Method for Measuring the Distance between Cementoenamel Junction and Alveolar Crest Apex Based on Oral Medical Images

Abstract

The present invention relates to a method for measuring the distance between the cementoenamel junction and the alveolar ridge crest based on oral medical images, comprising: S1. Obtaining the optimal observation plane from historical CBCT images, outlining the contours of all teeth, and obtaining the first training set; S2. Training a dental arch segmentation model according to the first training set; S3. Obtaining the contour of each tooth from the optimal observation plane, calculating the central position of each tooth and the angle of the maximum buccolingual cross-section; S4. Obtaining the maximum buccolingual cross-section of each tooth in the historical CBCT images, and marking the position information of the cementoenamel junction and the alveolar ridge crest as the second training set; S5. Training a feature point detection model according to the second training set, and obtaining the position information of the cementoenamel junction and the alveolar ridge crest in the patient's CBCT image; S6. Calculating the bone dehiscence spacing value, and determining whether it is greater than the spacing threshold. If so, a prompt message indicating the presence of bone dehiscence symptoms is output. Compared with the prior art, the present invention has the advantages of improving the diagnosis efficiency of bone dehiscence, etc.

Description

Technical Field

[0001] The present invention relates to the technical field of oral diagnosis, and in particular to a method for measuring the distance between the cementoenamel junction and the alveolar crest based on oral medical images. Background Art

[0002] Bone dehiscence is a common form of alveolar bone defect, which refers to a V-shaped defect in the alveolar bone on the labial / buccal or lingual / palatal side of a tooth, reaching the alveolar crest. The gold standard for bone dehiscence diagnosis is to visually observe the state of the alveolar bone near the tooth root after flipping back the gingival tissue to expose the alveolar bone. However, this process is highly traumatic to the patient and cannot be used as a routine method.

[0003] Currently, the non-invasive method for bone dehiscence diagnosis is to observe using oral medical images. First, manually find the maximum buccolingual cross-section of each tooth, and then find two feature points, namely the cementoenamel junction and the alveolar crest, in the maximum buccolingual cross-section. Whether bone dehiscence has occurred is judged based on the distance between the two feature points. Therefore, in the process of bone dehiscence diagnosis, determining the distance between the cementoenamel junction and the alveolar crest of each tooth is the most important step, and this step currently needs to be completed manually by a doctor, which consumes a lot of the doctor's time. If a method can be provided to automatically determine the distance between the cementoenamel junction and the alveolar crest of each tooth based on the patient's cone beam CT (CBCT) images, the doctor's diagnosis efficiency will be greatly improved. Summary of the Invention

[0004] The purpose of the present invention is to overcome the above-mentioned defects existing in the prior art and provide a method for measuring the distance between the cementoenamel junction and the alveolar crest based on oral medical images, effectively improving the diagnosis efficiency of bone dehiscence.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] A method for measuring the distance between the cementoenamel junction and the alveolar crest based on oral medical images specifically includes the following steps:

[0007] S1. Based on a preset method for obtaining an observation plane, respectively obtain the corresponding optimal observation plane from historical CBCT images, outline the contours of all teeth in the optimal observation plane, and obtain a first training set;

[0008] S2. Perform network training according to the first training set to obtain a dental arch segmentation model;

[0009] S3. Obtain the contour of each tooth from the optimal observation plane through the dental arch segmentation model, and then calculate the central position of each tooth and the angle of the maximum buccolingual cross-section of the tooth;

[0010] S4. Obtain the maximum buccolingual cross-section of each tooth in the historical CBCT images according to the central position of each tooth and the angle of the maximum buccolingual cross-section. In each image of the maximum buccolingual cross-section, mark the position information of the cementoenamel junction and the alveolar crest top as the second training set;

[0011] S5. Perform network training according to the second training set to obtain a feature point detection model, and obtain the position information of the cementoenamel junction and the alveolar crest top in the maximum buccolingual cross-section of the patient's CBCT image according to the feature point detection model;

[0012] S6. Calculate the bone dehiscence spacing value according to the position information of the cementoenamel junction and the alveolar crest top, and determine whether the bone dehiscence spacing value is greater than the spacing threshold. If so, output a prompt message indicating that the corresponding tooth has a bone dehiscence symptom.

[0013] The process of the observation plane acquisition method in step S1 is specifically as follows:

[0014] S11. Establish an XYZ coordinate system with the patient's face orientation as the X direction, the right side orientation as the Y direction, and the top of the head orientation as the Z direction. Use the CBCT data to obtain an XOZ plane image passing through the center of the head;

[0015] S12. Calculate the position of the pixel point with the highest gray value in the XOZ plane as the target pixel point;

[0016] S13. Subtract the optimal observation threshold from the Z value of the target pixel point as the height where the optimal observation plane of the lower anterior tooth area is located, and add the optimal observation threshold to the Z value of the target pixel point as the height where the optimal observation plane of the upper anterior tooth area is located, to obtain the corresponding XOY plane as the optimal observation plane of the lower anterior tooth area and the upper anterior tooth area.

[0017] Further, the optimal observation threshold is specifically 20.

[0018] The calculation formula for the central position of the tooth in step S3 is as follows:

[0019] ,

[0020] where, is the coordinate of the central position of the tooth, is the coordinate of the pixel inside the tooth contour, , N is the total number of pixel points inside the tooth contour.

[0021] The process of calculating the angle of the maximum buccolingual cross-section of the tooth in step S3 is specifically as follows:

[0022] S31. Take the calculated central position of the tooth as the center, and traverse all the angles where the buccolingual cross-section is located according to the preset step size;

[0023] S32. Detect that there are two intersection points between the buccolingual cross-section and the tooth contour at each angle, calculate the distance between the two intersection points, and take the angle corresponding to the buccolingual cross-section with the largest distance as the angle of the largest buccolingual cross-section.

[0024] Further, the preset step size in step S31 is specifically 1°.

[0025] The dentition segmentation model adopts an instance segmentation network.

[0026] The feature point detection model adopts an object detection network.

[0027] The spacing threshold in step S6 is specifically 3 mm.

[0028] The bone dehiscence spacing value in step S6 adopts the Euclidean distance, and the specific formula is as follows:

[0029] ,

[0030] Where, is the bone dehiscence spacing value, and are the abscissas of the positions of the cementoenamel junction and the alveolar crest respectively, and are the ordinates of the positions of the cementoenamel junction and the alveolar crest respectively.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] The present invention utilizes the CBCT image information of the patient to calculate and output the maximum buccolingual cross-section corresponding to each tooth, locates the positions of the cementoenamel junction and the alveolar crest of each tooth in each maximum buccolingual cross-section of the patient, and automatically calculates the distance between the cementoenamel junction and the alveolar crest of each tooth to determine whether the patient has symptoms of bone dehiscence, effectively improving the diagnosis efficiency of bone dehiscence. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a schematic flow chart of the present invention;

[0034] Figure 2 is a schematic flow chart of the present invention for obtaining the best observation plane;

[0035] Figure 3 is a schematic structural diagram of the best observation plane obtained by the present invention, where Figure 3 (a) is the best observation plane of the lower anterior tooth region, Figure 3 (b) is the best observation plane of the upper anterior tooth region;

[0036] Figure 4 is a schematic structural diagram of the dentition segmentation model of the present invention;

[0037] Figure 5 Schematic diagram of the tooth contour obtained in the embodiment of the present invention, where Figure 5 (a) is the original image, Figure 5 and (b) is the tooth contour obtained by the output of the model prediction;

[0038] Figure 6 Flow chart of calculating the tooth center and the maximum buccolingual cross-section angle of the present invention;

[0039] Figure 7 Schematic diagram of the tooth center and the corresponding angle of the maximum buccolingual cross-section obtained in the embodiment of the present invention;

[0040] Figure 8 Schematic diagram of the maximum buccolingual cross-section obtained in the embodiment of the present invention;

[0041] Figure 9 Schematic diagram of the structure of the feature point detection model of the present invention;

[0042] Figure 10 Schematic diagram of the feature point detection result in the embodiment of the present invention. Detailed implementation manners

[0043] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented on the premise of the technical solution of the present invention, and the detailed implementation manners and specific operation processes are given, but the protection scope of the present invention is not limited to the following embodiments.

[0044] Embodiment

[0045] As Figure 1 shown, a method for measuring the distance between the cementoenamel junction and the alveolar ridge crest based on oral medical images specifically includes the following steps:

[0046] S1. Based on the preset method for obtaining the observation plane, the corresponding optimal observation planes are respectively obtained from the historical CBCT images, and the contours of all teeth are outlined in the optimal observation planes to obtain the first training set;

[0047] S2. Network training is performed according to the first training set to obtain a dental arch segmentation model;

[0048] S3. The contour of each tooth is obtained from the optimal observation plane through the dental arch segmentation model, and then the central position of each tooth and the angle of the maximum buccolingual cross-section of the tooth are calculated;

[0049] S4. According to the central position of each tooth and the angle of the maximum buccolingual cross-section, the maximum buccolingual cross-section of each tooth is obtained in the historical CBCT image, and the position information of the cementoenamel junction and the alveolar ridge crest is marked in each image of the maximum buccolingual cross-section as the second training set;

[0050] S5. Perform network training based on the second training set to obtain a feature point detection model, and obtain the position information of the cementoenamel junction and the alveolar crest apex in the maximum buccolingual section of the patient's CBCT image according to the feature point detection model;

[0051] S6. Calculate the bone dehiscence spacing value according to the position information of the cementoenamel junction and the alveolar crest apex, and determine whether the bone dehiscence spacing value is greater than the spacing threshold. If so, output a prompt message indicating that the corresponding tooth has a bone dehiscence symptom.

[0052] As Figure 2 shown, the process of the observation plane acquisition method in step S1 is as follows:

[0053] S11. Establish an XYZ coordinate system with the patient's face orientation as the X direction, the right side orientation as the Y direction, and the head top orientation as the Z direction, and obtain an XOZ plane image passing through the center of the head using CBCT data;

[0054] S12. Calculate the position of the pixel point with the highest gray value in the XOZ plane as the target pixel point;

[0055] S13. Subtract the optimal observation threshold from the Z value of the target pixel point as the height where the optimal observation plane of the lower anterior tooth region is located, and add the optimal observation threshold to the Z value of the target pixel point as the height where the optimal observation plane of the upper anterior tooth region is located, and obtain the corresponding XOY plane as the optimal observation plane of the lower anterior tooth region and the upper anterior tooth region.

[0056] The optimal observation threshold is specifically 20.

[0057] In this experimental example, 200 CBCT data of patients are used to obtain 200 optimal observation plane images of the upper anterior tooth region and 200 optimal observation plane images of the lower anterior tooth region. The obtained optimal observation planes are as Figure 3 shown.

[0058] In this experimental example, 400 observation plane images of the upper anterior tooth region and the lower anterior tooth region are manually outlined to outline the contour of each tooth, and then the data is enhanced by using methods of translation, rotation, and scaling to obtain the first training set required for training the dental arch segmentation model.

[0059] The network structure of the dental arch segmentation model is as Figure 4 shown. The model adopts a MaskRCNN instance segmentation network structure and uses a two-stage method. In the first stage, the RPN network is used to scan the image and generate proposal regions. In the second stage, a convolutional network is used for class and boundary regression, and at the same time, a corresponding binary mask is predicted for each ROI region. In this experimental example, the schematic diagram of the tooth contour obtained by model prediction is as Figure 5 shown.

[0060] In step S3, the calculation formula for the central position of the tooth is as follows:

[0061] ,

[0062] where is the coordinate of the central position of the tooth, is the coordinate of the pixel within the tooth contour, , N is the total number of pixel points within the tooth contour.

[0063] As Figure 6 shown, the process of calculating the angle of the maximum buccolingual section of the tooth in step S3 is as follows:

[0064] S31. Taking the calculated central position of the tooth as the center, traverse all the angles where the buccolingual section is located according to a preset step size;

[0065] S32. Detect that there are two intersection points between the buccolingual section and the tooth contour at each angle, calculate the distance between the two intersection points, and take the angle corresponding to the buccolingual section with the maximum distance as the angle of the maximum buccolingual section.

[0066] The preset step size in step S31 is specifically 1°.

[0067] In this embodiment, the central position of the tooth and the angle corresponding to the maximum buccolingual section are as Figure 7 shown.

[0068] In this experimental example, the schematic diagram of the maximum buccolingual section is as Figure 8 shown.

[0069] The structural schematic diagram of the feature point detection model is as Figure 9 shown. The model adopts the object detection network YOLOv5 network structure. The model uses CSPDarknet53 as the backbone feature extraction network and outputs feature information at three scales. Further feature extraction and information fusion are performed through subsequent convolutional networks. At the same time, the method of adaptive anchor box calculation is used to achieve fast and accurate detection of the target.

[0070] In this experimental example, 1044 maximum buccolingual sections are labeled, and the data is enhanced by translation, rotation, and scaling methods to obtain the second training set required for training the feature point detection model.

[0071] The spacing threshold in step S6 is specifically 3 mm.

[0072] In this experimental example, the positions of the two feature points of the cementoenamel junction and the alveolar crest apex obtained are displayed, as Figure 10As shown, the bone cracking spacing value in step S6 adopts the Euclidean distance, and the specific formula is as follows:

[0073] ,

[0074] wherein, is the bone cracking spacing value, and are respectively the abscissas of the positions of the cementoenamel junction and the alveolar crest top, and are respectively the ordinates of the positions of the cementoenamel junction and the alveolar crest top.

[0075] In addition, it should be noted that for the specific embodiments described in this specification, the names taken may be different. The above content described in this specification is only an example of the structure of the present invention. Any equivalent changes or simple changes made according to the structure, features, and principles of the present invention are included in the protection scope of the present invention. Those skilled in the art of the present invention can make various modifications or supplements to the described specific examples or adopt similar methods, as long as they do not deviate from the structure of the present invention or exceed the scope defined by this claims, they should fall within the protection scope of the present invention.

Claims

1. A method for measuring the distance between the cementoenamel junction and the alveolar crest based on oral medical images, characterized in that, it specifically includes the following steps: S1. Based on the preset method for obtaining the observation plane, respectively obtain the corresponding optimal observation planes from historical CBCT images, outline the contours of all teeth in the optimal observation planes, and obtain the first training set; S2. Perform network training according to the first training set to obtain a dental arch segmentation model; S3. Obtain the contour of each tooth from the optimal observation plane through the dental arch segmentation model, and then calculate the central position of each tooth and the angle of the largest buccolingual cross-section of the tooth; S4. According to the central position of each tooth and the angle of the largest buccolingual cross-section, obtain the largest buccolingual cross-section of each tooth in the historical CBCT image, and mark the position information of the cementoenamel junction and the alveolar crest in each image of the largest buccolingual cross-section as the second training set; S5. Perform network training according to the second training set to obtain a feature point detection model, and obtain the position information of the cementoenamel junction and the alveolar crest in the largest buccolingual cross-section of the patient's CBCT image according to the feature point detection model; S6. Calculate the bone dehiscence spacing value according to the position information of the cementoenamel junction and the alveolar crest, and judge whether the bone dehiscence spacing value is greater than the spacing threshold. If so, output a prompt message indicating that the corresponding tooth has a bone dehiscence symptom; The process of the observation plane acquisition method in step S1 is specifically as follows: S11. Establish an XYZ coordinate system with the patient's face orientation as the X direction, the right side orientation as the Y direction, and the top of the head orientation as the Z direction, and use the CBCT data to obtain an XOZ plane image passing through the center of the head; S12. Calculate the position of the pixel point with the highest gray value in the XOZ plane as the target pixel point; S13. Subtract the optimal observation threshold from the Z value of the target pixel point as the height where the optimal observation plane of the lower anterior tooth area is located, and add the optimal observation threshold to the Z value of the target pixel point as the height where the optimal observation plane of the upper anterior tooth area is located, and obtain the corresponding XOY plane as the optimal observation plane of the lower anterior tooth area and the upper anterior tooth area; The calculation formula for the central position of the tooth in step S3 is as follows: ， wherein, is the coordinate of the central position of the tooth, is the coordinate of the pixel within the tooth contour, , N is the total number of pixel points within the tooth contour; The process of calculating the angle of the largest buccolingual cross-section of the tooth in step S3 is specifically as follows: S31. Take the calculated central position of the tooth as the center, and traverse all the angles where the buccolingual cross-section is located according to the preset step size; S32. Detect that there are two intersection points between the buccolingual cross-section and the contour of the tooth at each angle, calculate the distance between the two intersection points, and take the angle corresponding to the buccolingual cross-section with the largest distance as the angle of the largest buccolingual cross-section.

2. The method for measuring the distance between the cementoenamel junction and the alveolar crest based on oral medical images according to claim 1, characterized in that, the optimal observation threshold is specifically 20.

3. The method for measuring the distance between the cementoenamel junction and the alveolar crest based on oral medical images according to claim 1, characterized in that, the preset step size in step S31 is specifically 1°.

4. The method for measuring the distance between the cementoenamel junction and the alveolar crest based on oral medical images according to claim 1, characterized in that, The dental arch segmentation model uses an instance segmentation network.

5. A method for measuring the distance between the cementoenamel junction and the alveolar ridge crest based on oral medical images according to claim 1, wherein, the feature point detection model uses an object detection network.

6. A method for measuring the distance between the cementoenamel junction and the alveolar ridge crest based on oral medical images according to claim 1, wherein, the distance threshold in step S6 is specifically 3 mm.

7. A method for measuring the distance between the cementoenamel junction and the alveolar ridge crest based on oral medical images according to claim 1, wherein, the bone cracking distance value in step S6 uses the Euclidean distance, and the specific formula is as follows: ， Among them, is the value of the bone crack spacing, and are the abscissas of the positions of the cementoenamel junction and the alveolar crest respectively, and are the ordinates of the positions of the cementoenamel junction and the alveolar crest respectively.

Images