Artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system

Abstract

The present application relates to the technical field of region segmentation, in particular to an alveolar bone intelligent recognition and autogenous tooth transplantation scoring system based on artificial intelligence. The system acquires an alveolar crypt of a receiving area; according to the gray value carried by the in-point cloud of a voxel in the alveolar crypt of the receiving area, edge fuzzy voxels are screened and regarded as first-level fuzzy voxels, the first-level fuzzy voxels are divided into second-level voxels, and the second-level fuzzy voxels are acquired according to the similar situation of the fuzzy change direction of the second-level voxels and the first-level fuzzy voxels in which the second-level voxels are located; the second-level fuzzy voxels are divided to acquire third-level fuzzy voxels in the same way; the process is sequentially repeated until the stable performance of n-level fuzzy voxels and n-1-level fuzzy voxels in which the n-level fuzzy voxels are located is consistent, and the division is stopped; the inner edge of the alveolar crypt is acquired through the final fuzzy voxels; and the adaptation degree is acquired according to the matching of autogenous teeth and the inner edge of the alveolar crypt to score the autogenous tooth transplantation. The present application accurately acquires the inner edge of the alveolar crypt, effectively improves the accuracy of the matching of autogenous teeth and alveolar bone, and the transplantation effect of autogenous teeth.

Description

Technical Field

[0001] This invention relates to the field of region segmentation technology, specifically to an artificial intelligence-based intelligent alveolar bone recognition and autologous tooth transplantation scoring system. Background Technology

[0002] Tooth loss not only impairs a patient's chewing function, speech clarity, and facial aesthetics, but can also lead to secondary conditions such as temporomandibular joint disorder. Autologous tooth transplantation, as a highly biocompatible restorative method, involves transplanting the patient's own third molars or ectopic teeth to the edentulous area, avoiding rejection of foreign materials and effectively restoring dentition integrity and occlusal function.

[0003] Existing methods use CBCT (Cone-Beam Computed Tomography) reconstruction algorithms to backproject two-dimensional projection images of the patient's oral cavity acquired by cone-beam CT into a three-dimensional oral model. Then, the compatibility between the inner edge of the alveolar socket (i.e., part of the alveolar bone, the part where the tooth root contacts the alveolar bone) in the implant area and the autologous tooth (transplant tooth) in the three-dimensional oral model is analyzed to evaluate the effect of autologous tooth transplantation. This effectively reduces the time of autologous tooth removal and the number of times the alveolar socket in the implant area is matched with the autologous tooth, thereby improving the success rate of autologous tooth transplantation. However, in practice, considering that inflammation may occur in the alveolar socket of the implant site at the same time as tooth extraction and is difficult to control, autologous tooth transplantation surgery is usually performed in the early stage of tooth extraction (i.e., within 2-3 weeks after extraction). At this time, the periodontal ligament and bundle bone have been absorbed, and the alveolar bone in the implant site has undergone irreversible atrophy and absorption, the bone wall has become thinner and the bone density has decreased. When the thickness of the alveolar bone wall is less than that of the CBCT reconstructed voxel, the identification of the inner edge of the alveolar socket will be inaccurate, thus making it impossible to accurately assess the fit between the autologous tooth and the inner edge of the alveolar socket, affecting the transplantation effect of the autologous tooth. Summary of the Invention

[0004] To address the technical problem of inaccurate identification of the alveolar bone's inner edge in the implantation site, which leads to an inability to accurately assess the fit between the autologous tooth and the alveolar bone's inner edge and affects the outcome of autologous tooth transplantation, this invention aims to provide an artificial intelligence-based intelligent alveolar bone identification and autologous tooth transplantation scoring system. The specific technical solution adopted is as follows:

[0005] This invention provides an artificial intelligence-based intelligent alveolar bone recognition and autologous tooth transplantation scoring system, which includes:

[0006] The implantation area alveolar socket acquisition module is used to acquire the implantation area alveolar socket in the oral cavity three-dimensional model;

[0007] The edge-blurred voxel acquisition module is used to obtain the gray-level disorder of each voxel in the implantation area alveolar socket based on the disorder of gray-level values ​​carried by the point cloud in the neighboring voxels; and to filter out edge-blurred voxels based on the gray-level disorder.

[0008] The alveolar socket inner edge acquisition module is used to treat each edge fuzzy voxel as a first-level fuzzy voxel, and to obtain second-level voxels by pre-defined division of each first-level fuzzy voxel. Based on the similarity of the fuzzing change direction of each second-level voxel to the first-level fuzzy voxel, a second-level fuzzy voxel is obtained. Each second-level fuzzy voxel is then pre-defined to obtain third-level voxels, and based on the similarity of the fuzzing change direction of each third-level voxel to the second-level fuzzy voxel, a third-level fuzzy voxel is obtained. This process continues until each n-level fuzzy voxel has a stable performance consistent with its n-1-level fuzzy voxel, at which point the division stops. The final fuzzy voxels after the division stops are smoothly connected to the corresponding parts as the inner edge of the alveolar socket in the implant area.

[0009] The scoring module is used to obtain the degree of fit between the autogenous tooth and the alveolar bone based on the matching of the autogenous tooth and the inner edge of the alveolar socket; and to score the autogenous tooth transplant based on the degree of fit.

[0010] Furthermore, the method for obtaining the degree of grayscale disorder is as follows:

[0011] For any voxel in the alveolar socket of the implantation area, the mean value of the gray value carried by the point cloud in the voxel is used as the gray feature factor of the voxel.

[0012] The difference between the maximum and minimum gray-scale feature factors in the voxel and its neighboring voxels is obtained and used as the first fluctuation analysis value.

[0013] The mean of the gray-scale feature factors of this voxel and its neighboring voxels is used as the reference gray-scale feature factor.

[0014] The mean difference between the gray-scale feature factor of the voxel and each of its neighboring voxels and the reference gray-scale feature factor is used as the second fluctuation analysis value.

[0015] The normalized product of the first and second fluctuation analysis values ​​is used as the degree of grayscale disorder of the voxel.

[0016] Furthermore, the method for obtaining the edge-blurred voxels is as follows:

[0017] When the grayscale disorder level is greater than the preset grayscale disorder level threshold, the corresponding voxel is used as the edge blur voxel.

[0018] Furthermore, the method for obtaining the second-level fuzzy voxels is as follows:

[0019] For any first-level fuzzy voxel, the first-level fuzzy voxel is pre-divided into multiple sub-voxels, all of which are used as second-level voxels of the first-level fuzzy voxel.

[0020] Based on the difference in the direction of grayscale disorder between the first-level fuzzy voxel and its adjacent first-level fuzzy voxels, the direction of fuzziness change of the first-level fuzzy voxel is obtained.

[0021] For any second-level voxel of the first-level fuzzy voxel, the result of negatively correlated and normalized the difference between the fuzzy change direction of the second-level voxel and the first-level fuzzy voxel is taken as the fuzziness degree of the second-level voxel.

[0022] When the degree of ambiguity is greater than the preset ambiguity threshold, the second-level voxel is a second-level fuzzy voxel.

[0023] Furthermore, the method for obtaining the fuzzy change direction is as follows:

[0024] Based on the second derivative of the gray-level disorder degree between the first-level fuzzy voxel and other first-level fuzzy voxels in its first preset neighborhood as a function of spatial location, the growth curvature of the first-level fuzzy voxel is obtained through the Hessian matrix.

[0025] All adjacent first-level fuzzy voxels of the first-level fuzzy voxel are used as reference voxels of the first-level fuzzy voxel.

[0026] For any reference voxel, the center distance between the first-level fuzzy voxel and the reference voxel is taken as the first distance;

[0027] The magnitude of the difference in growth curvature between the first-level fuzzy voxel and the reference voxel is taken as the first value;

[0028] The arctangent of the first value and the ratio of the first distance is used as the reference direction of change between the first-level fuzzy voxel and the reference voxel.

[0029] The average of the reference change directions between the first-level fuzzy voxel and all its reference voxels is taken as the fuzzy change direction of the first-level fuzzy voxel.

[0030] In the process of obtaining the fuzzy change direction of the secondary voxel, the voxels and reference voxels in the first preset neighborhood of the secondary voxel are all other secondary voxels of the primary fuzzy voxel.

[0031] Furthermore, the method for obtaining the third-level fuzzy voxels is as follows:

[0032] For any second-level fuzzy voxel, the second-level fuzzy voxel is pre-divided into multiple sub-voxels, all of which are used as third-level voxels of the second-level fuzzy voxel.

[0033] For any tertiary voxel of the second-level fuzzy voxel, the result of negatively correlated and normalized the difference between the fuzzy change direction of the tertiary voxel and the second-level fuzzy voxel is taken as the fuzziness degree of the tertiary voxel; wherein, in the process of obtaining the fuzzy change direction of the tertiary voxel, the voxels and reference voxels in the first preset neighborhood of the tertiary voxel are all other tertiary voxels of the second-level fuzzy voxel.

[0034] When the degree of ambiguity is greater than the preset ambiguity threshold, the third-level voxel is a third-level fuzzy voxel.

[0035] Furthermore, the method for stopping the division is as follows:

[0036] The stability of the i-th (n-1)-th level fuzzy voxel is obtained based on the difference in grayscale disorder between the i-th (n-1)-th level fuzzy voxel and its adjacent (n-1)-th level fuzzy voxels, as well as the difference in the direction of fuzzy change; where, ;

[0037] All n-level fuzzy voxels divided by the i-th (n-1)th level fuzzy voxel are taken as target voxels;

[0038] For any target voxel, the division of the target voxel stops when the stability of the target voxel is equal to the stability of the i-th (n-1)th level fuzzy voxel.

[0039] Furthermore, the method for obtaining the stability level is as follows:

[0040] The (n-1)th level fuzzy voxel whose gray-level disorder is closest to that of the (i-1)th level fuzzy voxel is taken as the stable analysis voxel of the (i-1)th level fuzzy voxel.

[0041] The difference in grayscale disorder between the i-th (n-1)th level fuzzy voxel and its stable analysis voxel is taken as the first difference.

[0042] The difference in the direction of fuzzy change between the i-th (n-1)th level fuzzy voxel and its stable analysis voxel is taken as the second difference.

[0043] The sum of the first and second differences, followed by negative correlation and normalization, is used as the stability of the i-th (n-1)th level fuzzy voxel.

[0044] Furthermore, the method for obtaining the degree of adaptation is as follows:

[0045] Matching feature points in the inner edge of autogenous teeth and alveolar sockets are obtained using a key point detection algorithm;

[0046] Identify the structural centerline of the autogenous tooth and the alveolar socket in the implantation area, and perform rigid registration of matching feature points in the inner edge of the autogenous tooth and the alveolar socket by translation and rotation to obtain matching point pairs;

[0047] The mean of the distances between all matching points is negatively correlated and normalized to determine the degree of fit between the autogenous tooth and the alveolar bone.

[0048] Furthermore, the method for scoring autologous tooth transplantation based on fit is as follows:

[0049] When the degree of fit is greater than or equal to the first preset degree of fit threshold, there is a compression situation in autologous tooth transplantation;

[0050] When the degree of fit is greater than or equal to the second preset degree of fit threshold and less than the first preset degree of fit threshold, autologous tooth transplantation is reasonable.

[0051] When the degree of fit is less than the second preset degree of fit threshold, autologous tooth transplantation is not reasonable.

[0052] The present invention has the following beneficial effects:

[0053] This invention first determines the degree of grayscale disorder of each voxel in the implantation site's alveolar socket based on the disorder of grayscale values ​​carried by the point cloud of each voxel and its neighboring voxels, accurately reflecting the degree of disorder of the information contained in each voxel. Then, based on the degree of grayscale disorder, it accurately filters out marginally blurred voxels, which is beneficial for subsequent accurate and efficient identification of the alveolar socket's inner edge. To accurately identify the alveolar socket's inner edge and enable accurate analysis of the autologous tooth transplantation, each marginally blurred voxel is treated as a first-level blurred voxel. Each first-level blurred voxel is then pre-divided to obtain second-level voxels. Based on the similarity of the blurring direction between each second-level voxel and its corresponding first-level blurred voxel, third-level blurred voxels are obtained. Similarly, each second-level blurred voxel is pre-divided to obtain third-level voxels, thus enabling more accurate analysis of the location of the alveolar socket's inner edge. To avoid over-segmenting the fuzzy voxels at the edges and wasting unnecessary data resources, the segmentation stops when each n-level fuzzy voxel's stable performance matches that of its corresponding n-1-level fuzzy voxel. This accurately determines the voxel corresponding to the final edge of the alveolar socket. Therefore, the area corresponding to the smoothed connection of the final fuzzy voxels after segmentation is stopped is taken as the inner edge of the alveolar socket in the implantation area. This allows for accurate analysis of the fit between the autogenous tooth and alveolar bone. Based on the matching between the autogenous tooth and the inner edge of the alveolar socket, the degree of fit is obtained, accurately reflecting the matching status. This allows for accurate scoring of the autogenous tooth transplant based on the degree of fit, effectively reducing the number of matching attempts and the time required for tooth removal, thus improving the efficiency and effectiveness of autogenous tooth transplantation. Attached Figure Description

[0054] 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.

[0055] Figure 1 This is a structural block diagram of an artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system provided in one embodiment of the present invention;

[0056] Figure 2 This is a schematic diagram of a computer device provided according to an embodiment of the present invention. Detailed Implementation

[0057] 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 the artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system 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.

[0058] 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.

[0059] The specific solution of the artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0060] Example 1:

[0061] This invention proposes an artificial intelligence-based intelligent alveolar bone recognition and autologous tooth transplant scoring system. Please refer to [link / reference]. Figure 1 The diagram shows a structural block diagram of an artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system provided in an embodiment of the present invention. The system includes: a recipient alveolar socket acquisition module 10, a margin fuzzy voxel acquisition module 20, an alveolar socket inner margin acquisition module 30, and a scoring module 40.

[0062] The implantation area alveolar socket acquisition module 10 is used to acquire the implantation area alveolar socket in the three-dimensional oral cavity model.

[0063] Specifically, tooth extraction in the edentulous area can be categorized into simultaneous extraction, early extraction, and delayed extraction. Because simultaneous extraction results in severe and difficult-to-control inflammation of the extraction socket, autologous tooth transplantation is performed in the early extraction stage (i.e., within 2-3 weeks after extraction). This example uses a patient requiring autologous tooth transplantation in the early extraction stage as an example. It should be noted that subsequent patient references refer to the aforementioned patient requiring autologous tooth transplantation. The patient is seated with the orbitoauricular plane parallel to the horizontal plane and the sagittal plane perpendicular to the horizontal plane. An occlusal splint is worn, and the oral cavity remains still during the scan, with the upper and lower teeth in occlusion. This example uses a high-definition large-field cone-beam CT scanner to scan the patient's oral and maxillofacial region. The exposure voltage is set to 100kV, the current to 4mA, the scan time to 13s, and the field of view to 16cm × 8cm. The operator can adjust various parameters of the high-definition large-field cone-beam CT scanner according to the actual situation; these are not limited here. The oral cavity scan was exported in DICOM (Digital Imaging and Communications in Medicine) format, obtaining a set of two-dimensional projection images of the patient's oral cavity. These two-dimensional projection images were then back-projected into a three-dimensional model of the oral cavity using the CBCT (Cone-Beam Computed Tomography) reconstruction algorithm. The CBCT reconstruction algorithm is a well-known technique and will not be described in detail here.

[0064] To analyze autologous tooth transplantation more efficiently, this embodiment uses a semantic segmentation network to process the 3D oral cavity model and obtain the alveolar socket of the implant area from the 3D model. Specifically, the semantic segmentation network used in this embodiment is the U-net neural network. The input is the 3D oral cavity model; the output is a 3D model containing only the alveolar socket of the implant area. The U-net neural network is trained and labeled as follows: the area of ​​the alveolar socket of the implant area is marked as 1, and other areas are marked as 0. The loss function of the U-net neural network is the cross-entropy loss function. The U-net neural network is a well-known technology and will not be described in detail further.

[0065] The edge-blurred voxel acquisition module 20 is used to obtain the gray-level disorder of each voxel in the implantation area alveolar socket based on the disorder of gray-level values ​​carried by each voxel in the implantation area alveolar socket and its neighboring voxels; and to filter out edge-blurred voxels based on the gray-level disorder.

[0066] Specifically, after tooth loss in the implantation site, the alveolar bone gradually absorbs and atrophies, leading to decreased bone density and thinning of the bone wall. Given the limited voxel size corresponding to CBCT reconstruction, when the bone wall thickness is smaller than the reconstructed voxel grid, a single voxel in the implantation site's alveolar socket will simultaneously contain both the alveolar bone region and other periodontal tissue regions adjacent to the alveolar bone (such as reparative granulation tissue regions). This results in blurred edges within the alveolar socket, affecting the accuracy of matching the autogenous tooth with the alveolar bone. To accurately and efficiently identify the inner edges of the alveolar socket in the implantation site, this embodiment first needs to find the blurred edge voxels in the implantation site's alveolar socket that simultaneously contain both alveolar bone and periodontal tissue, and then perform higher-resolution sub-voxel segmentation on these blurred edge voxels. In this embodiment, for any voxel in the implantation site alveolar socket, if the voxel is a voxel with blurred edges, then the voxel must be surrounded by alveolar bone tissue and other periodontal tissues. Since alveolar bone tissue and other periodontal tissues are significantly different, the grayscale values ​​carried by the point cloud of this voxel and its neighboring voxels will inevitably fluctuate significantly. Therefore, this embodiment obtains the grayscale disorder level of each voxel in the implantation site alveolar socket based on the disorder of grayscale values ​​carried by the point cloud of each voxel and its neighboring voxels. The greater the grayscale disorder level, the more likely the corresponding voxel is to be a voxel with blurred edges. Therefore, filtering voxels with blurred edges based on the grayscale disorder level is beneficial for more efficient identification of the inner edge of the alveolar socket in the implantation site. It should be noted that for any voxel in the implantation site alveolar socket, this embodiment constructs a [database / structure] centered on this voxel. The window is used to define all other voxels belonging to the alveolar socket of the implantation area as neighboring voxels. The implementer can set the size and position of the window according to the actual situation, and there are no restrictions here.

[0067] Preferably, in one feasible embodiment of this invention, the method for obtaining the degree of grayscale disorder is as follows: For any voxel in the alveolar socket of the implantation area, the mean value of the grayscale values ​​carried by the point cloud in the voxel is taken as the grayscale feature factor of the voxel; the difference between the maximum and minimum grayscale feature factors in the voxel and its neighboring voxels is obtained as the first fluctuation analysis value; the mean value of the grayscale feature factors of the voxel and its neighboring voxels is taken as the reference grayscale feature factor; the mean value of the absolute values ​​of the differences between the grayscale feature factors of the voxel and each of its neighboring voxels and the reference grayscale feature factor is taken as the second fluctuation analysis value; when both the first and second fluctuation analysis values ​​are larger, it indicates that the grayscale distribution within the voxel is more disordered. In order to accurately characterize the grayscale disorder within the voxel, the product of the first and second fluctuation analysis values ​​is normalized, and the result is taken as the degree of grayscale disorder of the voxel. In this embodiment, the product of the first and second fluctuation analysis values ​​is normalized using the norm normalization function.

[0068] Thus, the gray-scale confusion degree of each voxel in the alveolar socket of the recipient area is obtained.

[0069] It is known that the greater the gray-scale confusion degree, the more likely the corresponding voxel is a voxel with blurred edges. Furthermore, in this embodiment, the preset gray-scale confusion degree threshold is set to 0.65. The implementer can set the size of the preset gray-scale confusion degree threshold according to the actual situation, which is not limited here. When the gray-scale confusion degree is greater than the preset gray-scale confusion degree threshold, the corresponding voxel is regarded as a voxel with blurred edges.

[0070] Thus, the voxels with blurred edges in the alveolar socket of the recipient area are determined.

[0071] The alveolar socket inner edge acquisition module 30 is configured to regard each voxel with blurred edges as a first-level blurred voxel, perform a preset division on each first-level blurred voxel to obtain second-level voxels, and obtain second-level blurred voxels according to the similarity of the blurred change direction between each second-level voxel and the first-level blurred voxel where it is located; perform a preset division on each second-level blurred voxel to obtain third-level voxels, and obtain third-level blurred voxels according to the similarity of the blurred change direction between each third-level voxel and the second-level blurred voxel where it is located; and so on, until the stable performance of each n-level blurred voxel is consistent with that of the n-1 level blurred voxel where it is located, and the division stops; the finally blurred voxels after the division stops are smoothly connected to the corresponding parts as the alveolar socket inner edge in the alveolar socket of the recipient area.

[0072] Specifically, in order to accurately identify the alveolar socket inner edge in the alveolar socket of the recipient area and at the same time avoid the normal area being submerged by noise and excessive waste of data resources caused by too small voxels, therefore, in this embodiment, each voxel with blurred edges is specifically divided into smaller voxels. In order to make the division of the voxels with blurred edges reasonable and avoid over-division of the voxels with blurred edges, therefore, in this embodiment, each voxel with blurred edges is first regarded as a first-level blurred voxel, and then a preset division is performed on each first-level blurred voxel to obtain second-level voxels. It should be noted that the rule of the preset division set in this embodiment is to trisect the side length of each first-level blurred voxel, that is, 1 first-level blurred voxel will be divided into 27 second-level voxels. The implementer can set the rule of the preset division according to the actual situation, which is not limited here.

[0073] To further determine the inner edge of the alveolar socket, this embodiment analyzes the similarity of the fuzzy change direction of each secondary voxel to that of its corresponding primary fuzzy voxel. Since the inner edge of the alveolar socket is continuous, its continuity characteristics remain constant in both primary and secondary voxels. Therefore, this embodiment compares the fuzzy change direction of each secondary voxel to that of its corresponding primary fuzzy voxel to screen for secondary fuzzy voxels that simultaneously contain alveolar bone and other periodontal tissues. The fuzzy change direction is essentially the change direction of the inner edge of the alveolar socket. To more accurately describe the process of obtaining the inner edge of the alveolar socket by dividing it into fuzzy voxels, this embodiment further follows the method for obtaining secondary fuzzy voxels. Each secondary fuzzy voxel is pre-divided into tertiary voxels. Based on the similarity of the fuzzy change direction of each tertiary voxel to the secondary fuzzy voxel, tertiary fuzzy voxels are obtained. This process is repeated until each n-level fuzzy voxel exhibits a stable performance consistent with its (n-1)-level fuzzy voxel, indicating that the corresponding n-level fuzzy voxel has maximally represented the inner edge features of the alveolar socket. At this point, the division of the corresponding n-level fuzzy voxels stops. The final fuzzy voxel after all fuzzy voxels have stopped dividing is the voxel corresponding to the inner edge of the alveolar socket. Therefore, in this embodiment, the corresponding part of the smooth connection of the final fuzzy voxels after the division stops is taken as the inner edge of the alveolar socket in the implantation area.

[0074] Preferably, in one feasible embodiment of this invention, the method for obtaining the second-level fuzzy voxels is as follows: for any first-level fuzzy voxel, the first-level fuzzy voxel is pre-divided into multiple sub-voxels, all of which are used as second-level voxels of the first-level fuzzy voxel; based on the difference in the direction of grayscale disorder between the first-level fuzzy voxel and its adjacent first-level fuzzy voxels, the fuzzy change direction of the first-level fuzzy voxel is obtained; wherein, adjacent indicates that the two voxels are closely adjacent. The method for obtaining the fuzzy change direction is as follows: based on the second derivative of the grayscale disorder with spatial position between the first-level fuzzy voxel and other first-level fuzzy voxels in its first pre-preset neighborhood, the growth curvature of the first-level fuzzy voxel is obtained through the Hessian matrix; wherein, the growth curvature is the eigenvector corresponding to the largest eigenvalue of the Hessian matrix. It should be noted that in this embodiment, the size of the first pre-preset neighborhood is set to... The implementer can set the size of the first preset neighborhood according to the actual situation, which is not limited here. The second derivative and Hessian matrix are well-known techniques and will not be elaborated further. All first-level fuzzy voxels adjacent to the first-level fuzzy voxel are taken as reference voxels of the first-level fuzzy voxel. For any reference voxel, the center distance between the first-level fuzzy voxel and the reference voxel is taken as the first distance. The magnitude of the difference in growth curvature between the first-level fuzzy voxel and the reference voxel is taken as the first value. The arctangent of the ratio of the first value to the first distance is taken as the reference change direction between the first-level fuzzy voxel and the reference voxel. To comprehensively characterize the change direction of the alveolar socket edge corresponding to the first-level fuzzy voxel, the average of the reference change directions between the first-level fuzzy voxel and all its reference voxels is taken as the fuzzy change direction of the first-level fuzzy voxel.

[0075] For any second-level voxel of the first-level fuzzy voxel, the fuzzy change direction of the second-level voxel is obtained according to the aforementioned method for obtaining the fuzzy change direction of the first-level fuzzy voxel. Specifically, in obtaining the fuzzy change direction of the second-level voxel, the voxels within the first preset neighborhood of the second-level voxel and the reference voxel are all other second-level voxels of the first-level fuzzy voxel. Specifically, when obtaining the grayscale disorder level of the second-level voxel, only other second-level voxels of the first-level fuzzy voxel within the window of the second-level voxel are analyzed; when obtaining the growth curvature of the second-level voxel, only other second-level voxels belonging to the first-level fuzzy voxel within the first preset neighborhood of the second-level voxel are analyzed; when obtaining the fuzzy change direction of the second-level voxel, all other second-level voxels adjacent to the second-level voxel belonging to the first-level fuzzy voxel are used as reference voxels of the second-level voxel.

[0076] When the direction of fuzziness change of the secondary voxel is more similar to that of the primary fuzzy voxel, the secondary voxel is more likely to be the voxel located at the inner edge of the alveolar socket. The absolute value of the difference between the fuzziness change directions of the secondary voxel and the primary fuzzy voxel is negatively correlated and normalized, and this result is taken as the fuzziness degree of the secondary voxel. In this embodiment, the negative of the absolute value of the difference is used as the power of an exponential function with the natural constant as the base. The output of this exponential function is the negatively correlated and normalized result of the absolute value of the difference. The greater the fuzziness degree, the more likely the secondary voxel is a new fuzzy voxel. Therefore, this embodiment sets a preset fuzziness degree threshold of 0.75. The implementer can set the size of the preset fuzziness degree threshold according to the actual situation, which is not limited here. When the fuzziness degree is greater than the preset fuzziness degree threshold, the secondary voxel is a secondary fuzzy voxel.

[0077] At this point, all secondary fuzzy voxels derived from the primary fuzzy voxels have been selected.

[0078] Preferably, in one feasible embodiment of this invention, the method for obtaining the third-level fuzzy voxels is as follows: For any second-level fuzzy voxel, the second-level fuzzy voxel is pre-divided into multiple sub-voxels, all of which are considered as third-level voxels of the second-level fuzzy voxel; for any third-level voxel of the second-level fuzzy voxel, the result of negatively correlated and normalized the difference in fuzzy change direction between the third-level voxel and the second-level fuzzy voxel is taken as the fuzziness degree of the third-level voxel; wherein, in the process of obtaining the fuzzy change direction of the third-level voxel, the voxels and reference voxels within the first pre-preset neighborhood of the third-level voxel are all other third-level voxels of the second-level fuzzy voxel; when the fuzziness degree is greater than a pre-preset fuzziness degree threshold, the third-level voxel is a third-level fuzzy voxel. The third-level fuzzy voxels are obtained through the above-described method for obtaining second-level fuzzy voxels, and the specific process will not be elaborated further. The method for obtaining the fuzzy change direction of the third-level voxels is the same as the method for obtaining the fuzzy change direction of the second-level voxels, and will not be elaborated further.

[0079] At this point, all the secondary fuzzy voxels have been screened out to identify the tertiary fuzzy voxels.

[0080] Preferably, in one feasible embodiment, the method for stopping the division is as follows: considering the smooth continuity of the inner edge of the alveolar socket, the performance characteristics of adjacent voxels corresponding to the inner edge of the alveolar socket should be similar. Therefore, in this embodiment, the stability of the i-th (n-1)th level fuzzy voxel is obtained based on the difference in grayscale disorder and the difference in fuzzy change direction between the i-th (n-1)th level fuzzy voxel and its adjacent (n-1)th level fuzzy voxels; wherein, The specific method for obtaining the stability level is as follows: The nearest neighboring voxel to the i-th (n-1)-th level fuzzy voxel with the closest grayscale disorder level to the i-th (n-1)-th level fuzzy voxel is taken as the stable analysis voxel of the i-th (n-1)-th level fuzzy voxel. Selecting a stable analysis voxel allows for a more accurate analysis of the stability of the i-th (n-1)-th level fuzzy voxel's changes. The absolute value of the difference in grayscale disorder between the i-th (n-1)-th level fuzzy voxel and its stable analysis voxel is taken as the first difference. The absolute value of the difference in the fuzzy change direction between the i-th (n-1)-th level fuzzy voxel and its stable analysis voxel is taken as the second difference. The smaller both the first and second differences are, the more stable the performance characteristics of the i-th (n-1)-th level fuzzy voxel. Therefore, in this embodiment, the result of negatively correlated and normalized summation of the first and second differences is taken as the stability level of the i-th (n-1)-th level fuzzy voxel. In this embodiment, the negative of the sum of the first difference and the second difference is used as the power of an exponential function with the natural constant as the base. The output of this exponential function is the result of adding the first difference and the second difference, followed by negative correlation and normalization.

[0081] All n-level fuzzy voxels derived from the i-th (n-1)-th level fuzzy voxel are considered as target voxels. For any target voxel, the closer its stability is to that of the i-th (n-1)-th level fuzzy voxel, the less likely it is to be further subdivided into smaller voxels. Therefore, subdivision of the target voxel stops when its stability equals that of the i-th (n-1)-th level fuzzy voxel. It should be noted that subdivision also stops when the edge length of the target voxel is less than 3 pixels.

[0082] This process identifies smaller fuzzy voxels from the final division of the fuzzy voxels, allowing for more accurate determination of the alveolar socket margin. This facilitates subsequent accurate analysis of the fit between the autogenous tooth and the alveolar socket margin, thereby improving the transplantation outcome of the autogenous tooth.

[0083] The scoring module 40 is used to obtain the degree of fit between the autogenous tooth and the alveolar bone based on the matching of the autogenous tooth and the inner edge of the alveolar socket; and to score the autogenous tooth transplant based on the degree of fit.

[0084] Specifically, in order to analyze the effect of autologous tooth transplantation, this embodiment first determines the degree of fit between the autologous tooth and the inner edge of the alveolar socket based on the matching situation; then, the autologous tooth transplantation is analyzed based on the degree of fit, which can predict the effect of autologous tooth transplantation in advance and facilitate the corresponding point grinding of the inner edge of the alveolar socket to improve the effect of autologous tooth transplantation.

[0085] Preferably, in one feasible embodiment of this invention, the method for obtaining the degree of fit is as follows: Matching feature points in the inner edge of the autogenous tooth and alveolar socket are obtained using a key point detection algorithm; the structural center lines of the autogenous tooth and the alveolar socket in the implantation area are identified; and the matching feature points in the inner edge of the autogenous tooth and alveolar socket are rigidly registered through translation and rotation to obtain matching point pairs; the smaller the distance between matching point pairs, the better the fit between the autogenous tooth and the inner edge of the alveolar socket. Furthermore, in this embodiment, the mean distances of all matching point pairs are negatively correlated and normalized, and this result is used as the degree of fit between the autogenous tooth and the inner edge of the alveolar socket. In this embodiment, the negative of the mean distances is used as the power of an exponential function with a natural constant as the base; the output of this exponential function is the result of negatively correlated and normalized mean distances of all matching point pairs. The key point detection algorithm and feature point matching are well-known techniques and will not be described in detail here.

[0086] Preferably, in one feasible way of this embodiment, the method for scoring autologous tooth transplantation based on the degree of fit is as follows: First, this embodiment sets a first preset degree of fit threshold of 0.9 and a second preset degree of fit threshold of 0.7. The implementer can set the size of the first preset degree of fit threshold and the second preset degree of fit threshold according to the actual situation, which is not limited here. When the fit is greater than or equal to the first preset fit threshold, the autogenous tooth is overfitted to the inner edge of the alveolar socket. Considering the potential presence of reparative granulation tissue within the alveolar socket, overfitting can lead to compression after transplantation. In this case, to ensure successful implantation and avoid excessive compression, the inner edge of the alveolar socket should be partially ground down. The ground-down portion should maintain a distance of 2mm from the buccal bone wall and 1mm from the lingual or palatal bone wall, without damaging important anatomical structures such as the mandibular canal, maxillary sinus floor, and nasal floor. When the fit is greater than or equal to the second preset fit threshold but less than the first preset fit threshold, the structural features of the autogenous tooth fit the alveolar socket, and the autogenous tooth transplantation is reasonable. When the fit is less than the second preset fit threshold, the autogenous tooth is too small and cannot fully fit the alveolar socket, making the autogenous tooth transplantation unreasonable.

[0087] Therefore, by scoring the autologous tooth transplant, the geometric features of the alveolar socket margin and the morphology of the autologous tooth can be evaluated in advance to assess the transplant effect. This reduces the number of times the autologous tooth is matched with the alveolar socket margin, and lowers the risk of damage to the periodontal ligament and anatomical structure of the implant area caused by the adaptive matching between the alveolar socket margin and the autologous tooth during surgery. In this way, the time of autologous tooth removal is reduced, and the efficiency and effect of autologous tooth transplantation are effectively improved.

[0088] In summary, this embodiment obtains the alveolar socket of the implantation site; based on the grayscale values ​​carried by the point cloud within the voxel in the alveolar socket, all edge-blurred voxels are selected as first-level blurred voxels. These first-level blurred voxels are then divided into second-level voxels, and second-level blurred voxels are obtained based on the similarity of their blurring change directions to those of their corresponding first-level voxels. These second-level blurred voxels are then divided into third-level voxels, and so on, until the nth-level blurred voxel exhibits a stable performance consistent with its corresponding n-1th-level blurred voxel. The inner edge of the alveolar socket is obtained through these final blurred voxels. The degree of fit between the autogenous tooth and the inner edge of the alveolar socket is assessed to score the autogenous tooth transplantation. This invention, by accurately obtaining the inner edge of the alveolar socket, effectively improves the accuracy of matching the autogenous tooth with the implantation site's alveolar socket and enhances the transplantation effect.

[0089] Example 2:

[0090] This invention also proposes an artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring device. This device includes a memory and a processor. The memory stores executable program code, and the processor is used to call and execute the executable program code to execute the artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system provided in the embodiments of this application. Specifically, the device may be a chip, component, or module. The chip may include a connected processor and memory; the memory stores instructions, and when the processor calls and executes the instructions, it enables the chip to execute the artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system provided in the above embodiments.

[0091] In addition, this embodiment also protects a computer device; please refer to [link to relevant documentation]. Figure 2 The computer device includes a memory 401, a processor 402, and a computer program 403 stored in the memory 401 and running on the processor 402. When the processor 402 executes the computer program 403, the computer device can execute any of the aforementioned artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring systems.

[0092] Example 3:

[0093] The present invention also provides a computer-readable storage medium storing computer program code, which, when executed on a computer, causes the computer to perform the aforementioned method steps to implement the artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system provided in the above embodiments.

[0094] Example 4:

[0095] The present invention also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned related steps to realize the artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system provided in the above embodiments.

[0096] In this embodiment, the device, computer-readable storage medium, computer program product, or chip are all used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0097] 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.

[0098] 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. An AI-based intelligent alveolar bone recognition and autologous tooth transplant scoring system, characterized in that, The system includes: The implantation area alveolar socket acquisition module is used to acquire the implantation area alveolar socket in the oral cavity three-dimensional model; The edge-blurred voxel acquisition module is used to obtain the gray-level disorder of each voxel in the implantation area alveolar socket based on the disorder of gray-level values ​​carried by the point cloud in the neighboring voxels; and to filter out edge-blurred voxels based on the gray-level disorder. The alveolar socket inner edge acquisition module is used to treat each edge fuzzy voxel as a first-level fuzzy voxel, and to obtain second-level voxels by pre-defined division of each first-level fuzzy voxel. Based on the similarity of the fuzzing change direction of each second-level voxel to the first-level fuzzy voxel, a second-level fuzzy voxel is obtained. Each second-level fuzzy voxel is then pre-defined to obtain third-level voxels, and based on the similarity of the fuzzing change direction of each third-level voxel to the second-level fuzzy voxel, a third-level fuzzy voxel is obtained. This process continues until each n-level fuzzy voxel has a stable performance consistent with its n-1-level fuzzy voxel, at which point the division stops. The final fuzzy voxels after the division stops are smoothly connected to the corresponding parts as the inner edge of the alveolar socket in the implant area. The scoring module is used to obtain the degree of fit between the autogenous tooth and the alveolar bone based on the matching of the autogenous tooth and the inner edge of the alveolar socket; and to score the autogenous tooth transplant based on the degree of fit. The method for obtaining the second-level fuzzy voxels is as follows: For any first-level fuzzy voxel, the first-level fuzzy voxel is pre-divided into multiple sub-voxels, all of which are used as second-level voxels of the first-level fuzzy voxel. Based on the difference in the direction of grayscale disorder between the first-level fuzzy voxel and its adjacent first-level fuzzy voxels, the direction of fuzziness change of the first-level fuzzy voxel is obtained. For any second-level voxel of the first-level fuzzy voxel, the result of negatively correlated and normalized the difference between the fuzzy change direction of the second-level voxel and the first-level fuzzy voxel is taken as the fuzziness degree of the second-level voxel. When the degree of ambiguity is greater than the preset ambiguity threshold, the second-level voxel is a second-level fuzzy voxel; The method for obtaining the direction of the fuzzy change is as follows: Based on the second derivative of the gray-level disorder degree between the first-level fuzzy voxel and other first-level fuzzy voxels in its first preset neighborhood as a function of spatial location, the growth curvature of the first-level fuzzy voxel is obtained through the Hessian matrix. All adjacent first-level fuzzy voxels of the first-level fuzzy voxel are used as reference voxels of the first-level fuzzy voxel. For any reference voxel, the center distance between the first-level fuzzy voxel and the reference voxel is taken as the first distance; The magnitude of the difference in growth curvature between the first-level fuzzy voxel and the reference voxel is taken as the first value; The arctangent of the first value and the ratio of the first distance is used as the reference direction of change between the first-level fuzzy voxel and the reference voxel. The average of the reference change directions between the first-level fuzzy voxel and all its reference voxels is taken as the fuzzy change direction of the first-level fuzzy voxel. In the process of obtaining the fuzzy change direction of the secondary voxel, the voxels and reference voxels in the first preset neighborhood of the secondary voxel are all other secondary voxels of the primary fuzzy voxel. The method for obtaining the third-level fuzzy voxels is as follows: For any second-level fuzzy voxel, the second-level fuzzy voxel is pre-divided into multiple sub-voxels, all of which are used as third-level voxels of the second-level fuzzy voxel. For any tertiary voxel of the second-level fuzzy voxel, the result of negatively correlated and normalized the difference between the fuzzy change direction of the tertiary voxel and the second-level fuzzy voxel is taken as the fuzziness degree of the tertiary voxel; wherein, in the process of obtaining the fuzzy change direction of the tertiary voxel, the voxels and reference voxels in the first preset neighborhood of the tertiary voxel are all other tertiary voxels of the second-level fuzzy voxel. When the degree of ambiguity is greater than the preset ambiguity threshold, the third-level voxel is a third-level fuzzy voxel.

2. The artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system as described in claim 1, characterized in that, The method for obtaining the grayscale disorder level is as follows: For any voxel in the alveolar socket of the implantation area, the mean value of the gray value carried by the point cloud in the voxel is used as the gray feature factor of the voxel. The difference between the maximum and minimum gray-scale feature factors in the voxel and its neighboring voxels is obtained and used as the first fluctuation analysis value. The mean of the gray-scale feature factors of this voxel and its neighboring voxels is used as the reference gray-scale feature factor. The mean difference between the gray-scale feature factor of the voxel and each of its neighboring voxels and the reference gray-scale feature factor is used as the second fluctuation analysis value. The normalized product of the first and second fluctuation analysis values ​​is used as the degree of grayscale disorder of the voxel.

3. The artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system as described in claim 2, characterized in that, The method for obtaining the edge-blurred voxels is as follows: When the grayscale disorder level is greater than the preset grayscale disorder level threshold, the corresponding voxel is used as the edge blur voxel.

4. The artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system as described in claim 1, characterized in that, The method for stopping the partitioning is as follows: The stability of the i-th (n-1)-th level fuzzy voxel is obtained based on the difference in grayscale disorder between the i-th (n-1)-th level fuzzy voxel and its adjacent (n-1)-th level fuzzy voxels, as well as the difference in the direction of fuzzy change; where, ; All n-level fuzzy voxels divided by the i-th (n-1)th level fuzzy voxel are taken as target voxels; For any target voxel, the division of the target voxel stops when the stability of the target voxel is equal to the stability of the i-th (n-1)th level fuzzy voxel.

5. The artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system as described in claim 4, characterized in that, The method for obtaining the stability level is as follows: The (n-1)th level fuzzy voxel whose gray-level disorder is closest to that of the (i-1)th level fuzzy voxel is taken as the stable analysis voxel of the (i-1)th level fuzzy voxel. The difference in grayscale disorder between the i-th (n-1)th level fuzzy voxel and its stable analysis voxel is taken as the first difference. The difference in the direction of fuzzy change between the i-th (n-1)th level fuzzy voxel and its stable analysis voxel is taken as the second difference. The sum of the first and second differences, followed by negative correlation and normalization, is used as the stability of the i-th (n-1)th level fuzzy voxel.

6. The artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system as described in claim 1, characterized in that, The method for obtaining the degree of adaptation is as follows: Matching feature points in the inner edge of autogenous teeth and alveolar sockets are obtained using a key point detection algorithm; Identify the structural centerline of the autogenous tooth and the alveolar socket in the implantation area, and perform rigid registration of matching feature points in the inner edge of the autogenous tooth and the alveolar socket by translation and rotation to obtain matching point pairs; The mean of the distances between all matching points is negatively correlated and normalized to determine the degree of fit between the autogenous tooth and the alveolar bone.

7. The artificial intelligence-based alveolar bone intelligent recognition and autologous tooth transplantation scoring system as described in claim 1, characterized in that, The method for scoring autologous tooth transplantation based on fit is as follows: When the degree of fit is greater than or equal to the first preset degree of fit threshold, there is a compression situation in autologous tooth transplantation; When the degree of fit is greater than or equal to the second preset degree of fit threshold and less than the first preset degree of fit threshold, autologous tooth transplantation is reasonable. When the degree of fit is less than the second preset degree of fit threshold, autologous tooth transplantation is not reasonable.

Images