Ideal arch line generation method and device based on AI alveolar bone grid segmentation

Abstract

The invention relates to an ideal arch line generation method based on AI alveolar bone grid segmentation. The method comprises the following steps: acquiring an alveolar bone region of interest according to a CT alveolar bone complete model; performing boundary smoothing processing on the alveolar bone region of interest; constructing a coordinate system, and transferring the coordinate system to a standard coordinate system; solving a point set representing the overall skeleton trend and fitting the point set into a B spline curve L1; n points are sampled on the L1 at equal intervals, cross sections are made, cross section intersection line centroids are obtained and arranged in sequence, and the centroids are recorded as L2; connecting a vector by taking the point on the L1 as a starting point and the point on the L2 as an ending point, solving intersection points of a straight line where the vector is located and the grid of the region of interest, arranging the intersection points according to the same sequence as the points on the L1 and the L2, and recording the intersection points as L3; and establishing an ideal dental arch line attached to the alveolar ridge crest according to the L3. According to the invention, when the alveolar arch line passes through each tooth hole, the Spee curvature is more suitable for the growth trend of the real alveolar bone of a patient, the compression force on the tooth root in the correction process is reduced as much as possible, and a safer and more reliable correction scheme is provided for the patient.

Description

Technical Field

[0001] The present invention relates to the technical field of oral computer-aided design, and in particular to a method and device for generating an ideal arch line based on AI alveolar bone mesh segmentation. Background Art

[0002] In the problem of dental scan tooth model alignment, if the personalized ideal arch shape is generated only by relying on tooth feature information, the generation of the ideal arch shape will be adversely affected by the presence of deformed teeth. This is a major challenge generally faced by dental computer-aided software. If the ideal arch shape is adjusted manually by traditional doctors, the operation process is complicated and it is also difficult to maintain the fit and smoothness of the ideal arch shape. Summary of the invention

[0003] The technical problem to be solved by the present invention is to design a method and device for generating a three-dimensional alveolar arch line based on the AI ​​segmentation of the alveolar bone area of ​​the CT alveolar bone model, so as to solve the existing technical problems.

[0004] In order to solve the above technical problems, the present invention provides an ideal archline generation method based on AI alveolar bone mesh segmentation, comprising the following steps:

[0005] Step S1: Acquire the alveolar bone region of interest based on the CT alveolar bone complete model.

[0006] Step S2: performing boundary smoothing processing on the acquired alveolar bone region of interest.

[0007] Step S3: construct a coordinate system for the region of interest after boundary smoothing and transfer it to the standard coordinate system.

[0008] Step S4: Obtain the point set P that represents the overall skeleton trend A , and according to the point set P A Connect the points in a smooth B-spline curve L1 so that it passes through the point set P as much as possible. A The point in .

[0009] Step S5: Sample N points equidistantly on the curve L1, calculate the tangent vector Dir at each point based on the sampled points and their adjacent points, and make N sections with each point as the plane origin and the tangent vector Dir as the plane normal vector. For each section, find its intersection with the grid of the area of ​​interest, and average the section intersections to find its centroid, then arrange all the intersection centroids in order, recorded as L2.

[0010] Step S6: Take the point on L1 as the starting point and the point on L2 as the end point, connect them into a vector ExtendDir, and shoot a straight line Line along the vector to find its intersection with the grid of the area of ​​interest. Finally, arrange all the intersections in the same order as the points on L1 and L2, recorded as L3.

[0011] Step S7: Establish an ideal dental arch line that fits the alveolar ridge top based on L3.

[0012] Furthermore, in step S1, the alveolar bone focus area is obtained based on data annotation and AI grid segmentation algorithm. The alveolar bone focus area only includes the alveolar bone morphology near the tooth cavity, and the palate structure that is too far away from the tooth cavity is removed.

[0013] Furthermore, in step S2, the method for boundary smoothing is: for the segmented alveolar bone area of ​​interest, find its maximum boundary, fit a cutting line with sufficiently high smoothness through the B-spline function, and then use the cutting line to re-cut a segmented area with a sufficiently smooth boundary on the complete alveolar bone model, and replace the original area of ​​interest.

[0014] Furthermore, in step S3, the plane is fitted using the least squares method, and the least squares matrix at this time is:

[0015]

[0016] According to X=(A T A) -1 A T b Solve the plane parameters a0, a1, a2 and use its normal vector as the current X-axis;

[0017] Then sum the normals of all the faces as the current Z axis, and use the center of mass as the origin to construct an orthogonal coordinate system of the current position according to the right-hand rule;

[0018] Finally, the grid of the area of ​​interest is transferred to the standard coordinate system through coordinate system transformation.

[0019] Furthermore, in step S4, the point set P is obtained by skeletonization method. A The specific method is: take the left tail area as the root node, take the right tail area as the terminal leaf node, and obtain the points sorted from left to right to form the point set P A .

[0020] Furthermore, in step S6, the method for obtaining the intersection of the line where the vector ExtendDir is located and the mesh of the area of ​​interest is as follows: for all triangular facets face of the mesh, the plane PL where the facets are located is obtained, and the intersection point InterPt between the line Line and the plane PL is obtained. If InterPt falls inside the face triangle, the intersection point InterPt is returned.

[0021] Furthermore, in step S7, the specific method for generating the dental arch line is as follows: equidistantly sample points on L3, and then project them onto the XOY plane of the standard coordinate system, and fit a symmetrical XOY plane curve with an even-order polynomial through the least squares method to obtain the curve parameter vector and determine the XOY plane even-order polynomial curve; fit an odd-order polynomial in the YOZ space to determine the Spee height corresponding to each point on the plane curve, and finally generate a symmetrical ideal curve with Spee curvature in R^3.

[0022] Furthermore, the following steps are also included between step S6 and step S7: along the extension direction ExtendDir of each point, L3 is extended or contracted by any length, specifically: TargetPoint=L3Point+ExtendLength*ExtendDir, wherein: TargetPoint is the point after extension or contraction, L3Point is the point on L3, and ExtendLength is the extension or contraction length.

[0023] Furthermore, when the arch curve needs to fit the crown, ExtendLength takes the average distance between the crown centroid and its corresponding nearest point on L3; when the arch curve needs to fit the root, ExtendLength takes the average distance between the root point and its corresponding nearest point on L3.

[0024] The present invention also provides an ideal archline generation device based on AI alveolar bone mesh segmentation, comprising:

[0025] at least one processor; and

[0026] at least one memory in communication with the processor;

[0027] The memory stores instructions executable by the processor, and the instructions are executed by the processor so that the device executes the aforementioned ideal archline generation method based on AI alveolar bone mesh segmentation.

[0028] Beneficial effects of the present invention:

[0029] The ideal archline generation method and device based on AI alveolar bone mesh segmentation of the present invention can achieve that when the alveolar archline passes through each tooth cavity, the Spee curvature is more in line with the growth trend of the patient's actual alveolar bone, thereby reducing the pressure on the tooth root during the correction process as much as possible, and providing patients with a safer and more reliable correction plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The specific implementation of the present invention will be further explained below in conjunction with the accompanying drawings.

[0031] Figure 1It is a flow chart of the ideal archline generation method based on AI alveolar bone mesh segmentation of the present invention;

[0032] Figure 2 Schematic diagram of the maxillofacial region of interest for alveolar bone segmentation in the present invention;

[0033] Figure 3 A frontal schematic diagram of the alveolar bone segmentation focus area in the invention;

[0034] Figure 4 Schematic diagram of boundary smoothing and normalization in the invention;

[0035] Figure 5 Schematic diagram of skeletonized splines in the invention;

[0036] Figure 6 Schematic diagram of the centroid of a local cross section in the invention;

[0037] Figure 7 Schematic diagram of the projection curve in the invention;

[0038] Figure 8 Schematic diagram of the final effect of the invention. DETAILED DESCRIPTION

[0039] Example 1

[0040] Combination Figure 1 The ideal archline generation method based on AI alveolar bone mesh segmentation of the present invention comprises the following steps:

[0041] Step S1: Acquire the alveolar bone region of interest based on the CT alveolar bone complete model.

[0042] In this embodiment, preferably, in step S1, the CT alveolar bone complete model maxillofacial Figure 2 As shown in the figure, the alveolar bone focus area is obtained based on data annotation and AI mesh segmentation algorithm ( Figure 2 The alveolar bone focus area only includes the alveolar bone morphology near the tooth cavity, and the palate structure that is too far from the tooth cavity is removed. Figure 3 shown.

[0043] Step S2: performing boundary smoothing processing on the acquired alveolar bone region of interest.

[0044] Preferably, in step S2 of this embodiment, the method for boundary smoothing is: for the segmented alveolar bone area of ​​interest, find its maximum boundary, fit a cutting line with sufficiently high smoothness through the B-spline function, and then use the cutting line to re-cut a segmented area with a sufficiently smooth boundary on the complete alveolar bone model, and replace the original area of ​​interest.

[0045] Step S3: construct a coordinate system for the region of interest after boundary smoothing and transfer it to the standard coordinate system.

[0046] In this embodiment, preferably, in step S3, the plane is fitted using the least square method according to the symmetry of the structure of the region of interest after boundary smoothing. The least square method matrix at this time is:

[0047]

[0048] According to X=(A T A) -1 A T b Solve the plane parameters a0, a1, a2 and use its normal vector as the current X-axis;

[0049] Then sum the normals of all the faces as the current Z axis, and use the center of mass as the origin to construct an orthogonal coordinate system of the current position according to the right-hand rule;

[0050] Finally, the grid of the area of ​​interest is transferred to the standard coordinate system through coordinate system transformation, such as Figure 4 shown.

[0051] Step S4: For the region of interest after boundary smoothing and coordinate normalization, Figure 5 As shown, in this embodiment, the point set P representing the overall skeleton trend is obtained by the skeletonization method. A (like Figure 5 As shown by the red dots in the figure), and according to the point set P A Connect the points in a smooth B-spline curve L1( Figure 5 The yellow curve in the figure) is made to pass through the point set P as much as possible. A The point in Figure 5 The yellow curve in the figure passes through the red point as much as possible.

[0052] In this embodiment, preferably, in step S4, the point set P is obtained by skeletonization method. A The specific method is: take the left tail area as the root node, take the right tail area as the terminal leaf node, and obtain the points sorted from left to right to form the point set P A .

[0053] Step S5: Sample N points equidistantly on the curve L1, calculate the tangent vector Dir at each point according to the sampled points and their adjacent points, and make N cross sections with each point as the plane origin and the tangent vector Dir as the plane normal vector. For each cross section, find its intersection with the grid of the area of ​​interest, and average the cross section intersections to find its centroid. Then arrange all the intersection centroids in order, recorded as L2, such as Figure 6 As shown, the green line is L2.

[0054] Specifically, in this embodiment, the number of sampling points N is in the range of 500≤N≤1200, and the more sampling points, the higher the accuracy. In this embodiment, preferably, considering both accuracy and data volume, N=1000 is taken.

[0055] Specifically, in this embodiment, the calculation method of the tangent vector Dir is: Dir=Normalize(Point(i+1)-Point(i)), where Point(i+1) and Point(i) represent two adjacent sampling points on L1.

[0056] Specifically, in this embodiment, each intersection line between the cross section and the grid of the region of interest is composed of a column of intersection points, and each intersection point falls on the cross section and the grid edge line segment of the region of interest at the same time, that is:

[0057]

[0058] In this embodiment, two lines L1 and L2 are obtained, each with 1000 points, where L1 is a skeletonized spline ( Figure 5 The yellow line in the middle), L2 is the line of the centroid of the cross section ( Figure 6 ), and both are sorted in the same way.

[0059] Step S6: Take the point on L1 as the starting point and the point on L2 as the end point, connect them into a vector ExtendDir, and shoot a straight line Line along the vector to find its intersection with the grid of the area of ​​interest. Finally, arrange all the intersections in the same order as the points on L1 and L2, recorded as L3, such as Figure 7 As shown, the red line is L3, L3 is the line of the alveolar ridge top, and the elongation vector corresponding to each point on L3.

[0060] Specifically, in this embodiment, 1000 pairs of points are connected into a vector ExtendDir, with the point on the yellow line as the starting point and the corresponding point on the green line as the end point.

[0061] Preferably, in step S6 of this embodiment, the method for obtaining the intersection of the straight line where the vector ExtendDir is located and the mesh of the area of ​​interest is as follows: for all triangular facets face of the mesh, the plane PL where the facets are located is obtained, and the intersection point InterPt between the straight line Line and the plane PL is obtained; if InterPt falls inside the face triangle, the intersection point InterPt is returned.

[0062] Preferably, this embodiment further includes the following steps: extending or contracting L3 to any length along the extension direction ExtendDir of each point, specifically:

[0063] TargetPoint=L3Point+ExtendLength*ExtendDir;

[0064] Among them: TargetPoint is the point on L3 after stretching or shrinking, L3Point is the point on L3, and ExtendLength is the length of stretching or shrinking.

[0065] In this embodiment, when the arch curve is required to fit the crown, ExtendLength is the average distance between the crown centroid and the corresponding point on the nearest L3, that is:

[0066]

[0067] ToothCentorid represents the center of mass of the tooth crown, ClosestL3Point represents the corresponding point on L3 that is closest to the center of mass of the tooth crown, and ExtendLength is calculated by taking the average of a total of N teeth.

[0068] In this embodiment, when the arch curve is required to fit the tooth root, ExtendLength takes the average distance between the tooth root point and the corresponding point on the nearest L3, that is:

[0069]

[0070] TootRoot represents the tooth root point, ClosestL3Point represents the corresponding point on L3 that is closest to the tooth root point, and ExtendLength is calculated by taking the average of a total of N teeth. Note that it should shrink inward at this time, so ExtendLength should be multiplied by -1 to take a negative value.

[0071] Step S7: Establish an ideal dental arch line that fits the alveolar ridge top based on L3.

[0072] Preferably, in step S7 of the present embodiment, the specific method for generating the dental arch line is as follows: equidistantly taking points on L3, and then projecting them onto the XOY plane of the standard coordinate system, and fitting a symmetrical XOY plane curve with an even-order polynomial through the least squares method, obtaining the curve parameter vector, and determining the XOY plane even-order polynomial curve; fitting with an odd-order polynomial in the YOZ space, determining the Spee height corresponding to each point on the plane curve, and finally generating a symmetrical ideal curve with Spee curvature in R^3.

[0073] Specifically, in this embodiment, the embodiment takes the case of no extension and no contraction (ie, ExtendLength=0) as an example to establish an ideal curve that fits the alveolar ridge top, and the specific method is:

[0074] First, take points at equal distances on the red line (L3), and then project them onto the XOY plane of the standard coordinate system. At this time, Ori (0, 0, 0), DirX (1, ​​0, 0), DirY (0, 1, 0), Figure 7 It can be viewed as a schematic diagram of the XOY plane.

[0075] And through the least squares method, a symmetrical XOY plane curve is fitted with an even-order polynomial. At this time, the independent variable matrix of the least squares method is recorded as A, and the first polynomial of X is taken as the value, that is:

[0076]

[0077] The dependent variable vector of the least squares method is recorded as b, with the corresponding Y as the value, that is:

[0078]

[0079] The curve parameter vector to be solved is recorded as c, then:

[0080] A T A*c=A T b;

[0081] c=(A T A) -1 *A T b;

[0082] The curve parameter vector is obtained by solving the least square matrix, thereby determining the even-order polynomial curve in the XOY plane.

[0083] Similarly, an odd-order polynomial is fitted in YOZ space to determine the Spee height corresponding to each point on the plane curve. I will not repeat it here, and finally generate a symmetrical ideal curve with Spee curvature in R^3, such as Figure 8 As shown by the pink line in the middle, the curve partially passes through the alveolar bone model and is covered by the model.

[0084] Example 2

[0085] The ideal archline generation device based on AI alveolar bone mesh segmentation of this embodiment includes:

[0086] at least one processor; and

[0087] at least one memory in communication with the processor;

[0088] The memory stores instructions executable by the processor, and the instructions are executed by the processor so that the device executes the ideal archline generation method based on AI alveolar bone mesh segmentation in Example 1.

[0089] Many specific details are described in the above description to facilitate a full understanding of the present invention. However, the above description is only a preferred embodiment of the present invention. The present invention can be implemented in many other ways different from those described herein, so the present invention is not limited to the specific implementation disclosed above. At the same time, any person familiar with the art can make many possible changes and modifications to the technical solution of the present invention using the methods and technical contents disclosed above without departing from the scope of the technical solution of the present invention, or modify it into an equivalent embodiment of equivalent changes. Any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the scope of protection of the technical solution of the present invention.

Claims

1. An ideal archline generation method based on AI alveolar bone mesh segmentation, characterized by: The steps include: Step S1: obtaining the alveolar bone area of ​​interest according to the complete CT alveolar bone model; Step S2: performing boundary smoothing processing on the acquired alveolar bone focus area; Step S3: construct a coordinate system for the region of interest after boundary smoothing and transfer it to the standard coordinate system; Step S4: Obtain the point set P that represents the overall skeleton trend A , and according to the point set P A Connect the points in a smooth B-spline curve L1; Step S5: Sample N points at equal intervals on the curve L1, calculate the tangent vector Dir at each point according to the sampled points and their adjacent points, make N cross sections with each point as the plane origin and the tangent vector Dir as the plane normal vector, find the intersection line of each cross section with the grid of the area of ​​interest, and average the intersection lines of the cross sections to find their centroid, and then arrange the centroids of all intersection lines in order, recorded as L2; Step S6: Take the point on L1 as the starting point and the point on L2 as the end point, connect them into a vector ExtendDir, and shoot a straight line Line along the vector to find its intersection with the grid of the area of ​​interest, and finally arrange all the intersections in the same order as the points on L1 and L2, recorded as L3; Step S7: Establish an ideal dental arch line that fits the alveolar ridge top based on L3.

2. The ideal archline generation method based on AI alveolar bone mesh segmentation according to claim 1 is characterized in that: In step S1, the alveolar bone focus area is obtained based on data annotation and AI grid segmentation algorithm. The alveolar bone focus area only includes the alveolar bone morphology near the tooth cavity, and the palate structure that is too far away from the tooth cavity is removed.

3. The ideal archline generation method based on AI alveolar bone mesh segmentation according to claim 1 is characterized in that: In step S2, the method for boundary smoothing is: find the maximum boundary of the segmented alveolar bone area of ​​interest, fit a cutting line with sufficiently high smoothness through the B-spline function, and then use the cutting line to re-cut a segmented area with a sufficiently smooth boundary on the complete alveolar bone model, and replace the original area of ​​interest.

4. The ideal archline generation method based on AI alveolar bone mesh segmentation according to claim 1, characterized in that: In step S3, the plane is fitted using the least squares method, and the least squares matrix at this time is: According to X=(A T A) -1 A T b Solve the plane parameters a0, a1, a2 and use its normal vector as the current X-axis; Then sum the normals of all the faces as the current Z axis, and use the center of mass as the origin to construct an orthogonal coordinate system of the current position according to the right-hand rule; Finally, the grid of the area of ​​interest is transferred to the standard coordinate system through coordinate system transformation.

5. The ideal archline generation method based on AI alveolar bone mesh segmentation according to claim 1, characterized in that: In step S4, the point set P is obtained by skeletonization method. A The specific method is: take the left tail area as the root node, take the right tail area as the terminal leaf node, and obtain the points sorted from left to right to form the point set P A .

6. The ideal archline generation method based on AI alveolar bone mesh segmentation according to claim 1, characterized in that: In step S6, the method for obtaining the intersection of the line where the vector ExtendDir is located and the mesh of the area of ​​interest is as follows: for all triangular facets face of the mesh, find the plane PL where the facets are located, and obtain the intersection point InterPt of the line Line and the plane PL. If InterPt falls inside the face triangle, then return the intersection point InterPt.

7. The ideal archline generation method based on AI alveolar bone mesh segmentation according to claim 1, characterized in that: In step S7, the specific method for generating the dental arch line is as follows: equidistant points are collected on L3, and then projected onto the XOY plane of the standard coordinate system, and a symmetrical XOY plane curve is fitted with an even-order polynomial through the least squares method to obtain the curve parameter vector and determine the XOY plane even-order polynomial curve; an odd-order polynomial is fitted in the YOZ space to determine the Spee height corresponding to each point on the plane curve, and finally a symmetrical ideal curve with Spee curvature is generated.

8. The ideal archline generation method based on AI alveolar bone mesh segmentation according to claim 1, characterized in that: The following steps are also included between step S6 and step S7: along the extension direction ExtendDir of each point, L3 is extended or contracted by any length, specifically: TargetPoint=L3Point+ExtendLength*ExtendDir, where: TargetPoint is the point after extension or contraction, L3Point is the point on L3, and ExtendLength is the extension or contraction length.

9. The ideal archline generation method based on AI alveolar bone mesh segmentation according to claim 8, characterized in that: When the arch curve needs to fit the crown, ExtendLength takes the average distance between the crown centroid and its corresponding nearest point on L3; when the arch curve needs to fit the root, ExtendLength takes the average distance between the root point and its corresponding nearest point on L3.

10. An ideal archline generation device based on AI alveolar bone mesh segmentation, characterized by: include: at least one processor; as well as at least one memory in communication with the processor; Wherein, the memory stores instructions executable by the processor, and the instructions are executed by the processor so that the device executes the ideal archline generation method based on AI alveolar bone mesh segmentation as described in any one of claims 1-9.

Images