An automatic positioning method for upper anterior tooth implants based on alveolar bone segmentation

Abstract

An automatic positioning method of upper anterior tooth implant based on alveolar bone segmentation, comprising the following steps: S1, inputting an original oral CBCT image to a segmentation model to obtain a tooth segmentation result I t and an alveolar bone segmentation result I e ; S2, generating an implant boundary according to the tooth segmentation result I t and the alveolar bone segmentation result I e , to obtain an implant reasonable implantation range of a surface set S p ; S3, generating an implant optimal axial direction according to the implant reasonable implantation range of the surface set S p , and obtaining visual data I p of the implant by slice-by-slice fitting. The present application realizes the alveolar bone segmentation of the upper anterior tooth for the CBCT data, and generates the implant optimal position prediction and the implant optimal image position for the immediate implant, and solves the problems of dependence on subjective experience of doctors, inaccurate implant and time-consuming in the current upper anterior tooth immediate implant surgery.

Description

Technical Field

[0001] The present invention belongs to the field of medical image processing, and in particular relates to an automatic positioning method for anterior maxillary dental implants based on alveolar bone segmentation. Background Art

[0002] Dental health plays a key role in oral function and appearance. In modern oral medical practice, immediate implantation is a cutting-edge oral restoration technology that not only restores the function and aesthetics of missing teeth, but also reduces the waiting time of patients after tooth extraction, thereby providing a more convenient oral rehabilitation solution. It is widely used in implantation of the upper anterior teeth. However, immediate implant surgery is highly dependent on the subjective experience of the doctor, and different doctors directly affect the effect of the implant surgery. Therefore, it is crucial to realize the automatic selection of the optimal position of the implant and the generation of preoperative implant influences to improve the efficiency and accuracy of immediate implant surgery.

[0003] For example, the Chinese invention patent application entitled "A Method for Rapidly Generating Virtual Implant Positions", with publication number CN116269748A and publication date 2023.06.23, discloses a method for rapidly generating virtual implant positions. This method imports oral CBCT and intraoral three-dimensional scanning data, performs three-dimensional reconstruction and data alignment, and uses three-dimensional surface data to generate implant positions. This method requires intraoral three-dimensional scanning data to assist oral CBCT, and both types of data are re-entered in three dimensions. Although the implant position is determined, there is no targeted use of constraints such as the thickness of a single alveolar bone wall to improve the segmentation accuracy of a single tooth position, and the constraint information of adjacent teeth on the implant position is not fully considered. Manual intervention is required, and an end-to-end method cannot be achieved. The problem of data accuracy loss in three-dimensional reconstruction is not solved, which affects the constraints of the near-labial and near-lingual bone walls of the alveolar bone, and affects the accuracy of the implant position and the efficiency of immediate implantation.

[0004] Another example is a Chinese invention patent application entitled "An oral implant design method, system and storage medium", with publication number CN116999195A and publication date 2023.11.07, which discloses an oral implant design method, system and storage medium. This method uses the center points of the identified teeth to intercept the crowns to fit the arc parabola of the upper and lower jaw dentition, and then fits the optimal axial position of the implant at the edentulous position based on the central axis of the tooth body of the remaining teeth. Although this method can generate the axial position of the implant, since the axial position of the implant is determined by the tooth body axis of the adjacent teeth and the upper and lower jaw dentition, there is a large error in the determination of the implant axis due to the influence of the adjacent teeth. In addition, the bone wall constraint of the alveolar bone at the edentulous position is not taken into account, resulting in the determination of the axial position of the implant not taking into account the initial stability of the implant, which will cause deviation in the optimal position of the implant and affect the efficiency and accuracy of immediate implantation. Summary of the Invention

[0005] In order to overcome the shortcomings of the existing technology, the present invention provides an automatic positioning method for maxillary anterior dental implants based on alveolar bone segmentation, which realizes the alveolar bone segmentation of the maxillary anterior teeth position based on CBCT data and predicts the optimal position of the implant for immediate implantation and generates the optimal imaging position of the implant, thereby solving the problems of current maxillary anterior dental immediate implant surgery that relies on the doctor's subjective experience, is inaccurate in implantation and is time-consuming.

[0006] The technical solution adopted by the present invention to solve its technical problem is:

[0007] A method for automatically positioning an upper anterior dental implant based on alveolar bone segmentation comprises the following steps:

[0008] S1. Input the original oral CBCT image into the segmentation model to obtain the tooth segmentation result I t and alveolar bone segmentation results I e ;

[0009] S2. According to the tooth segmentation result I t and alveolar bone segmentation results I e , generate the implant boundary and obtain the surface set S of the reasonable implant range p ;

[0010] S3, surface set S according to reasonable implant range p , generate the best axial direction of the implant by fitting each slice, and obtain the visualization data of the implant I p .

[0011] Furthermore, in S1, the segmentation model includes a feature encoding network and a feature decoding network. The feature encoding network includes five convolution blocks and downsampling operations. The feature decoding network includes four convolution blocks, upsampling operations and jump connections. The original oral CBCT image is input into the segmentation model to obtain the tooth segmentation result I t and alveolar bone segmentation results I e .

[0012] Preferably, the process of S1 is as follows:

[0013] Step 1.1: Input the original oral CBCT image into the convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map F1, whose dimension size is

[0014] Step 1.2: Perform maximum pooling on the feature map F1 and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map F2, whose dimension size is

[0015] Step 1.3, perform maximum pooling on the feature map F2 and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map F3, whose dimension size is

[0016] Step 1.4: Perform maximum pooling on the feature map F3 and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map F4, whose dimension size is

[0017] Step 1.5: Perform maximum pooling on the feature map F4 and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map F5, whose dimension size is

[0018] Step 1.6: Upsample the feature map F5, then concatenate it with the feature map F4 at the channel level and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the feature map F. up1 , its dimension is consistent with F4;

[0019] Step 1.7: The feature map F up1 Up-sample, then concatenate with the feature map F3 at the channel level and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the feature map F up2 , its dimension is consistent with F3;

[0020] Step 1.8, the feature map F up2Up-sample, then concatenate with the feature map F2 at the channel level and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the feature map F up3 , its dimension is consistent with F2;

[0021] Step 1.9: transform the feature map F up3 Up-sample, then concatenate with the feature map F1 at the channel level and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the feature map F up4 , its dimension size is consistent with F1;

[0022] Step 1.10, the feature map F up4 Input to the convolution block with a convolution kernel size of 1×1×1 to obtain the tooth segmentation result I t and alveolar bone segmentation results I e .

[0023] Furthermore, in said S2, according to the tooth segmentation result I t and alveolar bone segmentation results I e , generate the implant boundary and obtain the surface set S of the reasonable implant range p , the process is as follows:

[0024] Step 2.1, tooth segmentation result I t and alveolar bone segmentation results I e According to the different values ​​of the elements in the array, it is divided into the left adjacent tooth point set D l , right adjacent tooth point set D r , tooth point set D t and alveolar point set D e ;

[0025] Step 2.2, calculate the left adjacent tooth point set D l , right adjacent tooth point set D r and alveolar point set D e The normal vectors of these normal vector sets are obtained from their respective point sets, including the normal vectors of all points in each point set. First, a scalar field function f(x, y, z) is defined for each point set to calculate the normal vector of each point. The normal vector is calculated by solving each point p i The gradient of the function f The calculation formula is:

[0026]

[0027] Among them, p i Is a point in the point set, belonging to point set D l 、D r and D e . Left adjacent tooth point set D l , right adjacent tooth point set Dr and alveolar point set D e The normal vector sets are the left adjacent tooth normal vector set v l and the right adjacent tooth normal vector set v r And the alveolar bone normal vector set v e , each set contains the normal vectors of all points in the corresponding point set;

[0028] Step 2.3, according to the left adjacent tooth point set D l and the right adjacent tooth point set D r Extract the outer surface point set D of the left adjacent tooth lw , right adjacent tooth outer surface point set D rw and the highest point p of the left adjacent tooth l and the highest point p of the right adjacent tooth r , the left adjacent tooth outer surface point set D lw and the right adjacent tooth outer surface point set D rw According to the highest point p of the left adjacent tooth l and the highest point p of the right adjacent tooth r Divided into the right outer surface point set D of the left adjacent tooth lrw and the left outer surface point set D of the right adjacent tooth rlw ;

[0029] Step 2.4: Calculate the alveolar bone point set D e The major axis vector v el The long axis vector is obtained by calculating each point P from the alveolar point set i To the geometric centroid of the point set The long axis vector reflects the main extension direction of the alveolar bone in space and is the representative vector of the entire point set. The formula is as follows:

[0030]

[0031] Step 2.5, tooth point set D t The highest point p on the z-axis in the three-dimensional array t As the highest point of the tooth hole, create a section B parallel to the y-axis e Contains the long axis vector v of the alveolar bone el and the highest point of the tooth hole p t From section B e Set the alveolar bone point D e Divided into mesial alveolar point cluster D ef and the mesiolingual alveolar bone point set D eb ;

[0032] Step 2.6: Traverse the labial alveolar bone point set D using the y-axis slice ef and the mesiolingual alveolar bone point set D eb, connect the z-axis lower endpoints of the two point sets in the same slice and count them as the lower boundary points, and all points on each slice connection line are recorded as the lower boundary point set D d ;

[0033] Step 2.7, according to the alveolar bone point set D e Extract the alveolar bone outer surface point set D ew , the alveolar bone outer surface point set D ew The point set D ef and mesiolingual alveolar bone point cluster D eb Take the union and get the point set D of the outer surface of the alveolar bone on the near labial side efw and the point set D on the outer surface of the alveolar bone on the near lingual side ebw ;

[0034] Step 2.8, according to the right outer surface point set D of the left adjacent tooth lrw , the point set D of the left outer surface of the right adjacent tooth rlw and the point set D on the outer surface of the alveolar bone on the mesial labial side efw The corresponding left adjacent tooth normal vector set v l , right adjacent tooth normal vector set v r 、Alveolar bone normal vector set v e , expand the point set by a certain distance in the direction of the normal vector, and obtain the expanded point set D on the right side of the left adjacent tooth lrwc , point set D on the left side of the right adjacent tooth after expansion rlwc and the point set D on the outer surface of the mesial alveolar bone after expansion efwc , the coordinates of each point after movement P i The calculation formula is as follows:

[0035] P i ′=(x i +d·v xi ,y i +d·v yi ,z i +d·v zi )

[0036] Among them, P i =(x i ,y i ,z i ) represents the coordinates of the points in the point set, belonging to the point set D lrw 、D rlw and D efw , Indicates the normal vector corresponding to the point, which belongs to the normal vector set v l 、v r and v e , d is the distance moved;

[0037] Step 2.9, point set Dlrwc 、D rlwc 、D efwc 、D eb and D d The points are combined into an upper-open cuboid, and the space enclosed by the point set is the reasonable implant range of the implant, which is recorded as the surface set S of the reasonable implant range. p .

[0038] In said S3, according to the reasonable implant range surface set S p , the optimal axial direction of the implant is generated by slice-by-slice fitting, and the visualization data of the implant is obtained. The process is as follows:

[0039] Step 3.1: Traverse the expanded mesial alveolar bone outer surface point set D by slicing in the y-axis direction efwc and the point set D on the outer surface of the alveolar bone on the near lingual side ebw , respectively, use the least squares method to perform straight line fitting on the points of the current slice of the point set, and calculate the slope of the straight line corresponding to each slice, which is recorded as the slope k of the outer surface of the proximal labial side on each slice. ef , intercept d ef and the slope k of the outer surface of the mesiolingual side eb , intercept d eb , the formula is as follows:

[0040]

[0041] Among them, x i and z i represent the horizontal (x) and vertical (z) coordinates of each point in the y-axis slice, respectively, and n is the number of points in the slice;

[0042] Step 3.2: According to the slope k of the outer surface of the proximal labial side ef , intercept d ef and the slope k of the outer surface of the mesiolingual side eb , intercept d eb , get the slope k of the centered straight line on the y-axis slice e is (k ef +k eb ) / 2, intercept d e for (d ef +d eb ) / 2, record the points on the y-axis slice that the center line passes through, and record them as the y-plane axial point set D ep ;

[0043] Step 3.3: Traverse the point set D of the right outer surface of the left adjacent tooth after expansion by slicing in the y-axis direction lrwc , point set D on the left side of the right adjacent tooth after expansion rlwc, use the least squares method to perform straight line fitting on the points of the current slice of the point set, and calculate the slope of the straight line corresponding to each slice, which is recorded as the slope k of the straight line on the right outer surface of the left adjacent tooth on each slice tl , intercept d tl and the slope k of the straight line on the left side of the right adjacent tooth tr , intercept d tr , the formula is as follows:

[0044]

[0045] Among them, y i and z i denote the horizontal (x) and vertical (z) coordinates of each point in the x-axis slice, respectively, and n is the number of points in the slice;

[0046] Step 3.4: According to the slope k of the straight line on the right side of the left adjacent tooth tl , intercept d tl and the slope k of the straight line on the left side of the right adjacent tooth tr , intercept d tr , get the slope k of the centered straight line on the x-axis slice t is (k tl +k tr ) / 2, intercept d t for (d tl +d tr ) / 2, record the points that the center line on the x-axis slice passes through, and record them as the x-plane axial point set D tp ;

[0047] Step 3.5: Orient the y-plane axis to the point set D ep and the x-plane axial point set D tp Take the intersection to get the implant axial point set D p , project all points in the point set onto the y-axis plane, ignoring the y coordinates of all points, according to the x i and z i Coordinate fitting of a two-dimensional straight line to find the slope k xz , intercept d xz Similarly, project all points onto the x-axis plane, according to the y i and z i Coordinate fitting of a two-dimensional straight line to find the slope k yz , intercept d yz Finally, a straight line in three-dimensional space is obtained, which is recorded as the implant axial straight line L. The formula is as follows:

[0048] z=d xz +k xz x

[0049] z=d yz +kyz y

[0050] Step 3.6: Based on the implant axial line L and the lower boundary point set D d Calculate the intersection of the implant axis and the lower boundary, and record it as the implant starting point p ps The implant length d is extended from the starting point along the implant axis. p , get the implant endpoint p pe ;

[0051] Step 3.7, according to the implant starting point p ps and implant endpoint p pe As well as the implant width r, the points between the two points whose distance from the implant axial straight line L is less than the implant width r are marked as implants, and the visualization data I of the implant is obtained. p .

[0052] In this method, a segmentation model is first used to obtain the segmentation results of the teeth and alveolar bone. This segmentation automatically calculates the appropriate implant placement range. Based on this reasonable implant range, the optimal axial position of the implant is then selected slice by slice in two dimensions, ultimately generating visualization data of the optimal implant position.

[0053] The main benefits of this invention are: it enables alveolar bone segmentation of the upper anterior teeth based on CBCT data, predicts the optimal implant position for immediate implant placement, and generates the optimal implant image position. This solves the current problem of immediate implant placement in the upper anterior teeth, which relies on the surgeon's subjective experience, resulting in inaccurate and time-consuming implant placement. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a flow chart of an automatic positioning method for anterior maxillary dental implants based on alveolar bone segmentation;

[0055] Figure 2 This is a technical framework diagram of an automatic positioning method for anterior maxillary dental implants based on alveolar bone segmentation;

[0056] Figure 3 This is an example of the automatic positioning method for maxillary anterior dental implants based on alveolar bone segmentation. DETAILED DESCRIPTION

[0057] The present invention will be further described below with reference to the accompanying drawings.

[0058] Reference Figures 1 to 3 , an automatic positioning method for anterior maxillary dental implants based on alveolar bone segmentation, comprising the following steps:

[0059] S1. Input the original oral CBCT image into the segmentation model to obtain the tooth segmentation result I t and alveolar bone segmentation results Ie .

[0060] The segmentation model described in this application includes a feature encoding network and a feature decoding network. The feature encoding network includes five convolution blocks and downsampling operations. The feature decoding network includes four convolution blocks, upsampling operations and jump connections. The original oral CBCT image is input into the segmentation model to obtain the tooth segmentation result I t and alveolar bone segmentation results I e ; The process of S1 is as follows:

[0061] Step 1.1: Input the original oral CBCT image into the convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map F1, whose dimension size is

[0062] Step 1.2: Perform maximum pooling on the feature map F1 and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map F2, whose dimension size is

[0063] Step 1.3, perform maximum pooling on the feature map F2 and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map F3, whose dimension size is

[0064] Step 1.4: Perform maximum pooling on the feature map F3 and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map F4, whose dimension size is

[0065] Step 1.5: Perform maximum pooling on the feature map F4 and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map F5, whose dimension size is

[0066] Step 1.6: Upsample the feature map F5, then concatenate it with the feature map F4 at the channel level and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the feature map F. up1 , its dimension is consistent with F4;

[0067] Step 1.7: The feature map F up1 Up-sample, then concatenate with the feature map F3 at the channel level and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the feature map F up2 , its dimension is consistent with F3;

[0068] Step 1.8, the feature map F up2Up-sample, then concatenate with the feature map F2 at the channel level and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the feature map F up3 , its dimension is consistent with F2;

[0069] Step 1.9: transform the feature map F up3 Up-sample, then concatenate with the feature map F1 at the channel level and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the feature map F up4 , its dimension size is consistent with F1;

[0070] Step 1.10, the feature map F up4 Input to the convolution block with a convolution kernel size of 1×1×1 to obtain the tooth segmentation result I t and alveolar bone segmentation results I e .

[0071] Among them, the convolution block includes not only the convolution of the above-mentioned convolution kernel size, but also a normalization and a ReLU activation function.

[0072] S2. According to the tooth segmentation result I t and alveolar bone segmentation results I e , generate the implant boundary and obtain the surface set S of the reasonable implant range p .

[0073] like Figure 3 As shown, the present application describes the tooth segmentation result I t and alveolar bone segmentation results I e , generate the implant boundary and obtain the surface set S of the reasonable implant range p , the process is as follows:

[0074] Step 2.1, tooth segmentation result I t and alveolar bone segmentation results I e According to the different values ​​of the elements in the array, it is divided into the left adjacent tooth point set D l , right adjacent tooth point set D r , tooth point set D t and alveolar point set D e ;

[0075] Step 2.2, calculate the left adjacent tooth point set D l , right adjacent tooth point set D r and alveolar point set D e The normal vectors of these normal vector sets are obtained from their respective point sets, including the normal vectors of all points in each point set. First, a scalar field function f(x, y, z) is defined for each point set to calculate the normal vector of each point. The normal vector is calculated by solving each point p iThe gradient of the function f The calculation formula is:

[0076]

[0077] Among them, p i Is a point in the point set, belonging to point set D l 、D r and D e . Left adjacent tooth point set D l , right adjacent tooth point set D r and alveolar point set D e The normal vector sets are the left adjacent tooth normal vector set v l and the right adjacent tooth normal vector set v r And the alveolar bone normal vector set v e , each set contains the normal vectors of all points in the corresponding point set;

[0078] Step 2.3, according to the left adjacent tooth point set D l and the right adjacent tooth point set D r Extract the outer surface point set D of the left adjacent tooth lw , right adjacent tooth outer surface point set D rw and the highest point p of the left adjacent tooth l and the highest point p of the right adjacent tooth r , the left adjacent tooth outer surface point set D lw and the right adjacent tooth outer surface point set D rw According to the highest point p of the left adjacent tooth l and the highest point p of the right adjacent tooth r Divided into the right outer surface point set D of the left adjacent tooth lrw and the left outer surface point set D of the right adjacent tooth rlw ;

[0079] Step 2.4: Calculate the alveolar bone point set D e The major axis vector v el The long axis vector is obtained by calculating each point P from the alveolar point set i To the geometric centroid of the point set The long axis vector reflects the main extension direction of the alveolar bone in space and is the representative vector of the entire point set. The formula is as follows:

[0080]

[0081] Step 2.5, tooth point set D t The highest point p on the z-axis in the three-dimensional array t As the highest point of the tooth hole, create a section B parallel to the y-axis e Contains the long axis vector v of the alveolar bone el and the highest point of the tooth hole p t, from section B e Set the alveolar bone point D e Divided into mesial alveolar point cluster D ef and the mesiolingual alveolar bone point set D eb ;

[0082] Step 2.6: Traverse the labial alveolar bone point set D using the y-axis slice ef and the mesiolingual alveolar bone point set D eb , connect the z-axis lower endpoints of the two point sets in the same slice and count them as the lower boundary points, and all points on each slice connection line are recorded as the lower boundary point set D d ;

[0083] Step 2.7, according to the alveolar bone point set D e Extract the alveolar bone outer surface point set D ew , the alveolar bone outer surface point set D ew The point set D ef and mesiolingual alveolar bone point cluster D eb Take the union and get the point set D of the outer surface of the alveolar bone on the near labial side efw and the point set D on the outer surface of the alveolar bone on the near lingual side ebw ;

[0084] Step 2.8, according to the right outer surface point set D of the left adjacent tooth lrw , the point set D of the left outer surface of the right adjacent tooth rlw and the point set D on the outer surface of the alveolar bone on the mesial labial side efw The corresponding left adjacent tooth normal vector set v l , right adjacent tooth normal vector set v r 、Alveolar bone normal vector set v e , expand the point set by a certain distance in the direction of the normal vector, and obtain the expanded point set D on the right side of the left adjacent tooth lrwc , point set D on the left side of the right adjacent tooth after expansion lrwc and the point set D on the outer surface of the mesial alveolar bone after expansion efwc , the coordinates of each point after movement P i The calculation formula is as follows:

[0085] P i ′=(x i +d·v xi ,y i +d·v yi ,z i +d·v zi )

[0086] Among them, P i =(x i ,y i ,z i) represents the coordinates of the points in the point set, belonging to the point set D lrw 、D rlw and D efw , Indicates the normal vector corresponding to the point, which belongs to the normal vector set v l 、v r and v e , d is the distance moved;

[0087] Step 2.9, point set D lrwc 、D rlwc 、D efwc 、D eb and D d The points are combined into an upper-open cuboid, and the space enclosed by the point set is the reasonable implant range of the implant, which is recorded as the surface set S of the reasonable implant range. p ;

[0088] S3, surface set S according to reasonable implant range p , generate the best axial direction of the implant by fitting each slice, and obtain the visualization data of the implant I p .

[0089] The surface set S according to the reasonable implant range of the implant described in this application p , the optimal axial direction of the implant is generated by slice-by-slice fitting, and the visualization data of the implant is obtained. The process is as follows:

[0090] Step 3.1: Traverse the expanded mesial alveolar bone outer surface point set D by slicing in the y-axis direction efwc and the point set D on the outer surface of the alveolar bone on the near lingual side ebw , respectively, use the least squares method to perform straight line fitting on the points of the current slice of the point set, and calculate the slope of the straight line corresponding to each slice, which is recorded as the slope k of the outer surface of the proximal labial side on each slice. ef , intercept d ef and the slope k of the outer surface of the mesiolingual side eb , intercept d eb , the formula is as follows:

[0091]

[0092] Among them, x i and z i represent the horizontal (x) and vertical (z) coordinates of each point in the y-axis slice, respectively, and n is the number of points in the slice;

[0093] Step 3.2: According to the slope k of the outer surface of the proximal labial side ef , intercept d ef and the slope k of the outer surface of the mesiolingual side eb , intercept d eb, get the slope k of the centered straight line on the y-axis slice e is (k ef +k eb ) / 2, intercept d e for (d ef +d eb ) / 2, record the points on the y-axis slice that the center line passes through, and record them as the y-plane axial point set D ep ;

[0094] Step 3.3: Traverse the point set D of the right outer surface of the left adjacent tooth after expansion by slicing in the y-axis direction lrwc , point set D on the left side of the right adjacent tooth after expansion rlwc , use the least squares method to perform straight line fitting on the points of the current slice of the point set, and calculate the slope of the straight line corresponding to each slice, which is recorded as the slope k of the straight line on the right outer surface of the left adjacent tooth on each slice tl , intercept d tl and the slope k of the straight line on the left side of the right adjacent tooth tr , intercept d tr , the formula is as follows:

[0095]

[0096] Among them, y i and z i denote the horizontal (x) and vertical (z) coordinates of each point in the x-axis slice, respectively, and n is the number of points in the slice;

[0097] Step 3.4: According to the slope k of the straight line on the right side of the left adjacent tooth tl , intercept d tl and the slope k of the straight line on the left side of the right adjacent tooth tr , intercept d tr , get the slope k of the centered straight line on the x-axis slice t is (k tl +k tr ) / 2, intercept d t for (d tl +d tr ) / 2, record the points that the center line on the x-axis slice passes through, and record them as the x-plane axial point set D tp ;

[0098] Step 3.5: Orient the y-plane axis to the point set D ep and the x-plane axial point set D tp Take the intersection to get the implant axial point set D p , project all points in the point set onto the y-axis plane, ignoring the y coordinates of all points, according to the x i and z i Coordinate fitting of a two-dimensional straight line to find the slope kxz , intercept d xz Similarly, project all points onto the x-axis plane, according to the y i and z i Coordinate fitting of a two-dimensional straight line to find the slope k yz , intercept d yz Finally, a straight line in three-dimensional space is obtained, which is recorded as the implant axial straight line L. The formula is as follows:

[0099] z=d xz +k xz x

[0100] z=d yz +k yz y

[0101] Step 3.6: Based on the implant axial line L and the lower boundary point set D d Calculate the intersection of the implant axis and the lower boundary, and record it as the implant starting point p ps , extending the implant length d from the starting point along the implant axis p , get the implant endpoint p pe ;

[0102] Step 3.7, according to the implant starting point p ps and implant endpoint p pe As well as the implant width r, the points between the two points whose distance from the implant axial straight line L is less than the implant width r are marked as implants, and the visualization data I of the implant is obtained. p .

[0103] It should be noted that the example of the automatic positioning method of the upper anterior dental implant based on alveolar bone segmentation of the present application is shown in FIG. Figure 3 shown.

[0104] It should be noted that in this application, C is the number of channels, D is the depth of the image, H is the height of the image, W is the width of the image, and the subscripts of the letters represent the serial numbers, which are used to distinguish the dimensions of different feature maps.

[0105] This application uses a segmentation model to obtain the segmentation results of teeth and alveolar bone. Using data from adjacent teeth and alveolar bone at the edentulous site, the application adaptively determines the optimal implant placement area. Based on this reasonable implant range, the optimal axial position of the implant is selected slice by slice in two dimensions, ultimately generating visualization data of the optimal implant position.

[0106] The embodiments of this specification are merely examples of implementations of the invention and are provided for illustrative purposes only. The scope of protection of the present invention should not be considered limited to the specific embodiments described in these embodiments. The scope of protection of the present invention also extends to equivalent technical means that can be conceived by a person of ordinary skill in the art based on the invention.

Claims

1. A method for automatic positioning of anterior dental implants based on alveolar bone segmentation, characterized in that: The method comprises the following steps: S1. Input the original oral CBCT image into the segmentation model to obtain the tooth segmentation result and alveolar bone segmentation results ; S2. According to the tooth segmentation results and alveolar bone segmentation results , generate the implant boundary and obtain the surface set of the reasonable implant range , the process is as follows: Step 2.1: Tooth segmentation results and alveolar bone segmentation results According to the different values ​​of the elements in the array, it is divided into the left adjacent tooth point set , right adjacent tooth point set , tooth point set and alveolar point set ; Step 2.2, calculate the left adjacent tooth point set , right adjacent tooth point set and alveolar point set The normal vectors of , which are obtained from their respective point sets, including the normal vectors of all points in each point set; Step 2.3, according to the left adjacent tooth point set and right adjacent tooth point set Extract the outer surface point set of the left adjacent tooth , right adjacent tooth outer surface point set and the highest point of the left adjacent tooth and the highest point of the right adjacent tooth , the outer surface point set of the left adjacent tooth and the right adjacent tooth outer surface point set Based on the highest point of the left adjacent tooth and the highest point of the right adjacent tooth Divided into the right outer surface point set of the left adjacent tooth and the left outer surface point set of the right adjacent tooth ; Step 2.4: Alveolar bone point collection Mesial alveolar point cluster Point set with the mesiolingual alveolar bone ; Step 2.5: Traverse the alveolar bone point set on the mesial labial side using the y-axis slice Point set with the mesiolingual alveolar bone , connect the z-axis lower endpoints of the two point sets in the same slice and count them as the lower boundary points, and all points on each slice connection line are recorded as the lower boundary point set ; Step 2.6, according to the alveolar bone point set Extract the point set of the alveolar bone outer surface , the outer surface points of the alveolar bone are gathered The alveolar bone point clusters on the mesial labial side and mesiolingual alveolar point cluster Take the union and get the point set of the outer surface of the alveolar bone on the near labial side and the point set on the outer surface of the alveolar bone on the near lingual side ; Step 2.7, based on the right outer surface point set of the left adjacent tooth , the point set of the left outer surface of the right adjacent tooth and the point set of the outer surface of the alveolar bone on the mesial labial side The corresponding left adjacent tooth normal vector set , right adjacent tooth normal vector set , alveolar bone normal vector set , expand the point set by a certain distance in the direction of the normal vector, and obtain the point set of the right outer surface of the left adjacent tooth after expansion , Point set of the left outer surface of the right adjacent tooth after expansion and the point set of the outer surface of the mesial alveolar bone after expansion ; Step 2.8, point set 、 、 、 The points are combined into an upper-open cuboid, and the space enclosed by the point set is the reasonable implant range of the implant, which is recorded as the surface set of the reasonable implant range ; S3. Surface set based on reasonable implant range , generate the best axial direction of the implant by fitting each slice, and obtain the visual data of the implant .

2. The method for automatic positioning of anterior dental implants based on alveolar bone segmentation according to claim 1, characterized in that: In S1, the segmentation model includes a feature encoding network and a feature decoding network. The feature encoding network includes five convolution blocks and downsampling operations, and the feature decoding network includes four convolution blocks, upsampling operations and jump connections. The original oral CBCT image is input into the segmentation model to obtain the tooth segmentation result. and alveolar bone segmentation results .

3. The method for automatic positioning of anterior dental implants based on alveolar bone segmentation according to claim 2, characterized in that: The process of S1 is as follows: Step 1.1: Input the original oral CBCT image into the convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map , whose dimension is ; Step 1.2: Feature map Perform maximum pooling and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map , whose dimension is ; Step 1.3: Feature map Perform maximum pooling and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map , whose dimension is ; Step 1.4: Feature map Perform maximum pooling and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map , whose dimension is ; Step 1.5: Feature map Perform maximum pooling and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the output feature map , whose dimension is ; Step 1.6: Feature map Up-sample and then compare with the feature map Perform channel-level splicing and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the feature map , whose dimensions are the same as consistent; Step 1.7: Feature map Up-sample and then compare with the feature map Perform channel-level splicing and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the feature map , whose dimensions are the same as consistent; Step 1.8: Feature map Up-sample and then compare with the feature map Perform channel-level splicing and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the feature map , whose dimensions are the same as consistent; Step 1.9: Feature map Up-sample and then compare with the feature map Perform channel-level splicing and input it into a convolution block with a convolution kernel size of 3×3×3 to obtain the feature map , whose dimensions are the same as consistent; Step 1.10, feature map Input to the convolution block with a convolution kernel size of 1×1×1 to obtain the segmentation result of the teeth and alveolar bone segmentation results .

4. The method for automatically positioning an anterior dental implant based on alveolar bone segmentation according to any one of claims 1 to 3, characterized in that: In step 2.2, a scalar field function is first defined for each point set. , used to calculate the normal vector of each point. The normal vector is calculated by solving each point In the function The gradient in , the calculation formula is: in, It is a point in the point set, belonging to the point set 、 and , the left adjacent tooth point set , right adjacent tooth point set and alveolar point set The normal vector sets are the left adjacent tooth normal vector sets and right adjacent tooth normal vector set and the alveolar bone normal vector set , each set contains the normal vectors of all points in the corresponding point set; In step 2.4, the alveolar bone point set is calculated The major axis vector The long axis vector is obtained by calculating each point in the alveolar point set To the geometric centroid of the point set The long axis vector reflects the main extension direction of the alveolar bone in space and is the representative vector of the entire point set. The formula is as follows: Set the tooth points The highest point on the z-axis in a three-dimensional array As the highest point of the tooth hole, create a section parallel to the y-axis Contains the long axis vector of the alveolar bone and the highest point of the tooth hole , by the cross section Alveolar bone point collection Mesial alveolar point cluster Point set with the mesiolingual alveolar bone ; In step 2.7, the coordinates of each point after movement The calculation formula is as follows: in, Represents the coordinates of points in a point set, belonging to the point set 、 and , Indicates the normal vector corresponding to the point, which belongs to the normal vector set 、 and , is the distance moved.

5. The method for automatic positioning of anterior dental implants based on alveolar bone segmentation according to any one of claims 1 to 3, characterized in that: In said S3, according to the reasonable implant range surface set , the optimal axial direction of the implant is generated by slice-by-slice fitting, and the visualization data of the implant is obtained. The process is as follows: Step 3.1: Traverse the expanded mesial alveolar bone outer surface point set by slicing in the y-axis direction and the point set on the outer surface of the alveolar bone on the near lingual side , respectively, use the least squares method to perform straight line fitting on the points of the current slice of the point set, and calculate the slope of the straight line corresponding to each slice, which is recorded as the slope of the straight line on the outer surface of the proximal labial side of each slice. ,intercept and the slope of the straight line on the mesiolingual outer surface ,intercept , the formula is as follows: in, and Respectively expressed in The horizontal (x) and vertical (z) coordinates of each point in the axial slice, is the number of points in the slice; Step 3.2: According to the slope of the straight line on the outer surface of the proximal labial side ,intercept and the slope of the straight line on the mesiolingual outer surface ,intercept , get the slope of the centered straight line on the y-axis slice for( + ) / 2, intercept for( + ) / 2, record the points that the center line on the y-axis slice passes through, and record them as the y-plane axial point set ; Step 3.3: Traverse the right outer surface point set of the left adjacent tooth after expansion by slicing in the y-axis direction , Point set of the left outer surface of the right adjacent tooth after expansion , respectively, use the least squares method to perform straight line fitting on the points of the current slice of the point set, and calculate the slope of the straight line corresponding to each slice, which is recorded as the slope of the straight line on the right outer surface of the left adjacent tooth on each slice. ,intercept and the slope of the straight line on the left side of the right adjacent tooth ,intercept , the formula is as follows: in, and Respectively expressed in The horizontal (x) and vertical (z) coordinates of each point in the axial slice, is the number of points in the slice; Step 3.4: According to the slope of the straight line on the right side of the left adjacent tooth ,intercept and the slope of the straight line on the left side of the right adjacent tooth ,intercept , get the slope of the centered straight line on the x-axis slice for( + ) / 2, intercept for( + ) / 2, record the points that the center line on the x-axis slice passes through, and record them as the x-plane axial point set ; Step 3.5: Move the y-plane axis to the point set and the x-plane axial point set Take the intersection to get the implant axial point set , project all points in the point set onto Axis plane, ignoring all points Coordinates, according to and Coordinate fitting of a 2D straight line to find the slope ,intercept , similarly, project all points onto Axial plane, according to and Coordinate fitting of a 2D straight line to find the slope ,intercept Finally, a straight line in three-dimensional space is obtained, which is recorded as the implant axial line , the formula is as follows: Step 3.6, according to the implant axial line and the lower boundary point set Calculate the intersection of the implant axis and the lower boundary and record it as the implant starting point , extending the implant length from the starting point along the implant axis , get the implant endpoint ; Step 3.7, according to the implant starting point and implant endpoints and implant width , the distance between the two points to the implant axis Smaller than implant width The points are marked as implants, and the visualization data of implants is obtained. .

Images