented current finite metadata block is re-acquired, and the probability threshold judgment is continued. The finite metadata block that does not meet the probability threshold is further segmented until the probability threshold is met, and then the segmentation operation is stopped, resulting in several sub-finite metadata blocks and the sub-distribution probability set corresponding to the sub-finite metadata blocks.

[0038] In some embodiments of the present invention, based on the sub-distribution probability set, the consistency of soft tissue distribution between the sub-finite metadata block and all its neighboring sub-finite metadata blocks is analyzed, and the sub-finite metadata blocks are merged to obtain the soft tissue model structure of the current patient, including:

[0039] Based on the sub-distribution probability set, the information entropy of the soft tissue distribution between the sub-finite metadata block and all its neighboring sub-finite metadata blocks is obtained. Combining the volume of the neighboring sub-finite metadata block and the identity between the soft tissue in the neighboring sub-finite metadata block and the soft tissue in the sub-finite metadata block, the consistency of the soft tissue distribution between the sub-finite metadata block and all its neighboring sub-finite metadata blocks is obtained.

[0040] Based on the consistency of the soft tissue distribution, the sub-finite metadata blocks are merged, and an information entropy threshold is set to limit the merging, so as to obtain the soft tissue model structure of the current patient.

[0041] In some embodiments of the present invention, obtaining the historical model dataset includes:

[0042] Historical soft tissue data of the maxillofacial region were obtained from the sampled specimens using MRI technology.

[0043] Historical facial bone data of the sampled specimens were obtained using CBCT technology.

[0044] Historical facial surface data of the sampled specimens were obtained using 3D technology.

[0045] Based on the historical maxillofacial soft tissue data, the historical facial bone data, and the historical facial surface data, a historical model dataset of the sampled specimens is obtained.

[0046] In some embodiments of the present invention, the method for obtaining reference attachment points in the historical model dataset includes:

[0047] By semantic segmentation, the historical skeletal model of each historical model data in the historical model dataset is obtained;

[0048] Reference attachment points on the historical skeletal model are obtained through manual annotation.

[0049] As can be seen from the above embodiments, the method for generating a maxillofacial soft tissue model for the orthodontic process provided by the present invention has the following beneficial effects:

[0050] This invention first matches the facial bone data of the current patient with the historical model dataset to obtain simulated attachment points for the facial bone data. Then, based on the simulated attachment points and reference attachment points in the historical model dataset, it constructs the current finite metadata block and the reference finite metadata block, and obtains the corresponding reference finite metadata block for the current finite metadata block. It then analyzes the distribution of various soft tissues in the corresponding reference finite metadata block to obtain the distribution probability set of various soft tissues in the current finite metadata block. Based on the distribution probability set, the current finite metadata block is segmented to obtain several sub-finite metadata blocks and sub-distribution probability sets. Finally, based on the sub-distribution probability sets, it analyzes the consistency of soft tissue distribution between the sub-finite metadata blocks and all their neighboring sub-finite metadata blocks, merges the sub-finite metadata blocks, and obtains the soft tissue model structure of the current patient. In other words, this invention constructs a limited metadata block of the current patient's data by analyzing the matching degree between the current patient's data and historical data; then, by combining the organizational distribution of the historical data itself, it analyzes the soft tissue distribution probability within the limited metadata block corresponding to the current patient; and finally, based on the soft tissue distribution probability, it segments and reassembles the limited metadata block corresponding to the current patient, ultimately obtaining a more accurate soft tissue model structure for the current patient.

[0051] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0052] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, 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 the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 This is a schematic diagram of the basic process of a method for generating a soft tissue model of the maxillofacial region for the process of orthodontics, provided by an embodiment of the present invention.

[0054] Figure 2 A schematic diagram of reference attachment points for a historical skeletal model provided in an embodiment of the present invention;

[0055] Figure 3 A schematic diagram of reference attachment points for another historical skeleton model provided in an embodiment of the present invention. Detailed Implementation

[0056] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method for generating a maxillofacial soft tissue model for the orthodontic process proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. Terms such as “comprising,” “including,” or any other variations thereof are intended to cover a non-exclusive inclusion, such that a circuit structure, article, or device comprising a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such article or device. Without further limitation, an element defined by the phrase “comprising one…” does not exclude the presence of additional identical elements in the article or device that includes the element.

[0058] The following section, in conjunction with the accompanying drawings, provides a detailed description of a method for generating a maxillofacial soft tissue model for the orthodontic process, as provided in this embodiment.

[0059] Please see Figure 1 This illustrates the basic flow of a method for generating a maxillofacial soft tissue model for the orthodontic process, provided by an embodiment of the present invention.

[0060] like Figure 1As shown, an embodiment of the present invention provides a method for generating a maxillofacial soft tissue model for the orthodontic process, which specifically includes the following steps:

[0061] S100: Obtain the current patient's facial bone data and historical model dataset.

[0062] Acquire the current patient's facial bone data and historical model dataset. Specifically, acquire the current patient's facial bone data using CBCT (Cone Beam Computed Tomography), and perform Otsu threshold segmentation on the acquired CBCT data to obtain the current patient's facial bone data; simultaneously, acquire the current patient's facial surface data using 3D scanning technology, thus obtaining the basic data for constructing the maxillofacial soft tissue model of the current patient's orthodontic process.

[0063] Under laboratory conditions, historical soft tissue data of multiple maxillofacial specimens were acquired using MRI (Magnetic Resonance Imaging); historical facial bone data of multiple specimens were acquired using CBCT; and historical facial surface data of multiple specimens were acquired using 3D scanning technology. Based on these historical soft tissue, bone, and surface data, historical model data of the specimens were obtained. Specifically, this data was imported into 3D Slicer software (which provides various tools and modules for processing and analyzing medical imaging data, including facial model construction), outputting historical model data of the specimens. This historical model data includes historical bone and soft tissue models. The historical model data from multiple specimens forms a historical model dataset. This dataset can provide data support for the subsequent construction of maxillofacial soft tissue models for current patients.

[0064] At this point, we have obtained the current patient's facial bone data and the historical model dataset of the sampled specimens.

[0065] S200: Based on the reference attachment points in the historical model dataset, the facial bone data is matched with the facial bone information in the historical model dataset to obtain the simulated attachment points of the facial bone data.

[0066] Skeletal model data serves as the supporting structure in the soft tissue model construction process. Therefore, soft tissue models can be generated by matching known skeletal model data of the current patient with that of the sampled specimen. The purpose of matching is to align the acquired skeletal model data due to the diverse morphologies involved. This matching process is accomplished by aligning the mechanically critical points within the skeletal model data.

[0067] Based on the above analysis, in some embodiments of the present invention, facial bone data is matched with the historical model dataset using reference attachment points in the historical model dataset to obtain simulated attachment points for the facial bone data. Further, this includes:

[0068] First, the reference attachment points in the historical model dataset are obtained. Specifically, semantic segmentation is used to remove other interference from the historical model data, obtaining the historical skeletal model. Then, the mechanical points (also known as attachment points, and their annotation is essential anatomical knowledge for medical students) on the historical skeletal model are manually annotated to obtain the reference attachment points, such as... Figure 2 and Figure 3 As shown. Based on the above operations, all historical model datasets are analyzed to obtain the historical skeletal model and its reference attachment point for each historical model data in the historical model dataset.

[0069] Simultaneously, based on facial bone data, a current skeletal model of the current patient is constructed. Specifically, the facial bone data of the current patient is matched with all historical skeletal models in the historical model dataset to obtain the skeletal network information of the current patient. A smoothing filter is then applied to the skeletal network information to remove noise while maintaining clear boundaries, thus obtaining the current skeletal model of the current patient.

[0070] Then, since manually annotating the attachment points (key points) of the current patient's skeletal model is inefficient and places high demands on doctors, it is necessary to use a historical model dataset as a reference template for automated annotation. Therefore, in some embodiments of this invention, the current skeletal model is registered with historical skeletal models in the historical model dataset, and the reference attachment points in the historical model dataset are mapped to the current skeletal model to obtain the inherited mapping points of the current skeletal model. Specifically, rigid registration (ICP algorithm) is used to align the position and proportion of the current patient's skeletal model with each historical skeletal model in the historical template dataset; then, non-rigid deformation registration is used to deform the historical skeletal model to the shape of the current skeletal model. At this time, the attachment points or attachment region labels of each historical skeletal model in the historical template dataset will be mapped to the current skeletal model with the registration deformation. Thus, the attachment points of the current skeletal model are automatically inherited from the historical skeletal models, resulting in the inherited mapping points of the current skeletal model, without the need for manual annotation.

[0071] Next, the referentiality of the reference attachment points in the historical model dataset is analyzed. Specifically, this includes: based on the reference attachment points in the historical model dataset, arbitrary pairwise matching of historical model data corresponding to different sampled specimens in the historical model dataset is performed through projection. This method aligns sample data of different forms. Simultaneously, the matching count of each reference attachment point is obtained (the number of times a reference attachment point in a given historical model data set can successfully match with reference attachment points in other historical model data sets during the matching process), i.e., the matching count of the reference attachment point corresponding to each inherited mapping point is obtained. Then, based on the matching count and the number of sampled specimens in the historical model dataset, the referentiality of the reference attachment points corresponding to the inherited mapping points is obtained. Inherited mapping points are then constructed. In the The corresponding reference attachment point in the historical model data The formula for calculating the reference level is:

[0072]

[0073] In the formula, Indicates inheritance mapping point In the The corresponding reference attachment point in the historical model data The degree of reference value; Indicates inheritance mapping point In the The corresponding reference attachment point in the historical model data The number of times it can be successfully matched with reference attachment points in other historical model data (in the 1st) Each historical model data point inherits the mapping point when matching the historical model dataset through the reference attachment point. Corresponding reference attachment point (Number of matches that can be performed) This indicates that the mapping point is inherited when historical model data is matched multiple times through the reference attachment point. The location corresponds to the number of reference attachment points in the historical model data (i.e., how many reference attachment points in the historical model data are simultaneously mapped to the inherited mapping point). Location); This indicates the number of historical model data (samples) in the historical model dataset.

[0074] Next, by analyzing the shape feature matching degree between the inherited mapping point and the reference attachment point, and combining this with the referenceability, the reliability of the inherited mapping point is obtained. Specifically, the degree of difference in bone curvature and bone normal vector at the locations of the inherited mapping point and the reference attachment point is analyzed to obtain the shape feature matching degree between the inherited mapping point and the reference attachment point. Bone curvature and bone normal vector represent the shape features of the bone data; for example, the mandibular angle, zygomatic process, and maxillary ridge have stable geometric features on the bone surface. Combining the shape feature matching degree and the referenceability, the reliability of the inherited mapping point is obtained, and the inherited mapping point is constructed. The formula for calculating credibility is:

[0075]

[0076] In the formula, Indicates inheritance mapping point Credibility; Indicates inheritance mapping point In the The corresponding reference attachment point in the historical model data The degree of reference value; Indicates inheritance mapping point The curvature of the bone at the current location in the skeletal model; Represents the mapping points between historical model data and inheritance. The average skeletal curvature at the corresponding positions of all reference attachment points that match the location; Indicates inheritance mapping point The bone normal vector at the current position of the skeletal model; Represents the mapping points between historical model data and inheritance. The mean of the bone normal vectors at the corresponding positions of all reference attachment points that match the current position; Representing vectors with vector The angle between them; Representing vectors with vector The sine of the angle between them; Represented by natural constant An exponential function with base 0; Indicates taking the absolute value; This indicates taking the length of the vector by its modulus.

[0077] This indicates the degree of difference in bone curvature between the inherited mapping point and the reference attachment point. The smaller the value, the more similar the bone curvature between the inherited mapping point and the reference attachment point, and the higher the reliability of the inherited mapping point. This indicates the degree of difference between the bone normal vectors at the locations of the inherited mapping point and the reference attachment point. The smaller the value, the more similar the bone normal vectors at the locations of the inherited mapping point and the reference attachment point, and the higher the reliability of the inherited mapping point.

[0078] Finally, a preset confidence threshold is established, and simulated attachment points for the facial bone data are obtained based on the confidence level. Specifically, the preset confidence threshold is 0.8. Inherited mapping points are filtered based on their confidence level, and those with a confidence level greater than the confidence threshold are set as simulated attachment points for the current patient's facial bone data. The corresponding inheritance mapping point is set as the simulated attachment point of the current patient's facial bone data.

[0079] At this point, the simulated attachment points of the current patient's facial bone data have been obtained.

[0080] Step S200 completes the matching of the patient's skeletal model with the skeletal models in the historical model dataset. When further constructing the current patient's soft tissue model using the distribution of soft tissues attached to the bones in the historical model dataset, it is necessary to consider the differences between different populations (different sampled specimens), such as differences in scale (the size of each soft tissue may vary). Therefore, finite metadata blocks can be constructed using the reference attachment points of the historical skeletal model and the simulated attachment points of the current skeletal model. Then, based on the distribution of various soft tissues within the finite metadata blocks in the historical model dataset, the probability distribution of various soft tissues in the finite metadata blocks is analyzed. The finite metadata blocks are then segmented and combined, and their size is reconstructed to obtain a more refined soft tissue model distribution. This specifically includes steps S300 to S600.

[0081] S300: Based on the simulated attachment point and the reference attachment point, construct the current finite metadata block and the reference finite metadata block respectively, and obtain the corresponding reference finite metadata block of the current finite metadata block.

[0082] Based on the simulated attachment point and the reference attachment point, the current finite metadata block and the reference finite metadata block are constructed respectively, and the corresponding reference finite metadata block of the current finite metadata block is obtained. Further, this includes:

[0083] First, the facial surface data of the current patient is acquired. The specific method for acquiring the facial surface data has been described in detail in step S100 and will not be repeated here. Next, the soft tissue domain between the current patient's facial bone data and facial surface data is acquired, where the soft tissue domain refers to the skin, fat, muscles, blood vessels, nerves, and connective tissue covering the facial bones. Specifically, a 3D model import library (such as Assimp) can be used to import and process 3D model data (facial bone data and facial surface data) to obtain the soft tissue domain; existing techniques will not be elaborated here.

[0084] Then, starting from the simulated attachment points in the current patient's current bone model, construct 128 128 The current finite metadata block is 128 (preset, needs to be set to cube). The soft tissue volume domain between the facial bone data and the facial surface data is segmented by the current finite metadata block, and the number of the current finite metadata blocks is obtained.

[0085] Simultaneously, starting from the reference attachment point in the historical skeletal model of the sampled specimen, and using the number of current finite metadata blocks as the segmentation number, the soft tissue domain in the historical model dataset is segmented to obtain reference finite metadata blocks; that is, when segmenting the soft tissue domain in the historical model dataset, the number of reference finite metadata blocks after final segmentation is the same as the number of current finite metadata blocks. The size of the reference finite metadata blocks needs to be changed according to the registration situation between the current skeletal model and the historical skeletal model in step S100, that is, the size of each reference finite metadata block is not the same and is not the same as the size of the current finite metadata block.

[0086] Then, by using attachment point matching (data blocks are segmented by attachment points), the current finite metadata block and the reference finite metadata block are matched one by one according to the positional relationship between the two golden attachment points of the data blocks to obtain the corresponding reference finite metadata block of the current finite metadata block.

[0087] S400: Analyze the distribution of various soft tissues in the corresponding reference finite metadata block to obtain the distribution probability set of various soft tissues in the current finite metadata block.

[0088] The distribution of various soft tissues in the corresponding reference finite data block is analyzed to obtain the probability set of various soft tissue distributions in the current finite data block. Specifically, this includes: First, obtaining the reference volume of the reference finite data block, and annotating the soft tissue structures in the historical soft tissue model using prior anatomical knowledge. The annotated soft tissue structures include the frontalis muscle, orbicularis oculi muscle, dorsum nasalis muscle, alar expander muscle, masseter muscle, levator labii superioris muscle, depressor labii superioris muscle, mandibular muscle, corrugator supercilii muscle, temporalis muscle, levator nasal septum muscle, suprazygomaticus major muscle, infrazygomaticus major muscle, orbicularis oris muscle, buccinator muscle, platysma muscle, sternocleidomastoid muscle, and trapezius muscle. Then, the type names and corresponding volumes of various soft tissues in the reference finite data block are obtained, denoted as the reference tissue volume. Next, based on the reference tissue volume and the reference volume, the volume distribution of various soft tissues in the reference finite data block corresponding to the current finite data block is analyzed to obtain the probability set of various soft tissue distributions in the current finite data block. The construction of the first... In the current finite metadata block, the first The formula for calculating the probability distribution of a soft tissue is:

[0089]

[0090] In the formula, Indicates the first In the current finite metadata block, the first The probability distribution of each soft tissue; This indicates the number of historical model data (samples) in the historical model dataset; Indicates the first In the current finite metadata block, the first The soft tissue in the first The reference tissue volume of the corresponding soft tissue in the corresponding reference finite metadata block in each historical model data; Indicates the first In the historical model data, compared with the first The reference volume of the reference finite metadata block (corresponding reference finite metadata block) that is matched with the current finite metadata block.

[0091] By analyzing the proportion of material volume (the ratio of soft tissue volume to total volume) in all corresponding reference finite data blocks of the current finite data block in the historical model dataset, the proportion of soft tissue in the current finite data block can be determined, i.e., the probability information of the presence of other substances in the current finite data block can be determined. The closer the value is to 1, the better the segmentation effect of the current finite data block is, and the more accurate the material content in the current finite data block is.

[0092] Similarly, based on the above operations, for the current patient's... Quantizing the probability distributions of various soft tissues contained in the current finite metadata block can yield the first... The set of probability distributions of various soft tissues in a current finite metadata block is denoted as { Similarly, obtain the set of probability distributions corresponding to all current finite metadata blocks for the current patient.

[0093] S500: Based on the distribution probability set, the current finite metadata block is divided into several sub-finite metadata blocks and sub-distribution probability sets.

[0094] After obtaining the set of probability distributions corresponding to all current finite metadata blocks for the current patient, the current finite metadata blocks are segmented based on these probability distributions to obtain several sub-finite metadata blocks and sub-probability distribution sets. Specifically, this involves first obtaining the maximum probability in the probability distribution set, denoted as . It also presets a probability threshold (which can be 0.85); then determines whether the maximum probability is less than the probability threshold; if so, that is... If the segmentation effect of the current finite metadata block is poor, then the current finite metadata block is segmented into two parts. Then, the distribution probability set of various soft tissues in the segmented current finite metadata block is re-obtained (the acquisition method is the same as step S400), and the probability threshold judgment is continued. Finite metadata blocks that do not meet the probability threshold are segmented again until the probability threshold is met and the segmentation operation is stopped.

[0095] The above segmentation operation can divide the soft tissue domain of the current patient into sub-finite data blocks of different sizes. Each sub-finite data block contains the distribution probability of various substances, thus obtaining several sub-finite data blocks corresponding to the current patient and the sub-distribution probability set corresponding to the sub-finite data blocks.

[0096] S600: Based on the sub-distribution probability set, analyze the soft tissue distribution consistency between the sub-finite metadata block and all its neighboring sub-finite metadata blocks, merge the sub-finite metadata blocks, and obtain the soft tissue model structure of the current patient.

[0097] The sub-finite data blocks obtained through segmentation in step S500 still contain a significant amount of noise (due to the inherent diversity of historical model data). Therefore, it is necessary to merge and remove sub-data blocks from the current patient's soft tissue domains after segmentation. For example, a small adipose tissue data block may exist within muscle tissue, and the current adipose tissue data block may be noisy. When merging and optimizing the segmented sub-finite data blocks, it is necessary to analyze the consistency of various tissue distributions between the merged sub-finite data blocks and their surrounding sub-finite data blocks, while also considering the loss when the merged sub-finite data blocks are treated as a single tissue.

[0098] Based on the above analysis, in an embodiment of the present invention, based on the sub-distribution probability set, the consistency of soft tissue distribution between a sub-finite metadata block and all its neighboring sub-finite metadata blocks is analyzed, and the sub-finite metadata blocks are merged to obtain the soft tissue model structure of the current patient. Further, this includes:

[0099] First, based on the sub-distribution probability set, the information entropy of the soft tissue distribution among the sub-finite metadata block and all its neighboring sub-finite metadata blocks within its eight spatial neighborhoods is obtained. Combining this with the volume of the neighboring sub-finite metadata blocks and the identity between the soft tissues in the neighboring sub-finite metadata blocks and the soft tissues in the sub-finite metadata blocks, the consistency of the soft tissue distribution among the sub-finite metadata block and all its neighboring sub-finite metadata blocks is obtained. Specifically, based on the sub-distribution probability set, the information entropy of the soft tissue distribution among the sub-finite metadata block and all its neighboring sub-finite metadata blocks within its eight spatial neighborhoods is obtained. Information entropy is a measure of the uncertainty of discrete random variables; the higher the probability of an event occurring, the lower the information entropy, and vice versa. Here, the information entropy of the soft tissue distribution represents the degree of dispersion of the soft tissue distribution among the sub-finite metadata block and all its neighboring sub-finite metadata blocks. The higher the information entropy of the soft tissue distribution, the greater the consistency of the soft tissue distribution, and the more necessary it is to merge the sub-finite metadata block with its neighboring sub-finite metadata blocks; the lower the information entropy of the soft tissue distribution, the less consistent the soft tissue distribution. Furthermore, a larger volume of a neighborhood sub-finite metadata block indicates a larger organization, making it more likely to be within the same organization; that is, a greater consistency in the soft organization distribution within the neighborhood sub-finite metadata block itself. The identity between soft organizations within a neighborhood sub-finite metadata block and those within a sub-finite metadata block—that is, whether the soft organizations within a neighborhood sub-finite metadata block are the same as those within the current neighborhood sub-finite metadata block—takes a value of 0 or 1, where 0 indicates they are not the same soft organization and 1 indicates they are the same soft organization. The greater the identity between soft organizations within a neighborhood sub-finite metadata block and those within a sub-finite metadata block, the greater the consistency in the soft organization distribution between the sub-finite metadata block and all its neighboring neighborhood sub-finite metadata blocks. Therefore, by combining the information entropy of the soft organization distribution with the volume of the neighborhood sub-finite metadata block and the identity between soft organizations within a neighborhood sub-finite metadata block and those within a sub-finite metadata block, the consistency in the soft organization distribution between a sub-finite metadata block and all its neighboring neighborhood sub-finite metadata blocks can be obtained. Constructing the... The formula for calculating the soft tissue distribution consistency between a finite metadata block and all its neighboring finite metadata blocks is:

[0100]

[0101] In the formula, Indicates the first The soft organization distribution consistency between a sub-finite metadata block and all neighboring sub-finite metadata blocks within its eight-neighborhood; Indicates the first Information entropy of the soft tissue distribution among a sub-finite metadata block and all neighboring sub-finite metadata blocks within its eight-neighborhood; Indicates the first The number of all neighborhood sub-finite metadata blocks within the eight-neighborhood of the sub-finite metadata block space; Indicates the first The first of the sub-finite metadata blocks The volume of a finite number of neighborhood sub-metadata blocks; Indicates the first The finite metadata block and its first The soft organization distribution consistency among the finite metadata blocks of the neighborhood sub-domains (its value is 0 or 1, where 0 indicates that they are not the same type of soft organization, i.e., the soft organization distribution is inconsistent; and 1 indicates that they are the same type of soft organization, i.e., the soft organization distribution is consistent). This represents the linear normalization function.

[0102] Based on the consistency of soft tissue distribution, sub-finite data blocks are merged, and an information entropy threshold is set to limit the merging process, resulting in the current patient's soft tissue model structure. Specifically, a consistency threshold (which can be 0.6) is preset, and sub-finite data blocks with soft tissue distribution consistency greater than the consistency threshold are merged with all neighboring sub-finite data blocks within their eight spatial neighborhoods. Furthermore, to avoid excessive merging leading to the loss of noise information, an information entropy threshold (which can be 0.8) is set, and the normalized information entropy of various soft tissue probabilities within each merged finite data block is calculated (the information entropy is normalized). Merged finite data blocks with an information entropy greater than the information entropy threshold are not merged.

[0103] Based on the above operations, the segmented finite metadata blocks can be merged and removed to obtain the mesh structure of the current patient's soft tissue. The mesh structure of the soft tissue can be simulated using existing algorithms such as meshlab (Mesh Laboratory) and DiffTaichi (differentiable programming for physical simulation) to obtain the soft tissue model structure of the current patient.

[0104] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0105] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for generating a maxillofacial soft tissue model for the orthodontic process, characterized in that, The method includes: Obtain the current patient's facial bone data and historical model dataset; Based on the reference attachment points in the historical model dataset, the facial bone data is matched with the facial bone information in the historical model dataset to obtain the simulated attachment points of the facial bone data. Based on the simulated attachment point and the reference attachment point, a current finite metadata block and a reference finite metadata block are constructed respectively, and the corresponding reference finite metadata block of the current finite metadata block is obtained; Analyze the distribution of various soft tissues in the corresponding reference finite metadata block to obtain the distribution probability set of various soft tissues in the current finite metadata block; Based on the set of probability distributions, the current finite metadata block is divided to obtain several sub-finite metadata blocks and sub-sets of probability distributions; Based on the sub-distribution probability set, the consistency of soft tissue distribution between the sub-finite metadata block and all its neighboring sub-finite metadata blocks is analyzed, and the sub-finite metadata blocks are merged to obtain the soft tissue model structure of the current patient.

2. The method for generating a maxillofacial soft tissue model for the orthodontic process according to claim 1, characterized in that, Based on the reference attachment points in the historical model dataset, the facial bone data is matched with the facial bone information in the historical model dataset to obtain the simulated attachment points of the facial bone data, including: Based on the facial bone data, a current bone model of the current patient is constructed; The current skeletal model is registered with the historical skeletal models in the historical model dataset, and the reference attachment points in the historical model dataset are mapped to the current skeletal model to obtain the inheritance mapping points of the current skeletal model. Analyze the referenceability of the reference attachment point corresponding to the inherited mapping point; The shape feature matching degree between the inherited mapping point and the reference attachment point is analyzed, and the reliability of the inherited mapping point is obtained by combining the referenceability degree. A preset confidence threshold is used to obtain the simulated attachment points of the facial bone data based on the confidence level.

3. The method for generating a maxillofacial soft tissue model for the orthodontic process according to claim 2, characterized in that, Analyzing the referability of the reference attachment point corresponding to the inherited mapping point includes: Based on the reference attachment points in the historical model dataset, the historical model data corresponding to different sampled specimens in the historical model dataset are matched to obtain the number of times the reference attachment points are matched for each inherited mapping point. Based on the number of matching occurrences and the number of sampled specimens in the historical model dataset, the referenceability of the reference attachment point corresponding to the inherited mapping point is obtained.

4. The method for generating a maxillofacial soft tissue model for the orthodontic process according to claim 2, characterized in that, Analyzing the shape feature matching degree between the inherited mapping point and the reference attachment point includes: By analyzing the degree of difference in bone curvature and bone normal vector between the inherited mapping point and the reference attachment point, the shape feature matching degree between the inherited mapping point and the reference attachment point is obtained.

5. The method for generating a maxillofacial soft tissue model for the orthodontic process according to claim 1, characterized in that, Based on the simulated attachment point and the reference attachment point, a current finite metadata block and a reference finite metadata block are constructed respectively, and the corresponding reference finite metadata block of the current finite metadata block is obtained, including: Obtain the current patient's facial surface data; Starting from the simulated attachment point, a current finite metadata block is constructed. The soft tissue domain between the facial bone data and the facial surface data is segmented using the current finite metadata block to obtain the number of the current finite metadata blocks. Starting from the reference attachment point and using the number of current finite metadata blocks as the segmentation number, the soft tissue domain in the historical model dataset is segmented to obtain the reference finite metadata blocks; The current finite metadata block and the reference finite metadata block are matched one by one to obtain the corresponding reference finite metadata block of the current finite metadata block.

6. The method for generating a maxillofacial soft tissue model for the orthodontic process according to claim 1, characterized in that, Analyzing the distribution of various soft tissues in the corresponding reference finite metadata block yields a set of distribution probabilities for various soft tissues in the current finite metadata block, including: Obtain the reference volume of the reference finite metadata block and the reference tissue volume of various soft tissues in the reference finite metadata block; Based on the reference tissue volume and the reference volume, the volume distribution of various soft tissues in the reference finite metadata block corresponding to the current finite metadata block is analyzed to obtain the distribution probability set of various soft tissues in the current finite metadata block.

7. The method for generating a maxillofacial soft tissue model for the orthodontic process according to claim 1, characterized in that, Based on the aforementioned probability distribution set, the current finite metadata block is segmented to obtain several sub-finite metadata blocks and sub-probability distribution sets, including: Obtain the maximum probability in the set of probability distributions; Preset probability threshold; Determine whether the maximum probability is less than the probability threshold; If so, the current finite metadata block is divided into two. The distribution probability set of various soft tissues in the segmented current finite metadata block is re-acquired, and the probability threshold judgment is continued. The finite metadata block that does not meet the probability threshold is further segmented until the probability threshold is met, and then the segmentation operation is stopped, resulting in several sub-finite metadata blocks and the sub-distribution probability set corresponding to the sub-finite metadata blocks.

8. The method for generating a maxillofacial soft tissue model for the orthodontic process according to claim 1, characterized in that, Based on the sub-distribution probability set, the consistency of soft tissue distribution between the sub-finite metadata block and all its neighboring sub-finite metadata blocks is analyzed. The sub-finite metadata blocks are then merged to obtain the soft tissue model structure of the current patient, including: Based on the sub-distribution probability set, the information entropy of the soft tissue distribution between the sub-finite metadata block and all its neighboring sub-finite metadata blocks is obtained. Combining the volume of the neighboring sub-finite metadata block and the identity between the soft tissue in the neighboring sub-finite metadata block and the soft tissue in the sub-finite metadata block, the consistency of the soft tissue distribution between the sub-finite metadata block and all its neighboring sub-finite metadata blocks is obtained. Based on the consistency of the soft tissue distribution, the sub-finite metadata blocks are merged, and an information entropy threshold is set to limit the merging, so as to obtain the soft tissue model structure of the current patient.

9. The method for generating a maxillofacial soft tissue model for the orthodontic process according to claim 1, characterized in that, Obtain the historical model dataset, including: Historical soft tissue data of the maxillofacial region were obtained from the sampled specimens using MRI technology. Historical facial bone data of the sampled specimens were obtained using CBCT technology. Historical facial surface data of the sampled specimens were obtained using 3D technology. Based on the historical maxillofacial soft tissue data, the historical facial bone data, and the historical facial surface data, a historical model dataset of the sampled specimens is obtained.

10. The method for generating a maxillofacial soft tissue model for the orthodontic process according to claim 1, characterized in that, The method for obtaining reference attachment points in the historical model dataset includes: By semantic segmentation, the historical skeletal model of each historical model data in the historical model dataset is obtained; Reference attachment points on the historical skeletal model are obtained through manual annotation.

Images