A method and apparatus for processing an intraoral scan image

Abstract

The application discloses an oral cavity scanning image processing method and device. The method comprises the following steps: determining an alveolar process associated region based on a sagittal projection image corresponding to an oral cavity scanning image, wherein the alveolar process associated region comprises a mandibular alveolar process region and at least part of a mandibular ascending ramus region; obtaining an axial projection image corresponding to the alveolar process associated region in the oral cavity scanning image; performing image segmentation on the axial projection image to obtain a dental arch mask image; and obtaining a dental arch curve based on the dental arch mask image. The alveolar process associated region is not affected by the field of view and the tooth condition in the oral cavity, so the process of obtaining the dental arch curve does not need to be constrained by the field of view and the tooth condition in the oral cavity. In this way, a stable and accurate dental arch mask image can be obtained in the scene of a local small field of view CBCT image and a complex tooth condition (full edentulism, mixed dentition, metal artifacts, etc.), and a stable and accurate dental arch curve can be obtained based on the dental arch mask image.

Description

Technical Field

[0001] This application relates to the field of dental image processing technology, specifically to an oral cavity scanning image processing method, an oral cavity scanning image processing device, an electronic device, and a computer-readable storage medium. Background Technology

[0002] The dental arch curve is an ideal curve tangent to the dentition. Fitting and generating the dental arch curve is crucial for assisting orthodontic design and simulating tooth arrangement trials. With the development of oral treatment and orthodontics, the requirements for dental arch curve fitting are becoming increasingly stringent. Various geometric curves (parabolas, ellipses, catenaries, etc.) and mathematical functions (beta functions, the Fourier Series, polytropic cosine functions, etc.) are all applied to describe the morphology of the dental arch.

[0003] Currently, teeth are used as the basis for acquiring dental arch mask images, and high requirements are placed on the integrity of the field of view (e.g., the jawbone is required to be basically complete in the CBCT image, and the teeth are in good condition with no significant missing teeth). However, for CBCT images with small to medium field of view in local areas, or for scenarios with complex tooth conditions (complete missing teeth, mixed dentition, metal artifacts, etc.), it is impossible to generate stable and accurate dental arch curves. Summary of the Invention

[0004] This application provides a method, apparatus, electronic device, and computer-readable storage medium for processing oral scan images, in order to solve the problem in the prior art that it is impossible to generate stable and accurate dental arch curves for CBCT images with small to medium field of view and scenes with complex dental conditions.

[0005] To solve or improve the aforementioned technical problems to some extent, according to one aspect of this application, a method for generating dental arch curves is provided, the method comprising:

[0006] The alveolar process associated region is determined based on the sagittal projection image corresponding to the oral cavity scan image. The alveolar process associated region includes the mandibular alveolar process region and at least part of the mandibular ramus region.

[0007] Obtain the axial projection image corresponding to the alveolar process-associated region in the oral cavity scan image;

[0008] The axial projection image is segmented to obtain a dental arch mask image;

[0009] The dental arch curve is obtained based on the dental arch mask image.

[0010] In some embodiments, obtaining the dental arch curve based on the dental arch mask image includes:

[0011] Perform a distance transformation on the dental arch mask image to obtain a distance-transformed image.

[0012] The distance-transformed image is segmented using a skeleton extraction model to obtain an initial dental arch skeleton image. The skeleton extraction model is a pre-trained deep neural network model that outputs the corresponding dental arch skeleton image based on the input distance-transformed image.

[0013] A thinning algorithm is used to perform secondary skeleton extraction on the initial dental arch skeleton image to obtain the target dental arch skeleton image;

[0014] The dental arch curve is obtained based on the target dental arch skeleton image.

[0015] In some implementations, obtaining the dental arch curve based on the target dental arch skeleton image includes:

[0016] The target dental arch skeleton image is processed using cubic spline interpolation to obtain the dental arch curve.

[0017] In some embodiments, determining the alveolar process-related region based on the sagittal projection image corresponding to the oral cavity scan image includes:

[0018] The sagittal projection image corresponding to the oral cavity scan image is input into the alveolar process associated region extraction model to obtain the mask image of the alveolar process associated region output by the alveolar process associated region extraction model. The alveolar process associated region extraction model is a pre-trained deep neural network model for outputting the mask image of the corresponding alveolar process associated region based on the input sagittal projection image.

[0019] In some embodiments, obtaining the axial projection image corresponding to the alveolar process-associated region in the oral cavity scan image includes:

[0020] The region within the alveolar process associated area of ​​the oral cavity scan image is subjected to maximum density projection along the axial direction to obtain the axial projection image.

[0021] In some embodiments, the step of performing image segmentation on the axial projection image to obtain a dental arch mask image includes:

[0022] The axial projection image is input into the dental arch mask segmentation model to obtain the dental arch mask image output by the dental arch mask segmentation model. The dental arch mask segmentation model is a pre-trained deep neural network model used to output the corresponding dental arch mask image based on the input axial projection image.

[0023] In some embodiments, the method further includes:

[0024] The dental arch thickness is obtained based on the dental arch mask image;

[0025] The panoramic oral radiograph corresponding to the oral cavity scan image is reconstructed based on the dental arch curve and the dental arch thickness.

[0026] According to another aspect of this application, an oral cavity scanning image processing apparatus is provided, the apparatus comprising:

[0027] The alveolar process associated region determination unit is used to determine the alveolar process associated region based on the sagittal projection image corresponding to the oral cavity scan image. The alveolar process associated region includes the alveolar process region and at least part of the mandibular ramus region.

[0028] An axial projection image acquisition unit is used to acquire an axial projection image corresponding to the alveolar process associated region in the oral cavity scan image;

[0029] The dental arch mask image acquisition unit is used to perform image segmentation on the axial projection image to obtain the dental arch mask image;

[0030] The dental arch curve acquisition unit is used to obtain the dental arch curve based on the dental arch mask image.

[0031] According to another aspect of this application, an electronic device is provided, including a processor and a memory; wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the above-described method.

[0032] According to another aspect of this application, a computer-readable storage medium is provided that stores one or more computer instructions that are executed by a processor to implement the above-described method.

[0033] Compared with the prior art, this application has the following advantages:

[0034] The oral cavity scanning image processing method provided in this application determines the alveolar process associated region based on the sagittal projection image corresponding to the oral cavity scanning image. The alveolar process associated region includes the mandibular alveolar process region and at least part of the mandibular ramus region; obtains the axial projection image corresponding to the alveolar process associated region in the oral cavity scanning image; performs image segmentation on the axial projection image to obtain a dental arch mask image; and obtains the dental arch curve based on the dental arch mask image. This application determines the alveolar process associated region based on sagittal projection images and obtains the corresponding axial projection image of the alveolar process associated region. The axial projection image is then segmented to obtain a dental arch mask image. Since the method of obtaining the dental arch mask image depends on the alveolar process associated region, which includes the mandibular alveolar process region and at least part of the mandibular ramus region, this region is not affected by the field of vision or tooth condition in the oral cavity. Therefore, the process of obtaining the dental arch curve is not constrained by the field of vision or tooth condition in the oral cavity. In this way, stable and accurate dental arch mask images can be obtained in scenarios with local small and medium field of vision CBCT images and complex tooth conditions (complete missing teeth, mixed dentition, metal artifacts, etc.), and stable and accurate dental arch curves can be obtained based on the dental arch mask image.

[0035] Furthermore, this application performs a distance transformation on the dental arch mask image to obtain a distance-transformed image. A skeleton extraction model is then used to segment the distance-transformed image to obtain an initial dental arch skeleton image. The skeleton extraction model is a pre-trained deep neural network model designed to output the corresponding dental arch skeleton image based on the input distance-transformed image. A thinning algorithm is then used to perform secondary skeleton extraction on the initial dental arch skeleton image to obtain a target dental arch skeleton image. Finally, a dental arch curve is obtained based on the target dental arch skeleton image. This two-stage method for extracting the dental arch skeleton can obtain the correct target dental arch skeleton even when the imaging of the teeth and jawbone regions is incomplete, thereby obtaining an accurate dental arch curve. Attached Figure Description

[0036] Figure 1 This is a flowchart of the oral cavity scan image processing method provided in the embodiments of this application;

[0037] Figure 2 This is a schematic diagram of obtaining a sagittal projection image based on an oral cavity scan image provided in an embodiment of this application;

[0038] Figure 3 This is a schematic diagram of the alveolar process associated region provided in the embodiments of this application;

[0039] Figure 4 This is a schematic diagram of obtaining a mask image of the alveolar process associated region provided in an embodiment of this application;

[0040] Figure 5 This is a schematic diagram of obtaining an axial projection image provided in an embodiment of this application;

[0041] Figure 6 This is a schematic diagram of obtaining a dental arch mask image provided in an embodiment of this application;

[0042] Figure 7 This is a schematic diagram of how dental arch skeletons are obtained using existing technologies;

[0043] Figure 8 This is a schematic diagram of obtaining a dental arch skeleton provided in an embodiment of this application;

[0044] Figure 9 This is a schematic diagram of the panoramic image acquisition process provided in the embodiments of this application;

[0045] Figure 10 This is a schematic diagram of a partial panoramic image provided in an embodiment of this application;

[0046] Figure 11 This is a unit block diagram of the oral cavity scanning image processing device provided in the embodiments of this application;

[0047] Figure 12 This is a schematic diagram of the logical structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0048] Many specific details are set forth in the following description to provide a full understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this application; therefore, this application is not limited to the specific embodiments disclosed below.

[0049] To address scenarios involving dental arch curve generation and panoramic oral radiograph reconstruction, and in order to obtain stable and accurate dental arch mask images under conditions of severe tooth loss and significant metal artifacts, and to generate stable and accurate dental arch curves based on these mask images, as well as to achieve automatic panoramic radiograph reconstruction covering CBCT images, CT images, and other oral scan images in any scenario, this application provides an oral scan image processing method, a corresponding oral scan image processing device, an electronic device, and a computer-readable storage medium. The following embodiments provide a detailed description of the above-mentioned method, device, electronic device, and computer-readable storage medium.

[0050] This embodiment provides a method for processing oral scan images. The main application of this method can be a computing device used to automatically generate dental arch curves or automatically reconstruct panoramic oral images. Figure 1 This is a flowchart of the oral cavity scanning image processing method provided in the first embodiment of this application, which is described below in conjunction with... Figure 1 The method provided in this embodiment will be described in detail. The embodiments described below are used to explain the principle of the method and are not intended to limit actual use.

[0051] like Figure 1 As shown, the oral cavity scan image processing method provided in this embodiment includes the following steps:

[0052] S101, Determine the alveolar process associated region based on the sagittal projection image corresponding to the oral cavity scan image.

[0053] This step is used to determine the alveolar process associated region based on the sagittal projection image corresponding to the oral cavity scan image. The oral cavity scan image can be a CT (Computer Tomography) image, a CBCT (Cone Beam CT) image, or other scan images of the oral cavity. The alveolar process associated region includes the mandibular alveolar process region and at least part of the mandibular ramus region (e.g., the part of the mandible extending down to the mandibular ramus region). Figure 3 As shown, this is not a defined anatomical region. Since the alveolar process associated region includes the mandibular alveolar process region and part of the mandibular bone region extending to the mandibular ramus region, that is, this region spans the mandibular alveolar process and part of the mandibular ramus, the alveolar process associated region is not affected by the tooth condition or the visual field in the oral cavity. For example, in cases of complex tooth conditions (complete missing teeth, mixed dentition, metal artifacts, etc.) or poor visual field conditions such as small or medium-sized fields in the oral cavity (incomplete display of the jawbone), the alveolar process associated region can be stably and accurately extracted.

[0054] In this embodiment, the maximum intensity projection (MIP) method is used to project the three-dimensional oral cavity scan image along the sagittal direction to obtain a two-dimensional sagittal projection image (e.g., Figure 2 After (as shown), the sagittal projection image is input into the alveolar process associated region extraction model to obtain the mask image of the alveolar process associated region output by the model (e.g., ...). Figure 4 As shown), for example, using a predetermined number of sagittal projection image samples and corresponding alveolar process associated region image samples as training samples, deep neural network models with network structures such as U-Net model, variants of U-Net model (e.g., U-Net++ model), DeepLab series models, and Transformer are trained to obtain the above-mentioned alveolar process associated region extraction model for outputting a mask image of the corresponding alveolar process associated region based on the input sagittal projection image.

[0055] S102, obtain the axial projection image corresponding to the alveolar process associated region in the oral cavity scan image.

[0056] After determining the alveolar process-related region based on the sagittal projection image corresponding to the oral cavity scan image in the above steps, this step is used to obtain the axial projection image corresponding to the alveolar process-related region in the oral cavity scan image. In this embodiment, the region within the alveolar process-related region in the oral cavity scan image can be subjected to maximum density projection along the axial direction to obtain the axial projection image (e.g., Figure 5 (As shown).

[0057] S103, perform image segmentation on the axial projection image to obtain the dental arch mask image.

[0058] After obtaining the axial projection image corresponding to the alveolar process associated region in the above steps, this step is used to segment the axial projection image to obtain the dental arch mask image. Existing image segmentation methods mainly use traditional morphological processing methods such as thresholding. However, traditional morphological processing methods such as thresholding are sensitive to image grayscale features, and the segmentation stability and generalization are difficult to guarantee. In complex scenes (such as images with severe metal artifacts), they cannot accurately segment the target. To overcome this problem, this embodiment uses a deep neural network model to segment the axial projection image to obtain the dental arch mask image, specifically as follows: Figure 6 As shown, the axial projection image is input into the dental arch mask segmentation model to obtain the dental arch mask image output by the model. This dental arch mask segmentation model is a pre-trained deep neural network model. For example, a predetermined number of axial projection image samples and corresponding dental arch mask image samples are used as training samples to train deep neural network models with network structures such as U-Net, variants of U-Net (e.g., U-Net++), DeepLab series models, and Transformer, to obtain the aforementioned dental arch mask segmentation model that outputs the corresponding dental arch mask image based on the input axial projection image. By using the aforementioned deep neural network model to segment the axial projection image, stable and accurate segmentation of the axial projection image can be achieved in complex scenes with severe metal artifacts, thus obtaining an accurate dental arch mask image.

[0059] S104, dental arch curves are obtained based on dental arch mask images.

[0060] After segmenting the axial projection image to obtain the dental arch mask image in the above steps, this step is used to obtain the dental arch curve based on the dental arch mask image. Currently, a thinning algorithm is mainly used to extract the skeleton from the dental arch mask image, and cubic spline interpolation is used to process the extracted dental arch skeleton to obtain the dental arch curve. However, when the above oral scan image is a local field of view (e.g., when the imaging of the teeth and jawbone regions in the CBCT image is incomplete), the thinning algorithm cannot obtain the correct dental arch skeleton when extracting the skeleton from the dental arch mask image (e.g., ...). Figure 7 As shown, there is a significant difference between the dental arch skeleton obtained by the thinning algorithm and the ideal dental arch skeleton. Therefore, this embodiment uses the following two-stage skeleton extraction method to extract the dental arch skeleton and obtain the target dental arch skeleton image (e.g., Figure 8 As shown):

[0061] First, a distance transform is performed on the dental arch mask image to obtain a distance-transformed image. Specifically, the distance transform is performed on the dental arch mask image to obtain the nearest distance value between each point in the dental arch region and the background region. Then, the distance value is normalized (for example, the distance value is mapped to between 0 and 1) to obtain the distance-transformed image.

[0062] Secondly, the skeleton extraction model is used to segment the above distance-transformed image to obtain the initial dental arch skeleton image (i.e. Figure 8 The first stage of the dental arch skeleton is extracted using a pre-trained deep neural network model. For example, a predetermined number of distance-transformed image samples and corresponding dental arch skeleton image samples are used as training samples to train deep neural network models with network structures such as U-Net, variants of U-Net (e.g., U-Net++), DeepLab series models, and Transformer. This results in a skeleton extraction model that outputs the corresponding dental arch skeleton image based on the input distance-transformed image. Since the dental arch mask image does not have the image features of the dental arch skeleton lines, while the distance-transformed image contains the image features of the basic dental arch skeleton lines, the distance-transformed image makes the automatic extraction process of the dental arch skeleton lines learnable. That is, by using the distance-transformed image as the input of the skeleton extraction model, the ideal initial dental arch skeleton can be learned by the skeleton extraction model.

[0063] Finally, a thinning algorithm is used to perform secondary skeleton extraction on the initial dental arch skeleton image to obtain the target dental arch skeleton image (i.e., Figure 8 The final dental arch skeleton in the process.

[0064] After obtaining the target dental arch skeleton image through the above process, the target dental arch skeleton image is processed using cubic spline interpolation to obtain the dental arch curve. Compared with existing methods that use thinning algorithms to extract the skeleton from the dental arch mask image, the above two-stage method for extracting the dental arch skeleton can obtain the correct target dental arch skeleton even when the imaging of the teeth and jawbone regions is incomplete, thereby obtaining an accurate dental arch curve.

[0065] This embodiment determines the alveolar process associated region based on sagittal projection images and obtains the corresponding axial projection image of the alveolar process associated region. The axial projection image is then segmented to obtain a dental arch mask image. Since the method of obtaining the dental arch mask image depends on the alveolar process associated region, which only includes the mandibular alveolar process region and at least part of the mandibular ramus region, this region is not affected by the field of view and tooth condition in the oral cavity. Therefore, the subsequent process of obtaining the dental arch curve does not need to be constrained by the field of view and tooth condition in the oral cavity. In this way, a stable and accurate dental arch mask image can be obtained in scenarios with local small and medium field of view CBCT images and complex tooth conditions (complete missing teeth, mixed dentition, metal artifacts, etc.), and a stable and accurate dental arch curve can be obtained based on the dental arch mask image.

[0066] It should be noted that, before or after obtaining the dental arch curve based on the dental arch mask image, the dental arch thickness can also be obtained based on the dental arch mask image, and after obtaining the dental arch curve, the corresponding panoramic oral radiograph image can be reconstructed based on the dental arch curve and the dental arch thickness.

[0067] In this embodiment, the dental arch thickness can be obtained from the distance value after performing distance transformation on the dental arch mask image in step S104 above. For example, the maximum distance transformation value represents the farthest value from the background on the dental arch skeleton line. Therefore, the dental arch thickness can be approximately twice the maximum distance transformation value. The dental arch thickness is denoted as D. Considering the different occlusal relationships of the upper and lower teeth and the different tilt angles of the radiograph, in order to ensure that the dental arch thickness can cover all teeth, the dental arch thickness is increased by a predetermined multiple (e.g., 1.5 times, i.e., the final dental arch thickness is 1.5*D).

[0068] In this embodiment, the panoramic oral radiograph image can be reconstructed using the following existing methods (e.g. Figure 9 As shown): The obtained dental arch curve is sampled at equal arc lengths to obtain N sampling points. The perpendicular line from the dental arch curve at each sampling point is then calculated. Equal-distance sampling is performed along this perpendicular line to both sides to obtain M sampling points, where M equals the dental arch thickness. The corresponding pixel values ​​are extracted from the oral cavity scan image (CBCT image, CT image, etc.) along these sampling points and unfolded to obtain M two-dimensional multiplanar reconstruction (MPR) images. These M images are then synthesized into a panoramic image. In this embodiment, the synthesis can be performed using the following formula:

[0069]

[0070] Among them, X m Let m represent the m-th MPR image, where m ranges from 1 to M, and Y represent the panoramic image. To further improve the image contrast of the synthesized panoramic image Y, this invention employs the Unsharp Mask (USM) sharpening algorithm, as shown in the following formula:

[0071] Z = Y + α(YG(Y))

[0072] Where G represents the Gaussian filter function, α represents the weighting coefficient (which can be set to 2), and Z represents the final panoramic dental image (e.g., Figure 9 (As shown).

[0073] Automatic reconstruction of panoramic images from oral scans is essential for image analysis and diagnosis. Currently, most software and products require manual drawing of dental arch curves to reconstruct panoramic images. A few software programs can automatically reconstruct panoramic images, but their application scenarios are limited and the results are unsatisfactory. Current automatic reconstruction methods for panoramic images are designed for simple scenarios with complete fields of view (i.e., the jawbone is basically intact in the oral scan image) and good tooth condition (i.e., no significant tooth loss). They cannot automatically reconstruct usable panoramic images for scenarios with small to medium field of view (CBCT images, CT images, etc.) or complex tooth conditions (complete missing teeth, mixed dentition, metal artifacts, etc.).

[0074] To address the aforementioned shortcomings of existing methods, the oral cavity scanning image processing method provided in this embodiment obtains the dental arch mask image by relying on the alveolar process associated region. This region includes the alveolar process region and at least part of the mandibular ramus region. This region is unaffected by the field of view or tooth condition within the oral cavity. Therefore, the process of obtaining the dental arch curve is not constrained by the field of view or tooth condition within the oral cavity. This method can obtain stable and accurate dental arch mask images in scenarios with small to medium field of view in oral cavity scanning images and complex tooth conditions (complete missing teeth, mixed dentition, metal artifacts, etc.). Based on this dental arch mask image, a stable and accurate dental arch curve is obtained. Subsequently, based on this dental arch curve, automatic panoramic reconstruction of CBCT images, CT images, and other oral cavity scanning images covering any scenario (including small to medium field of view CBCT images, CT images, and other oral cavity scanning images, and scenarios with complex tooth conditions (complete missing teeth, mixed dentition, metal artifacts, etc.)) can be achieved. Figure 10 (As shown).

[0075] The above embodiments provide a method for processing oral scan images. Correspondingly, another embodiment of this application also provides an oral scan image processing device. Since the device embodiment is basically similar to the method embodiment, it is described in a relatively simple way. For details of the relevant technical features, please refer to the corresponding description of the method embodiment provided above. The following description of the device embodiment is merely illustrative.

[0076] Please refer to Figure 11 Understanding this embodiment, Figure 11 This is a unit block diagram of the oral cavity scanning image processing device provided in this embodiment, such as... Figure 11 As shown, the apparatus provided in this embodiment includes:

[0077] Alveolar process associated region determination unit 201 is used to determine the alveolar process associated region based on the sagittal projection image corresponding to the oral cavity scan image. The alveolar process associated region includes the mandibular alveolar process region and at least part of the mandibular ramus region.

[0078] Axial projection image acquisition unit 202 is used to acquire an axial projection image corresponding to the alveolar process associated region in the oral cavity scan image;

[0079] The dental arch mask image acquisition unit 203 is used to segment the axial projection image to obtain a dental arch mask image;

[0080] The dental arch curve acquisition unit 204 is used to acquire the dental arch curve based on the dental arch mask image.

[0081] In some embodiments, obtaining the dental arch curve based on the dental arch mask image includes:

[0082] Perform a distance transformation on the dental arch mask image to obtain a distance-transformed image.

[0083] The distance-transformed image is segmented using a skeleton extraction model to obtain an initial dental arch skeleton image. The skeleton extraction model is a pre-trained deep neural network model that outputs the corresponding dental arch skeleton image based on the input distance-transformed image.

[0084] A thinning algorithm is used to perform secondary skeleton extraction on the initial dental arch skeleton image to obtain the target dental arch skeleton image;

[0085] The dental arch curve is obtained based on the target dental arch skeleton image.

[0086] In some embodiments, obtaining the dental arch curve based on the target dental arch skeleton image includes: processing the target dental arch skeleton image using cubic spline interpolation to obtain the dental arch curve.

[0087] In some embodiments, determining the alveolar process associated region based on the sagittal projection image corresponding to the oral cavity scan image includes: inputting the sagittal projection image corresponding to the oral cavity scan image into an alveolar process associated region extraction model to obtain a mask image of the alveolar process associated region output by the alveolar process associated region extraction model, wherein the alveolar process associated region extraction model is a pre-trained deep neural network model for outputting a mask image of the corresponding alveolar process associated region based on the input sagittal projection image.

[0088] In some embodiments, obtaining the axial projection image corresponding to the alveolar process associated region in the oral cavity scan image includes: performing maximum density projection on the region in the oral cavity scan image that is within the range of the alveolar process associated region according to the axial direction to obtain the axial projection image.

[0089] In some embodiments, the step of segmenting the axial projection image to obtain a dental arch mask image includes: inputting the axial projection image into a dental arch mask segmentation model to obtain the dental arch mask image output by the dental arch mask segmentation model, wherein the dental arch mask segmentation model is a pre-trained deep neural network model for outputting a corresponding dental arch mask image based on the input axial projection image.

[0090] In some embodiments, the apparatus further includes:

[0091] Dental arch thickness acquisition unit, used to obtain dental arch thickness based on the dental arch mask image;

[0092] A panoramic image reconstruction unit is used to reconstruct the panoramic oral radiograph corresponding to the oral cavity scan image based on the dental arch curve and the dental arch thickness.

[0093] The oral cavity scanning image processing device provided in this application determines the alveolar process associated region based on the sagittal projection image and obtains the axial projection image corresponding to the alveolar process associated region. The axial projection image is segmented to obtain a dental arch mask image. Since the method of obtaining the dental arch mask image depends on the alveolar process associated region, which only includes the alveolar process region and at least part of the mandibular ramus region, this region is not affected by the visual field and tooth condition in the oral cavity. Therefore, the subsequent process of obtaining the dental arch curve does not need to be constrained by the visual field and tooth condition in the oral cavity. In this way, a stable and accurate dental arch mask image can be obtained even in cases of severe tooth loss and severe metal artifacts, and a stable and accurate dental arch curve can be obtained based on the dental arch mask image.

[0094] In the above embodiments, a method and apparatus for processing oral cavity scan images are provided. Furthermore, another embodiment of this application provides an electronic device. Since the electronic device embodiment is basically similar to the method embodiment, it is described simply. For details of the relevant technical features, please refer to the corresponding descriptions of the method embodiments provided above. The following description of the electronic device embodiment is merely illustrative. The electronic device embodiment is as follows:

[0095] Please refer to Figure 12 To understand this embodiment, Figure 12 This is a schematic diagram of the electronic device provided in this embodiment.

[0096] like Figure 12 As shown, the electronic device provided in this embodiment includes: a processor 301 and a memory 302;

[0097] The memory 302 is used to store computer instructions for data processing. When these computer instructions are read and executed by the processor 301, they perform the following operations:

[0098] The alveolar process associated region is determined based on the sagittal projection image corresponding to the oral cavity scan image. The alveolar process associated region includes the alveolar process region and at least part of the mandibular ramus region.

[0099] Obtain the axial projection image corresponding to the alveolar process-associated region in the oral cavity scan image;

[0100] The axial projection image is segmented to obtain a dental arch mask image;

[0101] The dental arch curve is obtained based on the dental arch mask image.

[0102] In some embodiments, obtaining the dental arch curve based on the dental arch mask image includes:

[0103] Perform a distance transformation on the dental arch mask image to obtain a distance-transformed image.

[0104] The distance-transformed image is segmented using a skeleton extraction model to obtain an initial dental arch skeleton image. The skeleton extraction model is a pre-trained deep neural network model that outputs the corresponding dental arch skeleton image based on the input distance-transformed image.

[0105] A thinning algorithm is used to perform secondary skeleton extraction on the initial dental arch skeleton image to obtain the target dental arch skeleton image;

[0106] The dental arch curve is obtained based on the target dental arch skeleton image.

[0107] In some implementations, obtaining the dental arch curve based on the target dental arch skeleton image includes:

[0108] The target dental arch skeleton image is processed using cubic spline interpolation to obtain the dental arch curve.

[0109] In some embodiments, determining the alveolar process-related region based on the sagittal projection image corresponding to the oral cavity scan image includes:

[0110] The sagittal projection image corresponding to the oral cavity scan image is input into the alveolar process association region extraction model to obtain the mask image of the alveolar process association region output by the alveolar process association region extraction model. The alveolar process association region extraction model is a pre-trained deep neural network model for outputting the mask image of the corresponding alveolar process association region based on the input sagittal projection image.

[0111] In some embodiments, obtaining the axial projection image corresponding to the alveolar process-associated region in the oral cavity scan image includes:

[0112] The region within the alveolar process associated area of ​​the oral cavity scan image is subjected to maximum density projection along the axial direction to obtain the axial projection image.

[0113] In some embodiments, the step of performing image segmentation on the axial projection image to obtain a dental arch mask image includes:

[0114] The axial projection image is input into the dental arch mask segmentation model to obtain the dental arch mask image output by the dental arch mask segmentation model. The dental arch mask segmentation model is a pre-trained deep neural network model used to output the corresponding dental arch mask image based on the input axial projection image.

[0115] In some implementations, it also includes:

[0116] The dental arch thickness is obtained based on the dental arch mask image;

[0117] The panoramic oral radiograph corresponding to the oral cavity scan image is reconstructed based on the dental arch curve and the dental arch thickness.

[0118] By using the electronic device provided in this embodiment, the alveolar process associated region can be determined based on the sagittal projection image, and the corresponding axial projection image of the alveolar process associated region can be obtained. The axial projection image is then segmented to obtain a dental arch mask image. Since the method of obtaining the dental arch mask image depends on the alveolar process associated region, which only includes the alveolar process region and at least part of the mandibular ramus region, this region is not affected by the visual field and tooth condition in the oral cavity. Therefore, the subsequent process of obtaining the dental arch curve does not need to be constrained by the visual field and tooth condition in the oral cavity. In this way, a stable and accurate dental arch mask image can be obtained even in cases of severe tooth loss and severe metal artifacts, and a stable and accurate dental arch curve can be obtained based on the dental arch mask image.

[0119] In the above embodiments, an oral cavity scanning image processing method, an oral cavity scanning image processing apparatus, and an electronic device are provided. Furthermore, another embodiment of this application provides a computer-readable storage medium for implementing the above-described oral cavity scanning image processing method. The computer-readable storage medium embodiments provided in this application are described in a relatively simple manner; relevant parts can be found in the corresponding descriptions of the above method embodiments. The embodiments described below are merely illustrative.

[0120] The computer-readable storage medium provided in this embodiment stores computer instructions, which, when executed by a processor, perform the following steps:

[0121] The alveolar process associated region is determined based on the sagittal projection image corresponding to the oral cavity scan image. The alveolar process associated region includes the alveolar process region and at least part of the mandibular ramus region.

[0122] Obtain the axial projection image corresponding to the alveolar process-associated region in the oral cavity scan image;

[0123] The axial projection image is segmented to obtain a dental arch mask image;

[0124] The dental arch curve is obtained based on the dental arch mask image.

[0125] In some embodiments, obtaining the dental arch curve based on the dental arch mask image includes:

[0126] Perform a distance transformation on the dental arch mask image to obtain a distance-transformed image.

[0127] The distance-transformed image is segmented using a skeleton extraction model to obtain an initial dental arch skeleton image. The skeleton extraction model is a pre-trained deep neural network model that outputs the corresponding dental arch skeleton image based on the input distance-transformed image.

[0128] A thinning algorithm is used to perform secondary skeleton extraction on the initial dental arch skeleton image to obtain the target dental arch skeleton image;

[0129] The dental arch curve is obtained based on the target dental arch skeleton image.

[0130] In some implementations, obtaining the dental arch curve based on the target dental arch skeleton image includes:

[0131] The target dental arch skeleton image is processed using cubic spline interpolation to obtain the dental arch curve.

[0132] In some embodiments, determining the alveolar process-related region based on the sagittal projection image corresponding to the oral cavity scan image includes:

[0133] The sagittal projection image corresponding to the oral cavity scan image is input into the alveolar process association region extraction model to obtain the mask image of the alveolar process association region output by the alveolar process association region extraction model. The alveolar process association region extraction model is a pre-trained deep neural network model for outputting the mask image of the corresponding alveolar process association region based on the input sagittal projection image.

[0134] In some embodiments, obtaining the axial projection image corresponding to the alveolar process-associated region in the oral cavity scan image includes:

[0135] The region within the alveolar process associated area of ​​the oral cavity scan image is subjected to maximum density projection along the axial direction to obtain the axial projection image.

[0136] In some embodiments, the step of performing image segmentation on the axial projection image to obtain a dental arch mask image includes:

[0137] The axial projection image is input into the dental arch mask segmentation model to obtain the dental arch mask image output by the dental arch mask segmentation model. The dental arch mask segmentation model is a pre-trained deep neural network model used to output the corresponding dental arch mask image based on the input axial projection image.

[0138] In some implementations, it also includes:

[0139] The dental arch thickness is obtained based on the dental arch mask image;

[0140] The panoramic oral radiograph corresponding to the oral cavity scan image is reconstructed based on the dental arch curve and the dental arch thickness.

[0141] By executing the computer instructions stored on the computer-readable storage medium provided in this embodiment, the alveolar process associated region can be determined based on the sagittal projection image, and the axial projection image corresponding to the alveolar process associated region can be obtained. The axial projection image is then segmented to obtain a dental arch mask image. Since the method of obtaining the dental arch mask image depends on the alveolar process associated region, which only includes the alveolar process region and at least part of the mandibular ramus region, this region is not affected by the visual field and tooth condition in the oral cavity. Therefore, the subsequent process of obtaining the dental arch curve does not need to be constrained by the visual field and tooth condition in the oral cavity. In this way, a stable and accurate dental arch mask image can be obtained even in cases of severe tooth loss and severe metal artifacts, and a stable and accurate dental arch curve can be obtained based on the dental arch mask image.

[0142] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0143] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0144] 1. Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include non-transitory computer-readable media, such as modulated data signals and carrier waves.

[0145] 2. Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0146] Although this application discloses preferred embodiments as described above, it is not intended to limit this application. Any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of this application. Therefore, the scope of protection of this application should be determined by the scope defined in the claims of this application.

Claims

1. A method for processing oral cavity scan images, characterized in that, include: The alveolar process associated region is determined based on the sagittal projection image corresponding to the oral cavity scan image. The alveolar process associated region includes the mandibular alveolar process region and at least part of the mandibular ramus region. Obtain the axial projection image corresponding to the alveolar process-associated region in the oral cavity scan image; The axial projection image is segmented to obtain a dental arch mask image; Obtaining a dental arch curve based on the dental arch mask image includes: performing a distance transformation on the dental arch mask image to obtain a distance-transformed image; segmenting the distance-transformed image using a skeleton extraction model to obtain an initial dental arch skeleton image, wherein the skeleton extraction model is a pre-trained deep neural network model for outputting a corresponding dental arch skeleton image based on the input distance-transformed image; performing secondary skeleton extraction on the initial dental arch skeleton image using a thinning algorithm to obtain a target dental arch skeleton image; and obtaining a dental arch curve based on the target dental arch skeleton image.

2. The method according to claim 1, characterized in that, The process of obtaining the dental arch curve based on the target dental arch skeleton image includes: The target dental arch skeleton image is processed using cubic spline interpolation to obtain the dental arch curve.

3. The method according to claim 1, characterized in that, The determination of the alveolar process associated region based on the sagittal projection image corresponding to the oral cavity scan image includes: The sagittal projection image corresponding to the oral cavity scan image is input into the alveolar process association region extraction model to obtain the mask image of the alveolar process association region output by the alveolar process association region extraction model. The alveolar process association region extraction model is a pre-trained deep neural network model for outputting the mask image of the corresponding alveolar process association region based on the input sagittal projection image.

4. The method according to claim 1, characterized in that, Obtaining the axial projection image corresponding to the alveolar process associated region in the oral cavity scan image includes: The region within the alveolar process associated area of ​​the oral cavity scan image is subjected to maximum density projection along the axial direction to obtain the axial projection image.

5. The method according to claim 1, characterized in that, The step of segmenting the axial projection image to obtain the dental arch mask image includes: The axial projection image is input into the dental arch mask segmentation model to obtain the dental arch mask image output by the dental arch mask segmentation model. The dental arch mask segmentation model is a pre-trained deep neural network model used to output the corresponding dental arch mask image based on the input axial projection image.

6. The method according to claim 1, characterized in that, The method further includes: The dental arch thickness is obtained based on the dental arch mask image; The panoramic oral radiograph corresponding to the oral cavity scan image is reconstructed based on the dental arch curve and the dental arch thickness.

7. An oral cavity scanning image processing device, characterized in that, The device includes: The alveolar process associated region determination unit is used to determine the alveolar process associated region based on the sagittal projection image corresponding to the oral cavity scan image. The alveolar process associated region includes the mandibular alveolar process region and at least part of the mandibular ramus region. An axial projection image acquisition unit is used to acquire an axial projection image corresponding to the alveolar process associated region in the oral cavity scan image; The dental arch mask image acquisition unit is used to perform image segmentation on the axial projection image to obtain the dental arch mask image; The dental arch curve acquisition unit is used to obtain a dental arch curve based on the dental arch mask image, including: performing a distance transformation on the dental arch mask image to obtain a distance-transformed image; segmenting the distance-transformed image using a skeleton extraction model to obtain an initial dental arch skeleton image, wherein the skeleton extraction model is a pre-trained deep neural network model for outputting a corresponding dental arch skeleton image based on the input distance-transformed image; performing secondary skeleton extraction on the initial dental arch skeleton image using a thinning algorithm to obtain a target dental arch skeleton image; and obtaining the dental arch curve based on the target dental arch skeleton image.

8. An electronic device, characterized in that, Includes processor and memory; among which, The memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the method as described in any one of claims 1-6.

9. A computer-readable storage medium storing one or more computer instructions thereon, characterized in that, The instruction is executed by the processor to implement the method as described in any one of claims 1-6.

Images