False tooth neck-edge line automatic extraction method and system
Through deep learning model and edge detection technology, the denture neck edge line is automatically extracted, which solves the problems of poor lighting and large calculations in the existing technology, and achieves rapid and accurate denture neck edge line repair, improving the quality of denture production.
Patent Information
- Application Number
- CN202510040374.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-10
- Publication Date
- 2025-04-15
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art has problems in the extraction process of denture neck edge lines, which lead to extraction deviation, large calculation amount, and difficulty in modifying the wrong parts separately, which affects the quality of denture production.
The deep learning model is used to combine edge detection and shape feature analysis to automatically extract the denture neck edge line, and optimize the model through the training set and loss function to achieve the function of quickly finding and modifying the error position.
It improves the accuracy and efficiency of denture neck edge line extraction, can quickly correct and repair neck edge defects, and improve the quality of denture production.
Smart Images
Figure CN119831979A_ABST
Abstract
Description
[Technical field]
[0001] The present invention relates to the field of dentures, and in particular to a method and system for automatically extracting a denture neck margin line. [Background technology]
[0002] Extracting the denture neck margin is an important step in the denture making process, which involves precise measurement and model making of teeth and gums. Therefore, the gum part in each oral image taken needs to be extracted separately for further refined modeling. In modern dentistry, digital impression methods are becoming more and more popular, especially in complex restorations and aesthetic restorations. The advantages of digital technology are more obvious. Dental clinics and laboratories are gradually turning from traditional impression materials to digital impressions.
[0003] However, in the prior art, an edge detection algorithm is usually used to segment and extract the denture neck margin line of the gingival edge in the oral image. The environment in which the oral image is shot and collected requires high lighting, and the light is likely to form reflective spots on the tooth surface, resulting in deviations in the segmentation and extraction of the denture neck margin line. In addition, the amount of calculation is extremely large. The denture neck margin line of an erroneous tooth cannot be modified individually, and repeated calculations must be relied upon to first find the correct denture neck margin lines on both sides of the erroneous denture neck margin line. This undoubtedly greatly increases the amount of calculation, causes trouble for later modifications, and reduces the quality of denture production.
[0004] Therefore, a method and system for automatically extracting the denture neck margin line is proposed, which can quickly find the erroneous denture neck margin line position and quickly modify and repair it, thereby improving the production quality of the denture. [Summary of the invention]
[0005] The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide a method and system for automatically extracting the denture neck margin line, which can quickly find the erroneous denture neck margin line position and quickly modify and repair it, thereby improving the production quality of the denture.
[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions:
[0007] A method for automatically extracting a denture cervical margin line, the method comprising:
[0008] S1: Scan the maxillary and mandibular tooth images of the patient, perform edge detection on the grayscale image of each tooth image, and obtain an edge image corresponding to each tooth image;
[0009] S2: Acquire multiple tooth neck edge images, obtain the shape feature degree of the corresponding neck edge, obtain at least one suspected neck edge line edge in the neck edge image, and mark the left and right end point values and the lowest point value of the neck edge of each tooth in each tooth neck edge image;
[0010] S3: Count the left and right end point values of each tooth and the lowest point value of the cervical margin as a number set, and use it as a training set for the deep learning model; establish a deep learning model; train the deep learning model according to the training set until the loss function value of the deep learning model no longer decreases;
[0011] S4: Determine the first neck edge line according to the left end point of the tooth and the lowest point of the neck edge, and determine the second neck edge line according to the right end point of the tooth and the lowest point of the neck edge. The angle formed by the first neck edge line and the second neck edge line is the target angle X. 1 , locate and record the target angle;
[0012] S5: Record the second lowest point value of the neck edge of each tooth, count the left and right end point values of each tooth and the second lowest point value of the neck edge as a number set, repeat S4, and get the target angle X 2 ;
[0013] S6: Repeat S3-S5 until all target angles X are recorded. n .
[0014] As a preference, the loss function formula is
[0015] Among them, W vac is the loss function, M is the total number of trainings, n is the number of trainings, X n Record the value for the target angle, is the predicted value.
[0016] As a preference, the target angle is from X 1 To X n The angle gradually increases.
[0017] As a preference, the target angle X 1 The angle is minimum.
[0018] As a preference, all target angle data are normalized; and the digital grid model is layered according to the angle size.
[0019] Preferably, a data set of tooth cervical margin images is obtained;
[0020] Preprocess the dataset;
[0021] Preprocessing includes image cropping and HU value normalization;
[0022] Label the contour of each tooth’s cervical margin in the dataset;
[0023] The labeled data set is input into the deep learning model and the deep learning model is trained until the deep learning model converges.
[0024] Preferably, the tooth neck edge image is subjected to sample amplification processing;
[0025] The sample augmentation process includes random rotation, scaling, radial transformation and cropping of the tooth cervical margin image.
[0026] Preferably, the method further includes S7, which specifically includes: when the target angle is smaller than a critical value, marking it as a diseased area, measuring the diseased area, and automatically calculating the depth and width of the diseased area.
[0027] A denture cervical margin automatic extraction system, comprising:
[0028] A scanning module is used to scan the maxillary and mandibular teeth images of the patient, perform edge detection on the grayscale image of each tooth image, and obtain an edge image corresponding to each tooth image;
[0029] A marking module is used to obtain multiple tooth neck edge images, obtain the shape feature degree of the corresponding neck edge, obtain at least one suspected neck edge line edge in the neck edge image, and mark the left and right end point values and the lowest point value of the neck edge of each tooth in each tooth neck edge image;
[0030] The reinforcement module is used to count the left and right end point values of each tooth and the lowest point value of the cervical margin as a number set and use it as a training set for the deep learning model; establish a deep learning model; train the deep learning model according to the training set until the loss function value of the deep learning model no longer decreases;
[0031] The positioning module is used to determine the first neck edge connection line according to the left end point of the tooth and the lowest point of the neck edge, and determine the second neck edge connection line according to the right end point of the tooth and the lowest point of the neck edge. The angle formed by the first neck edge connection line and the second neck edge connection line is the target angle X 1 , locate and record the target angle;
[0032] The statistical module is used to record the second lowest point value of the neck edge of each tooth, and count the left and right end point values of each tooth and the second lowest point value of the neck edge as a data set to obtain the target angle X 2 .
[0033] A computer-readable storage medium stores a computer program, which implements the steps of the above method when executed by a processor.
[0034] The beneficial effects of the present invention are as follows:
[0035] 1. S1: Scan the maxillary and mandibular tooth images of the patient, perform edge detection on the grayscale image of each tooth image, and obtain an edge image corresponding to each tooth image;
[0036] S2: Acquire multiple tooth neck edge images, obtain the shape feature degree of the corresponding neck edge, obtain at least one suspected neck edge line edge in the neck edge image, and mark the left and right end point values and the lowest point value of the neck edge of each tooth in each tooth neck edge image;
[0037] The left and right end point values of each tooth are fixed. The lowest point value of the neck edge can be obtained by measurement. Two points can be fixed on the semicircular path to determine one point. In order to improve the accuracy, Y can be set. n To record the horizontal coordinates of the tooth's cervical margin, this method can be used to register all points on the tooth's cervical margin and review them at any time to find the erroneous points and achieve rapid error correction. At the same time, the defects on the cervical margin can be repaired in detail.
[0038] S3: Count the left and right end point values of each tooth and the lowest point value of the cervical margin as a number set, and use it as a training set for the deep learning model; establish a deep learning model; train the deep learning model according to the training set until the loss function value of the deep learning model no longer decreases, thereby improving the accuracy of tooth cervical margin point recognition;
[0039] S4: Determine the first neck edge line according to the left end point of the tooth and the lowest point of the neck edge, and determine the second neck edge line according to the right end point of the tooth and the lowest point of the neck edge. The angle formed by the first neck edge line and the second neck edge line is the target angle X. 1 , locate and record the target angle, and measure the target angle multiple times to improve the measurement accuracy;
[0040] S5: Record the second lowest point value of the neck edge of each tooth, count the left and right end point values of each tooth and the second lowest point value of the neck edge as a number set, repeat S4, and get the target angle X 2 ;
[0041] S6: Repeat S3-S5 until all target angles X are recorded. n .
[0042] The method further includes S7. Specifically, when the target angle is smaller than a critical value, the target area is marked as a diseased area, the diseased area is measured, and the depth and width of the diseased area are automatically calculated.
[0043] For example, when the target angle is less than 10 degrees, the diseased area has gum atrophy. The two end points of the tooth neck defect are registered, and the high points of the two end points are used as the standard points to set up the horizontal axis. The angle θ is set at the connection between the horizontal axis and the deepest point of the tooth neck defect. The high points of the two end points and the deepest point of the tooth neck defect determine the line length a, and the depth d = a*sinθ. The high points of the two end points are used as the standard to prevent the low points of the two end points from having gum atrophy, resulting in inaccurate recognition.
[0044] 2. The formula of the loss function is Among them, W vac is the loss function, M is the total number of trainings, n is the number of trainings, X n Record the value for the target angle, is the predicted value. It reduces point recognition errors caused by shooting distortion.
[0045] 3. Target angle from X 1 To X n The angle gradually increases, and the target angle X 1 The angle is the smallest. If the angle becomes larger, it means that there is an error in the measurement and a reverse check is required. The error can be corrected quickly during the recording process, reducing the number of unqualified dentures produced and improving the production qualification rate.
[0046] These features and advantages of the present invention will be disclosed in detail in the following specific embodiments and drawings.
Brief Description of the Drawings
[0047] The present invention will be further described below in conjunction with the accompanying drawings:
[0048] Figure 1 This is a schematic diagram of a process of Embodiment 1 of the present invention;
[0049] Figure 2 Schematic diagram of the target angle of the present invention. [Specific implementation method]
[0050] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0051] The following first describes the concepts involved in this application in conjunction with the accompanying drawings. It should be noted that the following description of each concept is only to make the content of this application easier to understand and does not limit the scope of protection of this application.
[0052] Embodiment 1:
[0053] A method for automatically extracting the denture cervical margin line, such as Figure 1 , Figure 2 As shown, the method includes:
[0054] S1: Scan the maxillary and mandibular tooth images of the patient, perform edge detection on the grayscale image of each tooth image, and obtain an edge image corresponding to each tooth image;
[0055] S2: Acquire multiple tooth neck edge images, obtain the shape feature degree of the corresponding neck edge, obtain at least one suspected neck edge line edge in the neck edge image, and mark the left and right end point values and the lowest point value of the neck edge of each tooth in each tooth neck edge image;
[0056] The left and right end point values of each tooth are fixed. The lowest point value of the neck edge can be obtained by measurement. Two points can be fixed on the semicircular path to determine one point. In order to improve the accuracy, Y can be set. n To record the horizontal coordinates of the tooth's cervical margin, this method can be used to register all points on the tooth's cervical margin and review them at any time to find the erroneous points and achieve rapid error correction. At the same time, the defects on the cervical margin can be repaired in detail.
[0057] S3: Count the left and right end point values of each tooth and the lowest point value of the cervical margin as a number set, and use it as a training set for the deep learning model; establish a deep learning model; train the deep learning model according to the training set until the loss function value of the deep learning model no longer decreases, thereby improving the accuracy of tooth cervical margin point recognition;
[0058] S4: Determine the first neck edge line according to the left end point of the tooth and the lowest point of the neck edge, and determine the second neck edge line according to the right end point of the tooth and the lowest point of the neck edge. The angle formed by the first neck edge line and the second neck edge line is the target angle X. 1 ,like Figure 2 As shown, the target angle is located and recorded, and the target angle can be measured multiple times to improve the measurement accuracy;
[0059] S5: Record the second lowest point value of the neck edge of each tooth, count the left and right end point values of each tooth and the second lowest point value of the neck edge as a number set, repeat S4, and get the target angle X 2 ;
[0060] S6: Repeat S3-S5 until all target angles X are recorded. n .
[0061] Target angle X n Multiple measurements will be performed and recorded as measurement subset x i , get the target angular coordinates as (GX, GY), Wherein, H=N×X2Y+N×Y3-(X2+Y2)×Y;D=N×XY-X×Y;E=N×X3+N×XY2-(X2+Y2)×X;G=N×XY-Y×Y;C=N×X2-X×X;
[0062]
[0063] N represents the number of measurements; (x i ,y i) represents the i-th target angle X n The measured coordinates.
[0064] The leftmost coordinate of the tooth is (Q x ,Q y ), the rightmost coordinate of the tooth is (W x ,W y ), ZZ=W x -GX+Q x ; YY=W y -GY+Q y , when XX is greater than 0 and YY is greater than or equal to 0, a is the distance between the leftmost end of the tooth and the rightmost end of the tooth.
[0065] The method further includes S7. Specifically, when the target angle is smaller than a critical value, the target area is marked as a diseased area, the diseased area is measured, and the depth and width of the diseased area are automatically calculated.
[0066] For example, when the target angle is less than 10 degrees, the diseased area has gum atrophy. The two end points of the tooth neck defect are registered, and the high points of the two end points are used as the standard points to set up the horizontal axis. The angle θ is set at the connection between the horizontal axis and the deepest point of the tooth neck defect. The high points of the two end points and the deepest point of the tooth neck defect determine the line length a, and the depth d = a*sinθ. The high points of the two end points are used as the standard to prevent the low points of the two end points from having gum atrophy, resulting in inaccurate recognition.
[0067] The formula of the loss function is Among them, W vac is the loss function, M is the total number of trainings, n is the number of trainings, X n Record the value for the target angle, is the predicted value. It reduces point recognition errors caused by shooting distortion.
[0068] Target angle from X 1 To X n The angle gradually increases, and the target angle X 1 The angle is the smallest. If the angle becomes larger, it means that there is an error in the measurement and a reverse check is required. The error can be corrected quickly during the recording process, reducing the number of unqualified dentures produced and improving the production qualification rate.
[0069] Normalize all target angle data; layer the digital grid model according to the angle size, place the data reasonably, and realize fast data search when error correction is needed.
[0070] The normalization formula is:
[0071] Among them, x newis the normalized value, x is the original data value of the target value, and x max is the maximum value among the first target values, x min is the minimum value among the first target values.
[0072] Obtain a dataset of tooth cervical margin images;
[0073] Preprocess the dataset;
[0074] Preprocessing includes image cropping and HU value normalization;
[0075] Label the contour of each tooth’s cervical margin in the dataset;
[0076] The labeled data set is input into the deep learning model and the deep learning model is trained until the deep learning model converges.
[0077] Perform sample amplification processing on the tooth cervical margin image;
[0078] The deep learning model formula is:
[0079] The sample augmentation process includes random rotation, scaling, radial transformation and cropping of the tooth neck edge image, and finally integrating each tooth neck edge image to form a full tooth neck edge image of the photographer.
[0080] A denture cervical margin automatic extraction system, comprising:
[0081] A scanning module is used to scan the patient's maxillary and mandibular tooth images, perform edge detection on the grayscale image of each tooth image, and obtain an edge image corresponding to each tooth image;
[0082] A marking module is used to obtain multiple tooth neck edge images, obtain the shape feature degree of the corresponding neck edge, obtain at least one suspected neck edge line edge in the neck edge image, and mark the left and right end point values and the lowest point value of the neck edge of each tooth in each tooth neck edge image;
[0083] The reinforcement module is used to count the left and right end point values of each tooth and the lowest point value of the cervical margin as a number set and use it as a training set for the deep learning model; establish a deep learning model; train the deep learning model according to the training set until the loss function value of the deep learning model no longer decreases;
[0084] The positioning module is used to determine the first neck edge connection line according to the left end point of the tooth and the lowest point of the neck edge, and determine the second neck edge connection line according to the right end point of the tooth and the lowest point of the neck edge. The angle formed by the first neck edge connection line and the second neck edge connection line is the target angle X 1 , locate and record the target angle;
[0085] The statistical module is used to record the second lowest point value of the neck edge of each tooth, and count the left and right end point values of each tooth and the second lowest point value of the neck edge as a data set to obtain the target angle X 2 .
[0086] A computer-readable storage medium stores a computer program, which implements the steps of the above method when executed by a processor.
[0087] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Those skilled in the art should understand that the present invention includes but is not limited to the contents described in the drawings and the above specific embodiments. Any modification that does not deviate from the functional and structural principles of the present invention will be included in the scope of the claims.
[0088] Those of ordinary skill in the art will appreciate that the units (or modules, steps, etc., the same below) of the examples described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition of each example has been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.
[0089] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.
[0090] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0091] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-0nly Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk, etc., which can store program code.
[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.
Claims
1. A method for automatically extracting the denture neck edge line, characterized in that: The method comprises: S1: Scanning maxillary and mandibular dental images of a patient, performing edge detection on a grayscale image of each of the dental images, and obtaining an edge image corresponding to each of the dental images; S2: Acquire multiple tooth neck edge images, obtain the shape feature degree of the corresponding neck edge, obtain at least one suspected neck edge line edge in the neck edge image, and mark the left and right end point values and the lowest point value of the neck edge of each tooth in each tooth neck edge image; S3: Count the left and right end point values of each tooth and the lowest point value of the cervical margin as a number set, and use it as a training set for the deep learning model; establish the deep learning model; train the deep learning model according to the training set until the loss function value of the deep learning model no longer decreases; S4: determining a first neck edge connection line according to the left end point of the tooth and the lowest point of the neck edge, and determining a second neck edge connection line according to the right end point of the tooth and the lowest point of the neck edge, the angle formed between the first neck edge connection line and the second neck edge connection line is the target angle X1, and the target angle is located and recorded; S5: record the second lowest point value of the neck edge of each tooth, count the left and right end point values of each tooth and the second lowest point value of the neck edge as a number set, repeat S4 to obtain the target angle X2; S6: Repeat S3-S5 until all target angles X are recorded. n .
2. The method for automatically extracting the denture neck margin line according to claim 1, characterized in that: The formula of the loss function is Among them, W vac is the loss function, M is the total number of trainings, n is the number of trainings, X n Record the value for the target angle, is the predicted value.
3. The method for automatically extracting the denture neck margin line according to claim 1, characterized in that: The target angle is from X1 to X n The angle gradually increases.
4. The method for automatically extracting the denture neck margin line according to claim 3, characterized in that: The target angle X1 is the minimum angle.
5. The method for automatically extracting the denture neck margin line according to claim 1, characterized in that: All target angle data are normalized; and the digital grid model is layered according to the angle size.
6. The method for automatically extracting the denture neck margin line according to claim 1, characterized in that: Acquire a data set of the tooth neck edge image; Preprocessing the data set; The preprocessing includes image cropping and HU value normalization; Annotating the contour of the cervical margin of each tooth in the data set; The labeled data set is input into the deep learning model, and the deep learning model is trained until the deep learning model converges.
7. The method for automatically extracting the denture neck margin line according to claim 1, characterized in that: Performing sample amplification processing on the tooth neck edge image; The sample augmentation process includes randomly rotating, scaling, radially transforming and cropping the tooth neck edge image.
8. The method for automatically extracting the denture neck margin line according to claim 1, characterized in that: The method further includes S7. Specifically, when the target angle is smaller than a critical value, the target area is marked as a diseased area, the diseased area is measured, and the depth and width of the diseased area are automatically calculated.
9. An automatic denture neck edge line extraction system, characterized in that: include: A scanning module, used for scanning the maxillary and mandibular dental images of the patient, performing edge detection on the grayscale image of each of the dental images, and obtaining an edge image corresponding to each of the dental images; A marking module, used to obtain multiple tooth neck edge images, obtain the shape feature degree of the corresponding neck edge, obtain at least one suspected neck edge line edge in the neck edge image, and mark the left and right end point values and the lowest point value of the neck edge of each tooth in each tooth neck edge image; A reinforcement module is used to count the left and right end point values of each tooth and the lowest point value of the cervical margin as a number set and use it as a training set for the deep learning model; establish the deep learning model; train the deep learning model according to the training set until the loss function value of the deep learning model no longer decreases; A positioning module is used to determine a first neck edge line according to the left end point of the tooth and the lowest point of the neck edge, and to determine a second neck edge line according to the right end point of the tooth and the lowest point of the neck edge. The angle formed between the first neck edge line and the second neck edge line is the target angle X1, and the target angle is positioned and recorded; The statistical module is used to record the second lowest point value of the cervical margin of each tooth, count the left and right end point values of each tooth and the second lowest point value of the cervical margin as a number set, and obtain the target angle X2.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.