Dental registration method and readable storage medium

Abstract

The application provides a dental registration method and a readable storage medium, the dental registration method comprises the following steps: taking an oral virtual model established based on oral medical image data as a target point cloud, taking an oral scanning model obtained based on oral scanning as a point cloud to be registered, and performing coarse registration; selecting or generating a target key point set based on the oral virtual model, and selecting or generating a first constraint point in the oral virtual model; selecting or generating a source key point set corresponding to the target key point set based on the oral scanning model, and selecting or generating a second constraint point corresponding to the first constraint point in the oral scanning model; translating the point cloud to be registered after coarse registration so that the first constraint point coincides with the second constraint point; and performing fine registration of the point cloud to be registered and the target point cloud through iterative calculation of a bounded ICP algorithm based on at least three pairs of matched key points in the target key point set and the source key point set.

Description

Dental registration methods and readable storage media Technical Field

[0001] This invention relates to the field of medical device technology, and in particular to a dental registration method and a readable storage medium. Background Technology

[0002] In existing technologies, complete three-dimensional tooth models can be obtained through segmentation and reconstruction based on oral CBCT images. However, due to low scanning resolution and streak artifacts from metal restorations, this method cannot provide accurate occlusal surfaces and precise occlusal relationships.

[0003] The intraoral point cloud data obtained from laser scanning mainly contains three-dimensional information of the patient's tooth crowns, enabling precise tooth anatomy and accurate occlusal relationships. Therefore, in preoperative planning for dental implant surgery, the three-dimensional tooth model generated from CBCT images is primarily registered with the intraoral point cloud data to estimate the three-dimensional information of the tooth crown at the location of the missing tooth. The application of digital three-dimensional models of the patient's dentition significantly improves the efficiency of preoperative planning for dentists and the accuracy of dental implantation.

[0004] However, there are certain defects or shortcomings in the registration of 3D tooth models with intraoral scan point cloud data:

[0005] On the one hand, the registration of 3D tooth models reconstructed from CBCT images with intraoral scan point cloud data typically utilizes the Iterative Nearest Neighbor (ICP) algorithm to calculate the transformation matrix of certain feature points in the two sets of point clouds and find the registration transformation relationship. This method requires high accuracy of the corresponding feature points. If there are many feature points used for registration, the iterative calculation time is long, and it is easy to get trapped in local optima during the iteration process, making it impossible to obtain an accurate registration transformation relationship.

[0006] On the other hand, the registration of the 3D dental model with intraoral scan point cloud data places high demands on the initial pose of both the 3D dental model and the intraoral scan point cloud data. When the coarse registration error between the 3D dental model and the intraoral scan point cloud data is relatively large, the classic ICP algorithm or the registration calculation based on feature points may fail to converge, resulting in large errors. Unlike other hard tissues, dental implantation requires high registration accuracy, and in clinical practice, the registration error requirement is usually less than 0.5 mm. Therefore, the registration accuracy of traditional registration methods is unstable, and the shape of the tooth can affect the registration accuracy. Summary of the Invention

[0007] The purpose of this invention is to provide a dental registration method and a readable storage medium to solve the problems of long calculation time and unstable registration accuracy in existing registration methods.

[0008] To solve the above-mentioned technical problems, the present invention provides a dental registration method, which includes:

[0009] A virtual oral cavity model based on oral medical image data is used as the target point cloud, and an oral cavity scan model obtained from oral cavity scans is used as the point cloud to be registered, and coarse registration is performed.

[0010] Based on the oral cavity virtual model, a set of target key points is selected or generated, and a first constraint point is selected or generated in the oral cavity virtual model; based on the oral cavity scanning model, a set of source key points corresponding to the set of target key points is selected or generated, and a second constraint point corresponding to the first constraint point is selected or generated in the oral cavity scanning model.

[0011] The point cloud to be registered is translated after coarse registration so that the first constraint point coincides with the second constraint point; based on at least three pairs of compatible key points in the target key point set and the source key point set, the fine registration of the point cloud to be registered and the target point cloud is achieved through iterative calculation using the bounded ICP algorithm.

[0012] Optionally, in the dental registration method, the target key points in the target key point set are generated based on the following steps:

[0013] Extract the tooth model from the oral cavity virtual model;

[0014] Each point on the tooth model is configured according to a predetermined rule to correspond to the target key point of that tooth model.

[0015] Optionally, in the dental registration method, the step of configuring a point on each of the tooth models to correspond to the target key point of that tooth model according to a predetermined rule includes any of the following:

[0016] The intersection of the central axis of the outer bounding box of the tooth model and the crown is configured as the target key point;

[0017] Configure the centroid of the point cloud of the tooth model as the target key point;

[0018] The most prominent point on the lingual surface or lateral side of the tooth model is configured as the target key point;

[0019] The deepest point on the crown surface of the tooth model is configured as the target key point.

[0020] Optionally, in the dental registration method, the step of selecting the first constraint point includes:

[0021] Select at least three target key points located on the same plane from the set of target key points;

[0022] Using the centroid of the shape enclosed by at least three selected target key points as the base point, a point at a predetermined distance from the base point is selected as the first constraint point along the direction of the normal vector of the shape.

[0023] Optionally, in the dental registration method, the selected at least three target key points include target key points on the innermost tooth models located on both sides of the oral cavity in the oral virtual model, and also include target key points on the outermost tooth models located on both sides of the oral cavity in the oral virtual model.

[0024] Optionally, in the dental registration method, the target key points in the target key point set are generated based on the following steps:

[0025] Extract the tooth model from the oral cavity virtual model and determine a certain positioning point in the oral cavity virtual model;

[0026] The feature points on at least three adjacent or sub-adjacent tooth models on both sides of the positioning point are configured as the target key points.

[0027] Optionally, in the dental registration method, the step of selecting the first constraint point includes:

[0028] Using the sagittal plane as the interface, the point on the innermost tooth model on one side of the interface where the positioning point is located is selected as the first constraint point.

[0029] Optionally, in the dental registration method, the first constraint point is the intersection of the central axis of the outermost bounding frame of the tooth model on one side of the interface with the crown.

[0030] Optionally, in the dental registration method, the oral virtual model includes the location of the missing tooth, and the positioning point is the location of the missing tooth.

[0031] To address the aforementioned technical problems, the present invention also provides a readable storage medium having a program stored thereon, which, when executed, implements the steps of the dental registration method described above.

[0032] In summary, the dental registration method and readable storage medium provided by this invention include: performing coarse registration using a virtual oral model established based on oral medical image data as the target point cloud and an oral scanning model obtained based on oral scanning as the point cloud to be registered; selecting or generating a target key point set based on the virtual oral model and selecting or generating a first constraint point in the virtual oral model; selecting or generating a source key point set corresponding to the target key point set based on the oral scanning model and selecting or generating a second constraint point corresponding to the first constraint point in the oral scanning model; translating the coarsely registered point cloud to be registered so that the first constraint point coincides with the second constraint point; and achieving fine registration between the point cloud to be registered and the target point cloud by iterative calculation using a bounded ICP algorithm based on at least three pairs of compatible key points in the target key point set and the source key point set.

[0033] This configuration, based on the bounded ICP algorithm iterative calculation, uses relatively fixed first and second constraint points to constrain the fine registration process of the point cloud after coarse registration. On one hand, the number of point clouds required for registration is small, eliminating the need to use all point cloud information as a data source, thus reducing computational complexity and significantly lowering hardware requirements and iteration time. On the other hand, the first and second constraint points can serve as translation constraints, reducing the dependence of fine registration on the coarse registration results, lowering the risk of getting trapped in local optima during iteration, and improving the speed and accuracy of tooth registration. Attached Figure Description

[0034] Those skilled in the art will understand that the accompanying drawings are provided to better understand the invention and do not constitute any limitation on the scope of the invention. Wherein:

[0035] Figure 1 is a schematic diagram of the preoperative planning process for dental implant surgery according to the present invention;

[0036] Figure 2 is a schematic diagram of a virtual oral cavity model according to an embodiment of the present invention;

[0037] Figure 3 is a schematic diagram of laser scanning of the oral cavity according to an embodiment of the present invention;

[0038] Figure 4 is a schematic diagram of the lower teeth of the oral cavity virtual model according to an embodiment of the present invention;

[0039] Figure 5 is a schematic diagram of the lower teeth of the oral cavity scanning model according to an embodiment of the present invention;

[0040] Figure 6a is a schematic diagram of the first constraint point in an embodiment of the present invention;

[0041] Figure 6b is a schematic diagram of the second constraint point according to an embodiment of the present invention;

[0042] Figure 7 is a schematic diagram of fine registration of the target point cloud and the point cloud to be registered according to an embodiment of the present invention;

[0043] Figure 8a is a schematic diagram of the PCA algorithm for calculating the principal axis direction of the target point cloud according to an embodiment of the present invention;

[0044] Figure 8b is a schematic diagram of the PCA algorithm calculating the principal axis direction of the point cloud to be registered according to an embodiment of the present invention;

[0045] Figure 9 is a schematic diagram of coarse registration between the target point cloud and the point cloud to be registered according to an embodiment of the present invention;

[0046] Figure 10 is a schematic diagram of configuring the intersection of the central axis of the outer bounding frame of the tooth model and the crown as the target key point according to an embodiment of the present invention;

[0047] Figure 11a is a schematic diagram of the selection process of the first constraint point and the target key point in Embodiment 1 of the present invention;

[0048] Figure 11b is a schematic diagram of the selection process of the second constraint point and the source key point in Embodiment 1 of the present invention;

[0049] Figure 12 is a schematic diagram of a healthy permanent tooth oral cavity virtual model according to the present invention;

[0050] Figure 13a is a schematic diagram of the selection process of the first constraint point and the target key point in Embodiment 2 of the present invention;

[0051] Figure 13b is a schematic diagram of the selection process of the second constraint point and the source key point in Embodiment 2 of the present invention. Detailed Implementation

[0052] To make the objectives, advantages, and features of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the drawings are all in a very simplified form and are not drawn to scale, and are only used to facilitate and clarify the explanation of the embodiments of this invention. Furthermore, the structures shown in the drawings are often part of the actual structures. In particular, different figures may emphasize different aspects and may sometimes use different scales.

[0053] As used in this invention, the singular forms “a,” “an,” and “the” include plural objects; the term “or” is generally used to mean “and / or”; the term “a number” is generally used to mean “at least one”; and the term “at least two” is generally used to mean “two or more”. Furthermore, the terms “first,” “second,” and “third” are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as “first,” “second,” or “third” may explicitly or implicitly include one or at least two of that feature; “one end” and “the other end,” and “proximal end” and “distal end” generally refer to two corresponding parts, which include not only endpoints. Furthermore, the terms "installed," "connected," and "attached," as used in this invention, and the term "set" on one element from another, should be interpreted broadly. They generally only indicate a connection, coupling, cooperation, or transmission relationship between the two elements, which can be direct or indirect through an intermediate element. They should not be construed as indicating or implying a spatial relationship between the two elements, meaning one element can be located inside, outside, above, below, or to one side of another element, unless otherwise explicitly stated. Those skilled in the art can understand the specific meaning of these terms in this invention based on the specific circumstances. Additionally, directional terms such as above, below, up, down, upward, downward, left, and right are used relative to exemplary embodiments as shown in the figures, with upward or upper directions pointing towards the top of the corresponding figure, and downward or lower directions pointing towards the bottom of the corresponding figure.

[0054] The purpose of this invention is to provide a dental registration method and a readable storage medium to solve the problems of long calculation time and unstable registration accuracy in existing registration methods.

[0055] The following description refers to the accompanying drawings.

[0056] Figure 1 illustrates a preoperative planning process for dental implantation. As described in the background section, in order to determine the precise occlusal surface and accurate occlusal relationship during the preoperative planning process for dental implantation, it is first necessary to acquire oral medical image data (e.g., CBCT images) and oral scan data (e.g., intraoral point cloud data obtained from laser scanning). A virtual oral model is then established based on the oral medical image data, and an oral scan model is obtained based on the oral scan data. Next, the virtual oral model and the oral scan model need to be registered. After registration, the location of the missing tooth and implant information (such as implant size, implantation path, etc.) can be determined.

[0057] In traditional methods, the registration of a virtual oral cavity model and a scanned oral cavity model is often done manually, resulting in a large number of point clouds involved in fine registration and long computation time. Because it is impossible to select the points involved in fine registration, the registration effect is often unstable and susceptible to noise. To address these problems, this invention provides a dental registration method, comprising:

[0058] Step 1 S1: Use the oral virtual model established based on oral medical image data as the target point cloud and the oral scan model obtained based on oral scan as the point cloud to be registered, and perform coarse registration.

[0059] Step 2 S2: Select or generate a set of target key points based on the oral cavity virtual model, and select or generate a first constraint point in the oral cavity virtual model; select or generate a set of source key points corresponding to the set of target key points based on the oral cavity scanning model, and select or generate a second constraint point corresponding to the first constraint point in the oral cavity scanning model;

[0060] Step 3 (S3): Translate the coarsely registered point cloud to be registered so that the first constraint point coincides with the second constraint point; based on at least three pairs of matching key points in the target key point set and the source key point set, perform iterative calculations using the bounded ICP algorithm to achieve fine registration between the point cloud to be registered and the target point cloud.

[0061] Please refer to Figures 2 and 3. Figure 2 shows a virtual oral model 1 built based on oral medical image data (such as CBCT data). Figure 3 shows a scene of laser scanning of oral cavity 2. The process of building the virtual oral model in step S1 can be carried out manually or by using an AI network (artificial intelligence neural network) based on the characteristic relationship of voxels to construct the tooth model, thereby obtaining a three-dimensional and complete virtual oral model. The specific principles and methods of building a virtual oral model based on oral medical image data, and the principles and methods of obtaining an oral scanning model based on oral scanning, can be found in existing technologies and will not be elaborated here. After obtaining the virtual oral model and the oral scanning model, the virtual oral model is used as the target point cloud and the oral scanning model is used as the source point cloud to be registered. Then, coarse registration is performed between the target point cloud and the source point cloud. The coarse registration methods include, but are not limited to, the congruent four-point set (4PCS) algorithm and its improved algorithms, the local feature descriptor-based registration method, and the probability distribution-based NDT registration method, etc. The specific principles of coarse registration can be found in existing technologies and will not be elaborated here.

[0062] Please refer to Figures 4 and 5. In step S2, multiple target key points P can be selected or generated using the anatomical features of the teeth. t Target key point P tSpecific selection or generation rules can be set according to different scenarios and needs. For example, in the example shown in Figure 2, the oral cavity virtual model includes two rows of teeth, upper and lower. Taking the lower teeth as an example, as shown in Figure 4, the lower teeth include a total of 12 teeth. The center point of the upper surface of the crown of each tooth is selected as the target key point P. t These target key points P t The set of key points is the target key point set S. target {P t1 ,P t2 ,P t3 …P t12}. Then, based on the selected target key point P t Correspondingly, select or generate key points P for each target. t Corresponding source key point P s For example, in the example shown in Figure 5, corresponding to the oral cavity virtual model in Figure 2, the center point of the upper surface of the crown of each tooth is also selected as the source key point P. s These source key points P s The set of key points is the source key point set S. source {P s1 ,P s2 ,P s3 …P s12}

[0063] Understandably, based on the target key point set S obtained in step two S2 target {P t1 ,P t2 ,P t3 …P t12} and the source key point set S source {P s1 ,P s2 ,P s3 …P s12} are corresponding, where each target key point P ti With the corresponding source key point P si These are all matching key points (where i = 1, 2, 3... 12).

[0064] Furthermore, please refer to Figures 6a and 6b, and select the first constraint point P in the oral cavity virtual model. tc In the oral cavity scanning model, the second constraint point P is selected. sc This is to impose additional constraints on subsequent registration. The first constraint point is P. tc Preferably, it is far from the target key point P. t The far-end constraint point. The first constraint point P. tcThe selection can be aided by commonly used cephalometric landmarks, lines, and planes in oral medical images, or it can be manually selected by the surgeon based on the patient's occlusal features; this embodiment is not limited to either. This provides additional constraints for subsequent registration. Second constraint point P sc The selection or generation of [the parameter] should be related to the first constraint point P. tc Correspondingly, the same selection method and selection rules are used.

[0065] In particular, there is no requirement for the order of execution of step S1 and step S2. You can execute step S1 first for coarse registration, or you can execute step S2 first to select or generate key points and constraint points.

[0066] Please refer to Figure 7. Step S3 is the fine registration of the target point cloud and the point cloud to be registered. After coarse registration, the point cloud to be registered is translated so that the first constraint point P... tc With the second constraint point P sc They coincide. At this point, the centroid of the target point cloud (CBCT model centroid) C c With the first constraint point P tc Forming the first constraint axis P tc C c The point cloud centroid to be registered (Intraoral model centroid) C o With the second constraint point P sc Forming the second constraint axis P sc C o The registration transformation between the target point cloud and the point cloud to be registered can be viewed as the point cloud to be registered moving along the second constraint axis P. sc C o Rotate by an angle φ, then rotate by an angle θ along axis a, where axis a is the second constraint axis P. sc C o With the first constraint axis P tc C c The cross product of vectors. Furthermore, by performing multiple iterative calculations based on the bounded ICP algorithm, precise registration of the point cloud to be registered and the target point cloud can be achieved.

[0067] This configuration, based on the bounded ICP algorithm iterative calculation, uses relatively fixed first and second constraint points to constrain the fine registration process of the point cloud after coarse registration. On one hand, the number of point clouds required for registration is small, eliminating the need to use all point cloud information as a data source, thus reducing computational complexity and significantly lowering hardware requirements and iteration time. On the other hand, the first and second constraint points can serve as translation constraints, reducing the dependence of fine registration on the coarse registration results, lowering the risk of getting trapped in local optima during iteration, and improving the speed and accuracy of tooth registration.

[0068] The following is a detailed description with reference to several embodiments.

[0069] Example 1 uses CBCT images as dental image data for illustration.

[0070] Step S1 mainly achieves coarse registration. Please refer to Figures 8a and 8b. First, acquire CBCT image data, establish a virtual oral cavity model, and acquire an oral cavity scanning model. The virtual oral cavity model is the target point cloud C, and the oral cavity scanning model is the point cloud O to be registered. Then, the principal component analysis (PCA) method is used to extract the principal axis directions of the two point clouds, the target point cloud C and the point cloud O to be registered. Specifically, the centroid of the target point cloud C and the centroid of the point cloud O to be registered are calculated first using the following formula (1).

[0071]

[0072] Where C i O i Let C and O represent individual point clouds of the target point cloud and the point cloud to be registered, respectively. and Let C and O represent the centroids of the target point cloud C and the point cloud O to be registered, respectively.

[0073]

[0074] Then, according to the formula (2) above, the covariance matrix M of the target point cloud C and the point cloud O to be registered are calculated respectively. c M o By performing singular value decomposition on the covariance matrices, we can obtain the 3*3 eigenvectors E of the two covariance matrices. c E o That is, the main axis direction of the two point clouds, namely the target point cloud C and the point cloud O to be registered.

[0075] R0 = E c E o -1 (3)

[0076]

[0077] According to the principal axis direction matrix E of the target point cloud C in formula (3) above. c The principal axis direction matrix E of the point cloud O to be registered o Solve for the initial rotation matrix R0. Calculate the translation vector T0 between the target point cloud C and the point cloud O to be registered according to the formula (4) above. Multiply the calculated rotation matrix R0 and translation vector T0 on the left by the point cloud O to be registered, as shown in Figure 9, to achieve coarse registration between the target point cloud C and the point cloud O to be registered.

[0078] Step S2 primarily involves selecting or generating key points and constraint points. Optionally, the target key points in the target key point set are generated based on the following steps:

[0079] Step S21: Extract the tooth model from the oral cavity virtual model; the tooth model refers to the virtual model corresponding to a single tooth, that is, each tooth corresponds to a tooth model.

[0080] Step S22: Configure one point on each tooth model as a target keypoint corresponding to that tooth model according to predetermined rules. These predetermined rules can be reasonably selected based on the application scenario. Each tooth model corresponds to only one target keypoint, ensuring a certain distance between target keypoints, which helps improve registration accuracy.

[0081] Optionally, step S22, which configures a point on each tooth model according to a predetermined rule to correspond to a target key point of that tooth model, includes any of the following:

[0082] S221: Configure the intersection of the central axis of the outer bounding box of the tooth model and the crown as the target key point Pt; as shown in Figure 10;

[0083] S222: Configure the centroid of the point cloud of the tooth model as the target key point;

[0084] S223: Configure the most prominent point on the lingual surface or lingual side of the tooth model as the target key point;

[0085] S224: Configure the deepest point on the crown surface of the tooth model as the target key point.

[0086] Preferably, for the oral cavity virtual model, all tooth models are extracted. Referring to Figure 4, taking the lower teeth as an example, it contains 12 tooth models. Based on the above step S22 (e.g., step S221), 12 target key points can be obtained, thus obtaining the target key point set S. target {P t1 ,P t2 ,P t3 …P t12 Furthermore, a set S of target key points is generated by manually selecting or based on the nearest neighbor method after coarse registration. target The corresponding set of source key points S source {P s1 ,P s2 ,P s3 …P s12 As shown in Figure 5.

[0087] Please refer to Figure 11a. The steps for selecting the first constraint point include:

[0088] Step S24: In the target key point set S target Select at least three target key points located on the same plane;

[0089] Step S25: Using the centroid of the shape enclosed by the selected at least three target key points as the base point, select the point P at a predetermined distance from the base point along the direction of the normal vector of the shape as the first constraint point. tc .

[0090] Preferably, the selected at least three target key points include target key points on the innermost tooth models located on both sides of the oral cavity in the oral cavity virtual model, and also include target key points on the outermost tooth models located on both sides of the oral cavity in the oral cavity virtual model. Please refer to Figure 12 for an example of an oral cavity virtual model. Taking the sagittal plane 10 as the interface, the direction closer to the sagittal plane 10 is called the lateral direction, and the direction farther from the sagittal plane 10 is called the medial direction. Thus, in a healthy permanent tooth oral cavity virtual model, the tooth model of the central incisor 11 is the outermost tooth model, while the tooth model of the third molar 12 is the innermost tooth model.

[0091] To maximize the coverage area of ​​the graphic enclosed by the target key points in step S25, on one hand, target key points on the tooth models of the third molars 12 on both sides of the oral cavity virtual model can be selected, such as the center points m1 and m2 of the crown surface, as the target key points on the tooth models located on the innermost sides of the oral cavity. It is understood that if one or both third molars 12 are missing, target key points on the tooth models of the second molars 13 can be selected, and so on. On the other hand, target key points on the tooth models of the left or right central incisors 11 on the oral cavity virtual model can be selected, such as the center point m3 of the crown surface, as the target key point on the tooth models located on the outermost sides of the oral cavity. Figure 11a shows three target key points selected, and a feature triangle is established using m1, m2, and m3 as three vertices. Based on this feature triangle, its centroid is used as the base point, and along the normal vector direction of the plane containing the feature triangle, a point at a predetermined length l is selected as the first constraint point P. tc The predetermined length l can be determined by the surgeon based on the patient's condition. Specifically, the number of target key points selected in step S24 can be three, preferably no more than five, to reduce computational load.

[0092] Please refer to Figure 11b, the source key point and the second constraint point P. sc The selection should be based on the target key point and the first constraint point P. tc The selection corresponds to this, so you can refer to steps S24 and S25 above, which will not be explained in detail here.

[0093] Please refer to Figure 13a and Figure 7. Step S3 is for fine registration.

[0094] First, the point cloud O to be registered is translated so that the first constraint point P tc With the second constraint point P sc Overlap. The centroid C of the graphic formed by the selected target key points of the target point cloud C is calculated using the following formula (5). c The centroid C of the graph formed by the source key points selected by the point cloud O to be registered is calculated using the following formula (6). o .

[0095]

[0096]

[0097] Where P ci For at least three target key points selected in step S24, P oi For the source key point set S source China and P ci The corresponding source key points. Then, the centroid C of the target point cloud C is calculated according to the following formula (7). c To the first constraint point P tc Similarly, using the following formula (8), we can obtain the centroid C of the point cloud O to be registered, which is the vector q. o To the second constraint point P sc The vector s.

[0098] q = C c -P tc (7)

[0099] s=C o -P sc (8)

[0100] Using the formula for the angle between vectors (9), the angle θ between vectors q and s can be calculated, which is the rotation angle of the centroid of the point cloud O to be registered and the target point cloud C during the registration process.

[0101]

[0102] After rotating the point cloud O to be registered by an angle θ around the vector a, the vector q and the vector s have coincided. At this time, the point cloud O to be registered needs to spin around the vector q or the vector s to complete the final registration. Calculate the spin angle φ of the point cloud O to be registered.

[0103] Furthermore, the individual target keypoint P in the target point cloud C is calculated using the following formula (10). ci To the first constraint point Ptc vector m i And the following formula (11) is used to calculate a single source key point P in the point cloud O to be registered. oi To the second constraint point P sc vector n i .

[0104] m i =P ci -P sc (10)

[0105] n i =P oi -P sc (11)

[0106] Where, vector m i The part b perpendicular to vector s mi As shown in equation (12), vector n i The part b perpendicular to vector s ni As shown in equation (13):

[0107]

[0108]

[0109] Then, the single source key point P is obtained by using the formula for the angle between vectors (14). oi During the registration process, the rotation angle φ around s i :

[0110]

[0111] The source key point P is obtained using the following formula (15). oi The average rotation angle φ of the set, where S s This is the set of source key points for the point cloud O to be registered.

[0112]

[0113] Move the point cloud O to be registered along P tc C o After rotating φ, the registered point cloud is obtained, and the process is iterated until the accuracy requirements are met, thereby achieving fine registration from the point cloud O to the target point cloud C, which is to say, achieving fine registration from the oral cavity scanning model to the oral cavity virtual model.

[0114] Understandably, the registration steps in Example 1 are particularly suitable for application scenarios where patients have a small number of remaining teeth, as they can effectively utilize existing teeth for registration.

[0115] Example 2, please refer to Figures 13a and 13b.

[0116] In Example 2, the coarse registration process in step S1 can be the same or similar to that in Example 1. Please refer to Example 1 above, and it will not be repeated here.

[0117] Step S2, which involves selecting or generating key points and constraint points, differs from Example 1. Specifically, in Example 2, the target key points in the target key point set are generated based on the following steps:

[0118] Step S26: Extract the tooth model from the oral cavity virtual model and determine a specific positioning point within the model. Similarly, a tooth model refers to a virtual model corresponding to a single tooth; each tooth corresponds to one tooth model. The positioning point is a selected reference point within the oral cavity virtual model, which can be chosen by the surgeon or automatically calculated based on the model.

[0119] Step S27: Configure feature points on at least three adjacent or second-adjacent tooth models on both sides of the positioning point as target key points.

[0120] Preferably, the oral cavity virtual model includes the location of the missing tooth, with the location of the missing tooth serving as the positioning point. In some application scenarios, registration focuses on certain local locations in the oral cavity. For example, in applications where several teeth are missing and dental implants are needed, the registration accuracy of teeth adjacent to the missing tooth is relatively important, while the registration accuracy of areas far from the missing tooth has less impact on the implantation process. Therefore, the location of the missing tooth in the oral cavity virtual model can be set as the positioning point, and the tooth models of one or two adjacent or second-adjacent teeth on both sides of the missing tooth location can be used as regions of interest (ROIs) for registration. Here, adjacent teeth refer to teeth that are immediately next to the location of the missing tooth in the tooth arrangement sequence, while second-adjacent teeth refer to teeth that are immediately next to the aforementioned adjacent teeth in the tooth arrangement sequence, i.e., separated from the location of the missing tooth by one tooth. This configuration effectively improves the registration accuracy near the missing tooth location, reduces computational load, and increases registration speed, since the target key points are all near the location of the missing tooth.

[0121] Referring to Figures 13a and 13b, in one example, the left central incisor of sagittal plane 10 is missing, while the right central incisor 11 is intact, and both lateral incisors 14 are intact. In this case, the missing central incisor on the left side of sagittal plane 10 is referred to as the missing tooth, and in the established virtual oral model, the location of this missing central incisor is called the missing tooth location and is configured as a positioning point. The tooth model selected in step S27 can be the tooth model of the right central incisor 11, which is symmetrical about the location of the missing tooth with respect to sagittal plane 10, and the tooth models of the two lateral incisors 14 on both sides. Furthermore, the feature points can be selected based on the anatomical characteristics of the teeth.

[0122] Furthermore, the steps for selecting the first constraint point include:

[0123] Step S28: Using the sagittal plane 10 as the interface, select a point on the innermost tooth model on one side of the interface as the first constraint point. For example, in the examples shown in Figures 13a and 13b, the first constraint point is the intersection of the central axis of the outermost bounding frame of the tooth model on one side of the interface (i.e., the left side) and the crown. Similarly, the innermost point here can refer to the definition in Embodiment 1 above. When the third molar 12 is present, the innermost tooth model is the tooth model of the third molar 12. If the third molar 12 is missing, the tooth model of the second molar 13 can be selected as the innermost tooth model, and so on.

[0124] It is understandable that the selection of the source key point and the second constraint point should correspond to the selection of the target key point and the first constraint point. Therefore, you can refer to steps S26 to S28 above, which will not be explained in detail here.

[0125] The fine registration process in step S3 can be the same or similar to that in Example 1. Please refer to Example 1 above, and it will not be repeated here.

[0126] Understandably, the registration steps in Example 2 are particularly suitable for application scenarios where patients retain a large number of teeth. The key registration points are all located near the missing teeth, which can effectively improve the registration accuracy near the missing teeth, reduce the amount of calculation, and increase the registration speed.

[0127] This invention also provides a readable storage medium storing a program that, when executed, implements the steps of the dental registration method described above. Furthermore, this invention provides a surgical robot system that uses the dental registration method described above to register a virtual oral model and an oral scan model. The surgical robot system then performs surgical operations based on the registration results, such as dental implant surgery. Optionally, the readable storage medium can be set up independently or integrated into the surgical robot system; this invention is not limited in this regard. For the structure and principles of other components of the surgical robot system, please refer to the prior art; this invention will not elaborate on them further.

[0128] In summary, the dental registration method and readable storage medium provided by this invention include: coarse registration using a virtual oral model established based on oral medical image data as the target point cloud and an oral scanning model obtained based on oral scanning as the point cloud to be registered; selecting or generating a target key point set based on the virtual oral model and selecting or generating a first constraint point in the virtual oral model; selecting or generating a source key point set corresponding to the target key point set based on the oral scanning model and selecting or generating a second constraint point corresponding to the first constraint point in the oral scanning model; translating the point cloud to be registered after coarse registration so that the first constraint point coincides with the second constraint point; and achieving fine registration between the point cloud to be registered and the target point cloud by iterative calculation using a bounded ICP algorithm based on at least three pairs of compatible key points in the target key point set and the source key point set.

[0129] This configuration, based on the bounded ICP algorithm iterative calculation, uses relatively fixed first and second constraint points to constrain the fine registration process of the point cloud after coarse registration. On one hand, the number of point clouds required for registration is small, eliminating the need to use all point cloud information as a data source, thus reducing computational complexity and significantly lowering hardware requirements and iteration time. On the other hand, the first and second constraint points can serve as translation constraints, reducing the dependence of fine registration on the coarse registration results, lowering the risk of getting trapped in local optima during iteration, and improving the speed and accuracy of tooth registration.

[0130] It should be noted that the above embodiments can be combined with each other. The above description is only a description of preferred embodiments of the present invention and is not intended to limit the scope of the present invention in any way. Any changes or modifications made by those skilled in the art based on the above disclosure shall fall within the protection scope of the claims.

Claims

1. A dental registration method, characterized in that, include: A virtual oral cavity model based on oral medical image data is used as the target point cloud, and an oral cavity scan model obtained from oral cavity scans is used as the point cloud to be registered, and coarse registration is performed. Based on the oral cavity virtual model, a set of target key points is selected or generated, and a first constraint point is selected or generated in the oral cavity virtual model. Taking the centroid of the shape enclosed by the selected at least three target key points as the base point, along the direction of the normal vector of the shape, a point at a predetermined distance from the base point is selected as the first constraint point. Based on the oral cavity scanning model, a source key point set corresponding to the target key point set is selected or generated, and a second constraint point corresponding to the first constraint point is selected or generated in the oral cavity scanning model to provide additional constraints for subsequent registration; the point cloud to be registered after coarse registration is translated so that the first constraint point and the second constraint point coincide. Based on at least three pairs of compatible key points in the target key point set and the source key point set, the fine registration of the point cloud to be registered with the target point cloud is achieved through iterative calculation using the bounded ICP algorithm.

2. The dental registration method according to claim 1, characterized in that, The target key points in the target key point set are generated based on the following steps: extracting tooth models from the oral cavity virtual model; configuring a point on each tooth model as the target key point corresponding to that tooth model according to a predetermined rule.

3. The dental registration method according to claim 2, characterized in that, The step of configuring a point on each of the tooth models as a target key point corresponding to the tooth model according to a predetermined rule includes any of the following: configuring the intersection of the central axis of the outer bounding frame of the tooth model and the crown as the target key point; Configure the centroid of the point cloud of the tooth model as the target key point; The most prominent point on the lingual surface or lateral side of the tooth model is configured as the target key point; The deepest point on the crown surface of the tooth model is configured as the target key point.

4. The dental registration method according to claim 2, characterized in that, The selection step of the first constraint point includes: selecting at least three target key points located on the same plane from the set of target key points; taking the centroid of the figure enclosed by the selected at least three target key points as the base point, and selecting a point at a predetermined distance from the base point along the direction of the normal vector of the figure as the first constraint point.

5. The dental registration method according to claim 4, characterized in that, The selected at least three target key points include target key points on the innermost tooth models located on both sides of the oral cavity in the oral cavity virtual model, and also include target key points on the outermost tooth models located on both sides of the oral cavity in the oral cavity virtual model.

6. The dental registration method according to claim 1, characterized in that, The target key points in the target key point set are generated based on the following steps: extracting the tooth model from the oral cavity virtual model, determining a certain positioning point in the oral cavity virtual model; configuring feature points on at least three adjacent or sub-adjacent tooth models on both sides of the positioning point as the target key points.

7. The dental registration method according to claim 6, characterized in that, The selection step of the first constraint point includes: taking the sagittal plane as the interface, selecting a point on the tooth model on the innermost side of the interface as the first constraint point.

8. The dental registration method according to claim 7, characterized in that, The first constraint point is the intersection of the central axis of the outermost bounding frame of the tooth model on one side of the interface with the crown.

9. The dental registration method according to claim 6, characterized in that, The oral cavity virtual model includes the location of the missing tooth, and the positioning point is the location of the missing tooth.

10. A readable storage medium having a program stored thereon, characterized in that, When the program is executed, it performs the steps of the dental registration method according to any one of claims 1 to 9.

Images