Method, apparatus, electronic device, and storage medium for segmenting mandibular nerve canal

Abstract

An embodiment of the present invention discloses a method, apparatus, electronic device, and storage medium for segmenting the mandibular nerve canal. The method includes: obtaining an original oral image of a target object, preprocessing the original oral image to obtain a target oral image; determining a target mandible image of the target object according to the target oral image and a pre-established rough mandible segmentation model; determining a target mental foramen localization image of the target object according to the target mandible image and a pre-established mental foramen detection model; and determining a segmentation result of the mandibular nerve canal of the target object according to the target mental foramen localization image and a pre-established nerve canal segmentation model. The technical solution of the embodiment of the present invention can achieve automatic and accurate segmentation of the mandibular nerve canal, and improve the recognition efficiency of the mandibular nerve canal reduced due to external factors.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of image processing technology, and in particular, to a method, device, electronic device and storage medium for segmenting the mandibular nerve canal. Background Art

[0002] In recent years, with the aggravation of population aging, the continuous improvement of per capita disposable income, and the development of medical technology, the oral medical market in China has developed rapidly, and dental implant surgery has become more and more popular. In dental implant surgery, the position and course of the mandibular nerve canal are issues that must be noted in oral surgery tooth implantation. During the operation, it is necessary to avoid the mandibular nerve canal to prevent damage to the mandibular nerve, which may lead to problems such as mandibular numbness.

[0003] Currently, in the existing technology, the positioning of the mandibular nerve canal completely depends on the judgment of dentists, which requires certain professional knowledge and experience. The manual recognition efficiency is low and there is a dependence on manual experience. Summary of the Invention

[0004] The embodiments of the present invention provide a method, device, electronic device and storage medium for segmenting the mandibular nerve canal to achieve automatic segmentation of the mandibular nerve canal.

[0005] In a first aspect, the embodiments of the present invention provide a method for segmenting the mandibular nerve canal, the method comprising:

[0006] Obtaining an original oral image of a target object, and preprocessing the original oral image to obtain a target oral image;

[0007] Determining a target mandibular bone image of the target object according to the target oral image and a pre-established rough mandibular bone segmentation model;

[0008] Determining a target mental foramen positioning image of the target object according to the target mandibular bone image and a pre-established mental foramen detection model;

[0009] Determining a segmentation result of the mandibular nerve canal of the target object according to the target mental foramen positioning image and a pre-established nerve canal segmentation model.

[0010] In a second aspect, the embodiments of the present invention further provide a device for segmenting the mandibular nerve canal, the device comprising:

[0011] A preprocessing module, configured to obtain an original oral image of a target object, and preprocess the original oral image to obtain a target oral image;

[0012] A segmentation image determination module, configured to determine a target mandibular bone image of the target object according to the target oral image and a pre-established rough mandibular bone segmentation model;

[0013] A positioning image determination module, configured to determine a target mental foramen positioning image of the target object according to the target mandible image and a pre-established mental foramen detection model;

[0014] A segmentation result determination module, configured to determine a segmentation result of the mandibular nerve canal of the target object according to the target mental foramen positioning image and a pre-established nerve canal segmentation model.

[0015] In a third aspect, an embodiment of the present invention further provides an electronic device, which includes:

[0016] One or more processors;

[0017] A storage device, configured to store one or more programs,

[0018] When the one or more programs are executed by the one or more processors, the one or more processors implement the segmentation method of the mandibular nerve canal provided in any embodiment of the present invention.

[0019] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the segmentation method of the mandibular nerve canal provided in any embodiment of the present invention is implemented.

[0020] The technical solution of the embodiment of the present invention obtains a preprocessed target oral cavity image, inputs the target oral cavity image into a pre-established mandible rough segmentation model to obtain a target mandible image, and inputs it into a pre-established mental foramen detection model to obtain a target mental foramen positioning image. Further, the target mental foramen positioning image is input into a pre-established nerve canal segmentation model, and finally, a segmentation result of the mandibular nerve canal of the target object is obtained, solving the problems in the prior art such as difficult recognition of the mandibular nerve canal and high dependence on manual experience, realizing automatic and accurate segmentation of the mandibular nerve canal, and effectively shortening the segmentation time and improving the segmentation accuracy by segmenting the mandibular nerve canal based on a deep learning segmentation network. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the introduced drawings are only the drawings of a part of the embodiments to be described in the present invention, rather than all the drawings. For those of ordinary skill in the art, other drawings can be obtained according to these drawings without creative efforts.

[0022] Figure 1 It is a schematic flowchart of a method for segmenting a mandibular nerve canal provided in Embodiment 1 of the present invention;

[0023] Figure 2Schematic flowchart of a method for segmenting the mandibular nerve canal provided in the second embodiment of the present invention;

[0024] Figure 3 Schematic flowchart of a method for segmenting the mandibular nerve canal in an application scenario provided in the second embodiment of the present invention;

[0025] Figure 4 Schematic structural diagram of a device for segmenting the mandibular nerve canal provided in the third embodiment of the present invention;

[0026] Figure 5 Schematic structural diagram of an electronic device provided in the fourth embodiment of the present invention. Detailed implementation manners

[0027] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It can be understood that the specific embodiments described herein are only used to explain the present invention, rather than limiting the present invention. Additionally, it should be noted that for the sake of description, only parts related to the present invention are shown in the drawings rather than all structures.

[0028] Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be performed in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when its operations are completed, but it can also have additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, and so on.

[0029] Embodiment 1

[0030] Figure 1 Schematic flowchart of a method for segmenting the mandibular nerve canal provided in the first embodiment of the present invention. This embodiment is applicable to the situation of image positioning and segmentation. This method can be executed by a device for segmenting the mandibular nerve canal, and this device can be implemented through software and / or hardware, and can be configured in a terminal and / or a server to implement the method for segmenting the mandibular nerve canal in the embodiments of the present invention.

[0031] As Figure 1 shown, the method of this embodiment can specifically include:

[0032] S110. Obtain the original oral image of the target object, and preprocess the original oral image to obtain the target oral image.

[0033] Among them, the target object can be understood as the object for which the mandibular nerve canal needs to be located at the current moment. Exemplarily, the target object can be a human or an animal, and this embodiment does not limit this. The oral image of the target object can refer to an image reflecting the anatomical morphology and structure of the teeth and root canal system in the oral cavity of the target object. In the embodiments of the present invention, the original oral image can be understood as the unprocessed oral image obtained, or in other words, the oral image before processing. Exemplarily, the original oral image can be an oral Computed Tomography (CT) image, an oral Cone beam Computer Tomography (CBCT) image, or an oral Magnetic Resonance Imaging (MRI) image, and this embodiment does not limit this. For example, if the original oral image is an oral CT image, it can reflect the tissue conditions from a three-dimensional perspective, and can detect lesions that cannot be detected by the projection angle of oral X-ray films or more subtle lesions, and can accurately evaluate the bone tissue conditions and mandibular joint conditions. It should be noted that the original oral image can be obtained in real time from a medical imaging device, from an image database, or received from an external device for oral data transmission, and this embodiment does not limit this.

[0034] Among them, the preprocessing of the original oral image can be processes such as resampling and normalization of the original oral image. Exemplarily, the resampling process can be understood as the process of sampling the image from the original resolution to other resolutions. The normalization process refers to the process of performing a series of standard processing transformations on the image to transform it into a fixed standard form. For example, by selecting appropriate upper and lower thresholds, the original oral image is normalized to an image between 0 and 1 or between -1 and 1. It can be understood that the target oral image can refer to the oral image obtained after the original oral image undergoes preprocessing, which can facilitate the subsequent operations. For example, the target oral image can be an oral image that meets the segmentation requirements of a pre-established rough mandible segmentation model after preprocessing. Optionally, the preprocessing of the original oral image can also include image cropping and / or filtering, etc., and the specific image preprocessing method can be set according to actual needs and is not limited here.

[0035] Specifically, after the image processing terminal device obtains the original oral image of the target object, a series of preprocessing such as resampling and normalization is performed on the original oral image to obtain the target oral image, so that the automatic segmentation operation of the mandibular nerve canal can be continued based on the target oral image.

[0036] S120. Determine the target mandible image of the target object according to the target oral image and the pre-established rough mandible segmentation model.

[0037] Among them, the pre-established rough mandible segmentation model can be a neural network model that has been trained and is used for rough segmentation of the oral mandible. Specifically, the rough mandible segmentation model can be implemented by training a pre-established rough mandible segmentation model to be trained. When training the rough mandible segmentation model, first, obtain the original oral sample image dataset for training, and the mandible annotation image obtained by annotating the mandible position in the original oral sample image. Among them, the mandible annotation image can be understood as a representation of the ground truth image and can be used as a basis for evaluating subsequent prediction results; second, preprocess the original oral sample image dataset to obtain an input sample image dataset that can be input into the rough mandible segmentation model to be trained; then, input the input sample image dataset into the rough mandible segmentation model to be trained to obtain an initial training result; adjust the parameters of the rough mandible segmentation model based on the loss function generated by the preset mandible annotation image and the initial training result until the training end condition is met to obtain a trained rough mandible segmentation model.

[0038] It should be noted that the image preprocessing can include operations such as resampling and normalization. For example, the resolution of the original oral image is 0.25mm×0.25mm×0.25mm, and it is resampled to an oral image with a resolution of 1.2mm×1.2mm×1.2mm, then normalized, and input into the rough mandible segmentation model to be trained.

[0039] Exemplarily, the rough mandible segmentation model can be composed of at least one of a fully convolutional neural network, a recurrent neural network, or a deep neural network structure, etc. The loss function can be at least one of a cross-entropy loss function, a mean squared error loss function, or a Dice loss function, etc. This embodiment does not limit the specific structure of the rough mandible segmentation model and the loss function applied. For example, the rough mandible segmentation model is a deep convolutional neural network model established based on the V-net network, and the loss function adopts the weighted sum of the cross-entropy loss function and the Dice coefficient loss function.

[0040] Among them, the target mandible image can be understood as an image that can represent the mandible position in the target oral image after being segmented by the rough mandible segmentation model. It should be noted that the target mandible image can be obtained by cropping the target oral image according to the output result of the rough mandible segmentation model, or the mandible position in the target oral image can be marked as the region of interest according to the output result of the rough mandible segmentation model, and the region of interest is displayed differently. This embodiment does not limit this.

[0041] Specifically, the preprocessed target oral image is input into a pre-established rough mandible segmentation model to obtain a target mandible image. The purpose of this is, on the one hand, to reduce the size of the input image in subsequent operations, so as to achieve the effects of reducing the video memory occupancy and the running time of subsequent algorithms; on the other hand, it can roughly locate the position of the mental nerve canal in the target mandible image, which is beneficial to the accurate segmentation of the mental nerve canal in the subsequent steps.

[0042] S130. Determine the target mental foramen localization image of the target object according to the target mandible image and the pre-established mental foramen detection model.

[0043] Among them, the pre-established mental foramen detection model can be understood as a neural network model that has been trained and is used to locate the position of the mental foramen. Specifically, the establishment of the mental foramen detection model can be achieved by training the mental foramen detection model to be trained. When training the mental foramen detection model, first, obtain the original oral sample image dataset for training, and the mental foramen annotation coordinate information obtained by annotating the position of the mental foramen in the original oral sample image; secondly, preprocess the original oral sample image dataset to obtain an input sample image dataset that can be used to input into the mental foramen detection model to be trained; then, input the input sample image dataset into the mental foramen detection model to obtain an initial training result; adjust the parameters of the mental foramen detection model based on the loss function generated by the preset mental foramen annotation position and the initial training result until the training end condition is met, and obtain the trained mental foramen detection model. Similarly, the image preprocessing can include operations such as resampling and normalization. For example, the resolution of the original oral image is 0.25mm×0.25mm×0.25mm, and it is resampled to an oral image with a resolution of 1.0mm×1.0mm×1.0mm, then normalized, and input into the mental foramen detection model for training.

[0044] Exemplarily, the mental foramen detection model can be composed of at least one of a fully convolutional neural network, a recurrent neural network, or a deep neural network structure, etc. The loss function can be at least one of a cross-entropy loss function, a mean square error loss function, or a Dice loss function, etc. This embodiment does not limit the structure of the mental foramen detection model and the composition of the loss function. In this embodiment, optionally, the mental foramen detection model includes a Gaussian heatmap regression model, and the loss function is a weighted Adaptive Wing loss.

[0045] Among them, the Gaussian heatmap regression model can be understood as a fully convolutional regression model applied to image key point detection. Through the Gaussian heatmap, a large amount of data can be simply aggregated and represented using a gradient color band. The final effect is generally better than the direct representation of discrete points, and it can intuitively show the density or frequency of spatial data. WeightedAdaptive Wing loss is a loss function used when training the Gaussian heatmap regression model.

[0046] Among them, the target mental foramen localization image can be understood as the image information used to display the position of the mental foramen in the mandibular bone image. Optionally, the target mental foramen localization image may include the left mental foramen localization image and the right mental foramen localization image. It should be noted that the target mental foramen localization image can be obtained by cropping the target mandibular bone image according to the output result of the mental foramen detection model, or the position of the mental foramen in the target mandibular bone image can be used as the region of interest according to the output result of the mental foramen detection model and displayed differently, etc. This embodiment does not make a limitation on this.

[0047] Specifically, the target mandibular bone segmentation image is input into the mental foramen detection model, and the mental foramen detection model is used to locate the position of the mental foramen in the target mandibular bone segmentation image, and the left mental foramen localization image and the right mental foramen localization image are output, so that the left and right neural canals can be located according to the left and right mental foramen position information.

[0048] S140. Determine the segmentation result of the mandibular nerve canal of the target object according to the target mental foramen localization image and the pre-established nerve canal segmentation model.

[0049] Among them, the pre-established nerve canal segmentation model can be understood as a neural network model that has been trained and can be used to segment the nerve canal. Specifically, the establishment of the nerve canal segmentation model can be achieved by training the nerve canal segmentation model to be trained. When training the nerve canal segmentation model, first, obtain the original oral sample image dataset for training, and the mandibular nerve canal annotation image obtained by annotating the position of the mandibular nerve canal in the original oral sample image; secondly, preprocess the original oral sample image dataset to obtain the input sample image dataset; then, input the input sample image dataset into the nerve canal segmentation model to be trained to obtain the initial training result; adjust the parameters of the nerve canal segmentation model based on the loss function generated by the preset mandibular nerve canal annotation image and the initial training result until the training end condition is met to obtain the trained nerve canal segmentation model. Similarly, image preprocessing can include operations such as resampling and normalization. For example, the resolution of the original oral image is 0.25mm×0.25mm×0.25mm, and it is resampled to an oral image with a resolution of 0.3mm×0.3mm×0.3mm, then normalized, and input into the nerve canal segmentation model to be trained.

[0050] Exemplarily, the neural tube segmentation model can be composed of at least one of a fully convolutional neural network, a recurrent neural network, or a deep neural network structure, etc. The loss function can be at least one of a cross-entropy loss function, a mean squared error loss function, or a Dice loss function, etc. This embodiment does not limit the specific structure of the neural tube segmentation model and the loss function applied. For example, the neural tube segmentation model is a fully convolutional neural network model established based on the V-net network, and the loss function is a weighted sum of a cross-entropy loss function, a Dice loss function, and a clDice loss function.

[0051] Among them, the segmentation result of the mandibular neural tube can be understood as an image or data information used to represent the specific position and contour information of the mandibular neural tube in the oral cavity image. It should be noted that the segmentation result of the mandibular neural tube can be an image or data information that can reflect the position or contour information of the mandibular neural tube, or it can be to mark the mandibular neural tube part as the region of interest and display it differently from the non-region of interest. This embodiment does not limit this.

[0052] Specifically, input the target mental foramen localization image of the target object into the pre-established neural tube segmentation model, then the segmentation result of the mandibular neural tube of the target object can be output, and further, the specific position and contour information of the mandibular neural tube of the target object can be obtained according to the segmentation result.

[0053] It should be noted that when training the above three neural network models, in order to prevent the training sample image data from causing overfitting of the subsequent three neural network models, data augmentation processing can be performed on the oral sample images to expand the number of input sample images and enhance the generalization function of the neural network models. Among them, the data augmentation processing methods can include random scaling, random elastic deformation, translation, rotation, grayscale histogram adjustment, and mirror data, etc.

[0054] It should also be noted that the training processes of the above three models can be independent of each other, can be carried out simultaneously, or can be carried out sequentially. This embodiment does not limit this.

[0055] In the technical solution of the embodiment of the present invention, by obtaining a preprocessed target oral image, inputting the target oral image into a pre-established rough mandible segmentation model to obtain a target mandible image, and inputting it into a pre-established mental foramen detection model to obtain a target mental foramen localization image. Further, inputting the target mental foramen localization image into a pre-established nerve canal segmentation model, finally obtaining the segmentation result of the mandibular nerve canal of the target object, which solves the problems of difficult recognition of the mandibular nerve canal and high dependence on manual experience in the prior art, realizes the full-automatic and accurate segmentation of the mandibular nerve canal, and effectively shortens the segmentation time and improves the segmentation accuracy based on the deep learning segmentation network for segmenting the mandibular nerve canal.

[0056] Embodiment 2

[0057] Figure 2 FIG. is a schematic flowchart of a method for segmenting the mandibular nerve canal provided in Embodiment 2 of the present invention. On the basis of the above technical solution, the technical solution of this embodiment can be further refined. Optionally, determining the target mandible image of the target object according to the target oral image and the pre-established rough mandible segmentation model includes: inputting the target oral image into the established rough mandible segmentation model to obtain a mandible mask image; cropping the target oral image based on the mandible mask image to obtain a target mandible image.

[0058] On the basis of the above optional technical solutions, further, the target mental foramen localization image may include a left mental foramen localization image and a right mental foramen localization image; determining the target mental foramen localization image of the target object according to the target mandible image and the pre-established mental foramen detection model may include: inputting the target mandible image into the pre-established mental foramen detection model to obtain a preliminary mental foramen localization image, where the preliminary mental foramen localization image includes the first mental foramen position information of the left mental foramen and the second mental foramen position information of the right mental foramen; cropping the target mandible image according to the first mental foramen position information and the second mental foramen position information to obtain a left mental foramen localization image corresponding to the left mandibular nerve canal of the target object and a right mental foramen localization image corresponding to the right mandibular nerve canal of the target object respectively.

[0059] On the basis of the above optional technical solutions, optionally, determining the segmentation result of the mandibular nerve canal of the target object according to the target mental foramen localization image and the pre-established nerve canal segmentation model includes: determining the left nerve canal segmentation image and the right nerve canal segmentation image of the target object according to the left mental foramen localization image, the right mental foramen localization image and the pre-established nerve canal segmentation model; determining the segmentation result of the mandibular nerve canal of the target object according to the left nerve canal segmentation image and the right nerve canal segmentation image.

[0060] For the specific implementation manners of the above technical solutions, reference may be made to the detailed description of this embodiment. Among them, the same or similar technical terms as those in the above embodiment will not be elaborated again.

[0061] See Figure 2 As shown, the method of the embodiment of the present invention may specifically include the following steps:

[0062] S210. Obtain the original oral cavity image of the target object, and preprocess the original oral cavity image to obtain the target oral cavity image.

[0063] S220. Input the target oral cavity image into the established rough mandible segmentation model to obtain a mandible mask image.

[0064] Among them, the mandible mask image can be understood as a filter template for identifying the position and contour of the mandible in the oral cavity image, and can screen out the mandible part in the oral cavity image by blocking other parts of the oral cavity image. Exemplarily, the mandible mask image may include key point information about the position of the mandible, or may include the contour information of the mandible, or may include mandible position identification data information. This embodiment does not make any limitation thereto.

[0065] Among them, the size of the mandible mask image obtained by inputting the target oral cavity image into the established rough mandible segmentation model is often the same as the size of the target oral cavity image.

[0066] S230. Crop the target oral cavity image based on the mandible mask image to obtain a target mandible image.

[0067] In specific implementation, input the preprocessed target oral cavity image into the rough mandible segmentation model, and a mandible mask image can be obtained. According to the mandible mask image, the border closest to the outer part of the mandible can be determined. Furthermore, crop the target oral cavity image according to the determined outer border of the mandible, and a target mandible image that can reflect the position and contour information of the mandible can be obtained, eliminating the background data in the target oral cavity image, reducing the interference of the background data, and reducing the calculation amount caused by the background data.

[0068] S240. Input the target mandible image into the pre-established mental foramen detection model to obtain a preliminary mental foramen localization image.

[0069] Among them, the preliminary mental foramen localization image includes the first mental foramen position information of the left mental foramen and the second mental foramen position information of the right mental foramen.

[0070] Specifically, the target mandible image is input into the mental foramen detection model, and the mental foramen detection model can mark the coordinates or key point information of the mental foramina on the left and right sides, so as to obtain a preliminary mental foramen positioning image containing the position information of the mental foramina on both sides, so that the target mandible image can be processed according to the mental foramen coordinate information in the preliminary mental foramen positioning image in the subsequent process.

[0071] It should be noted that "first" and "second" are only used to distinguish the left and right sides, and do not represent the serial number of arrangement, nor can they be understood as indicating or implying relative importance.

[0072] S250. Crop the target mandible image according to the first mental foramen position information and the second mental foramen position information, and respectively obtain a left mental foramen positioning image corresponding to the left mandibular nerve canal of the target object and a right mental foramen positioning image corresponding to the right mandibular nerve canal.

[0073] In a specific implementation, the target mandible image is cropped according to the first mental foramen position information and the second mental foramen position information obtained, which includes the left and right mental foramen coordinate information. The target mandible image is respectively cropped into a left mental foramen image corresponding to the left mandibular nerve canal of the target object and a right mental foramen positioning image corresponding to the right mandibular nerve canal through the determined mental foramen coordinate information, that is, the target mandible image is cropped into two mental foramen positioning images corresponding to the mandibular nerve canals of the target object. Thus, the exact position of the mandibular nerve canal in the target image can be further approximated, the interference of background data can be eliminated, and the calculation amount caused by background data can be reduced.

[0074] S260. Determine the left nerve canal segmentation image and the right nerve canal segmentation image of the target object according to the left mental foramen positioning image, the right mental foramen positioning image and the pre-established nerve canal segmentation model.

[0075] Optionally, the left mental foramen positioning image is input into the pre-established nerve canal segmentation model to obtain the left nerve canal segmentation image of the target object; the right mental foramen positioning image is input into the pre-established nerve canal segmentation model to obtain the right nerve canal segmentation image of the target object.

[0076] Among them, the left nerve canal segmentation image can be understood as an image containing the position or contour information of the left mandibular nerve canal of the target object, and the right nerve canal segmentation image can be understood as an image containing the position or contour information of the right mandibular nerve canal of the target object. Specifically, inputting the left mental foramen positioning image into the nerve canal segmentation model can obtain the left nerve canal image including the area where the left nerve canal is located. Similarly, inputting the right mental foramen positioning image into the nerve canal segmentation model can obtain the right nerve canal image including the area where the right nerve canal is located.

[0077] In this embodiment, optionally, the neural tube segmentation model includes a cascaded network. Among them, the input parameters of the subsequent network in the cascaded network are the output parameters of the previous network adjacent to the subsequent network and the input parameters of the first-level network. For example, the input parameters and output parameters of the first-level network are used as two-channel input parameters and input into the second-level network, and then the input parameters of the first-level network and the output parameters of the second-level network are used as two-channel input parameters and input into the third-level network, and so on. In each level of network, the subsequent network is not only related to the output of the previous network, but also takes into account the input of the first-level network, reducing the occurrence probability of mandibular neural tube fracture and improving the robustness of the model. Thus, the complete segmentation of the mandibular neural tube is ensured.

[0078] In this embodiment, optionally, the loss function of the neural tube segmentation model includes clDice loss. Among them, clDice loss can be understood as a topology-preserving loss function for tubular structure segmentation.

[0079] It should be noted that the cascaded network included in the neural tube segmentation model and the clDice loss included in the loss function of the neural tube segmentation model can both be used to maintain the consistency of the topology of the mandibular neural tube and can reduce the occurrence probability of neural tube fracture to a certain extent.

[0080] S270. Determine the segmentation result of the mandibular neural tube of the target object according to the left neural tube segmentation image and the right neural tube segmentation image.

[0081] Optionally, the left neural tube segmentation image, the right neural tube segmentation image and the original oral image are fused to obtain the segmentation result of the mandibular neural tube of the target object.

[0082] Among them, fusing the left neural tube segmentation image, the right neural tube segmentation image and the original oral image can be to mark them in the original oral image according to the position or contour information of the two neural tubes included in the left neural tube segmentation image and the right neural tube segmentation image, or to resample the left neural tube segmentation image and the right neural tube segmentation image to the resolution of the original oral image and merge the resampled left neural tube segmentation image and the right neural tube segmentation image. Of course, there can also be other image fusion methods, which are not limited in this embodiment.

[0083] Specifically, the left mental foramen localization image and the right mental foramen localization image are respectively input into the neural tube segmentation model, and then the left neural tube segmentation image containing the position or contour information of the left neural tube and the right neural tube segmentation image containing the position or contour information of the right neural tube can be obtained. Furthermore, fusing the two neural tube segmentation images output by the neural tube segmentation model on the basis of the original oral image can obtain the final segmentation result of the mandibular neural tube of the target object.

[0084] To clearly introduce the specific implementation of this embodiment, it can be illustrated with a specific example. For example, in the first step, an original oral image is obtained; in the second step, the original oral image is preprocessed to obtain a target oral image; in the third step, the target oral image is input into the rough mandible segmentation model to obtain a target mandible image; in the fourth step, the target mandible image is input into the mental foramen detection model to obtain a left mental foramen localization image and a right mental foramen localization image; in the fifth step, the left mental foramen localization image and the right mental foramen localization image are respectively input into the nerve canal segmentation model to obtain a left nerve canal segmentation image and a right nerve canal segmentation image; in the sixth step, the left nerve canal segmentation image and the right nerve canal segmentation image are integrated into the original oral image to obtain the final mandibular nerve canal segmentation result.

[0085] The specific implementation process of the above example can be referred to Figure 3 as shown in the process schematic diagram.

[0086] The technical solution of this embodiment obtains a preprocessed target oral image, inputs it into the rough mandible segmentation model to obtain a mandible mask image, crops the target oral image according to the mandible mask image to obtain a target mandible image. Further, the target mandible image is input into the mental foramen detection model to obtain a preliminary mental foramen localization image, and the target mandible image is cropped according to the position information of the left and right mental foramens to obtain a left mental foramen localization image and a right mental foramen localization image. Then, the left and right mental foramen localization images are respectively input into the nerve canal segmentation model to obtain left and right nerve canal segmentation results, and the left and right nerve canal segmentation results are integrated to obtain the final mandibular nerve canal segmentation result, which solves the problems of low recognition accuracy of the mandibular nerve canal and high dependence on manual experience in the prior art, realizes the automatic and accurate segmentation of the mandibular nerve canal, and effectively improves the recognition efficiency of the mandibular nerve canal.

[0087] Embodiment III

[0088] Figure 4 FIG. is a schematic structural diagram of a mandibular nerve canal segmentation device provided in Embodiment III of the present invention. The device specifically includes: a preprocessing module 310, a segmentation image determination module 320, a localization image determination module 330, and a segmentation result determination module 340.

[0089] Among them, the preprocessing module 310 is configured to obtain an original oral image of a target object, and preprocess the original oral image to obtain a target oral image;

[0090] The segmentation image determination module 320 is configured to determine a target mandible image of the target object according to the target oral image and a pre-established rough mandible segmentation model;

[0091] A positioning image determination module 330, configured to determine a target mental foramen positioning image of a target object according to a target mandible image and a pre-established mental foramen detection model;

[0092] A segmentation result determination module 340, configured to determine a segmentation result of the mandibular nerve canal of the target object according to the target mental foramen positioning image and a pre-established nerve canal segmentation model.

[0093] The technical solution of the embodiment of the present invention obtains a preprocessed target oral cavity image, inputs the target oral cavity image into a pre-established mandible rough segmentation model to obtain a target mandible image, and inputs it into a pre-established mental foramen detection model to obtain a target mental foramen positioning image. Further, the target mental foramen positioning image is input into a pre-established nerve canal segmentation model, and finally, a segmentation result of the mandibular nerve canal of the target object is obtained, solving the problems of difficult recognition of the mandibular nerve canal and high dependence on manual experience in the prior art, realizing automatic and accurate segmentation of the mandibular nerve canal, and effectively shortening the segmentation time and improving the segmentation accuracy by segmenting the mandibular nerve canal based on a deep learning segmentation network.

[0094] Optionally, the segmentation image determination module 320 is further configured to input the target oral cavity image into the established mandible rough segmentation model to obtain a mandible mask image; and crop the target oral cavity image based on the mandible mask image to obtain a target mandible image.

[0095] Optionally, the target mental foramen positioning image includes a left mental foramen positioning image and a right mental foramen positioning image;

[0096] The positioning image determination module 330 is further configured to input the target mandible image into the pre-established mental foramen detection model to obtain a preliminary mental foramen positioning image, where the preliminary mental foramen positioning image includes first mental foramen position information of the left mental foramen and second mental foramen position information of the right mental foramen; and crop the target mandible image according to the first mental foramen position information and the second mental foramen position information to respectively obtain a left mental foramen positioning image corresponding to the left mandibular nerve canal of the target object and a right mental foramen positioning image corresponding to the right mandibular nerve canal of the target object.

[0097] Optionally, the mental foramen detection model includes a Gaussian heatmap regression model, and the loss function of the mental foramen detection model is weighted Adaptive Wing loss.

[0098] Optionally, the segmentation result determination module 340 further includes an image determination unit and a segmentation result determination unit.

[0099] Among them, an image determination unit is configured to determine a left neural tube segmentation image and a right neural tube segmentation image of a target object according to a left mental foramen localization image, a right mental foramen localization image, and a pre-established neural tube segmentation model; a segmentation result determination unit is configured to determine a segmentation result of the mandibular neural tube of the target object according to the left neural tube segmentation image and the right neural tube segmentation image.

[0100] Optionally, the image determination unit is further configured to input the left mental foramen localization image into the pre-established neural tube segmentation model to obtain a left neural tube segmentation image of the target object; input the right mental foramen localization image into the pre-established neural tube segmentation model to obtain a right neural tube segmentation image of the target object.

[0101] Optionally, the segmentation result determination unit is further configured to perform image fusion on the left neural tube segmentation image, the right neural tube segmentation image, and the original oral cavity image to obtain a segmentation result of the mandibular neural tube of the target object.

[0102] The above-mentioned mandibular neural tube segmentation device can execute the mandibular neural tube segmentation method provided by any embodiment of the present invention, and has corresponding functional modules and beneficial effects for executing the method.

[0103] It should be noted that the various units and modules included in the above-mentioned image recognition device are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of the functional units are only for the convenience of mutual distinction and do not limit the protection scope of the embodiments of the present invention.

[0104] Embodiment IV

[0105] Figure 5 It is a schematic structural diagram of an electronic device provided by Embodiment IV of the present invention. Figure 5 A block diagram of an exemplary electronic device 40 suitable for implementing the embodiments of the present invention is shown. Figure 5 The displayed electronic device 40 is only an example and should not bring any limitation to the functions and usage scope of the embodiments of the present invention.

[0106] As Figure 5 shown, the electronic device 40 is presented in the form of a general-purpose computing device. The components of the electronic device 40 may include, but are not limited to: one or more processors or processing units 401, a system memory 402, and a bus 403 connecting different system components (including the system memory 402 and the processing unit 401).

[0107] The bus 403 represents one or more of several types of bus architectures, including a memory bus or memory controller, a peripheral bus, an Accelerated Graphics Port, a processor bus, or a local bus using any of the several bus architectures. By way of example, and not limitation, these architectures include the Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MAC) bus, Enhanced ISA bus, Video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus.

[0108] The electronic device 40 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device 40, including both volatile and nonvolatile media, removable and non-removable media.

[0109] System memory 402 can include computer system readable media in the form of volatile memory, such as random access memory (RAM) 404 and / or cache memory 405. The electronic device 40 may further include other removable / non-removable, volatile / nonvolatile computer system storage media. By way of example only, storage system 406 can be used for reading and writing non-removable, nonvolatile magnetic media ( Figure 5 not shown and typically called a "hard disk drive"). Although Figure 5 not shown in the figures, a disk drive for reading and writing a removable nonvolatile magnetic disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing a removable, nonvolatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) can be provided. In these instances, each drive can be connected to the bus 403 by one or more data media interfaces. Memory 402 can include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of the embodiments of the present invention.

[0110] A program / utility 408 having a set (at least one) of program modules 407 can be stored, for example, in memory 402, such program modules 407 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which examples or some combination thereof may include an implementation of a network environment. The program modules 407 generally carry out the functions and / or methods of the embodiments described herein.

[0111] The electronic device 40 can also communicate with one or more external devices 409 (such as a keyboard, a pointing device, a display 410, etc.), and can also communicate with one or more devices that enable a user to interact with the electronic device 40, and / or communicate with any device that enables the electronic device 40 to communicate with one or more other computing devices (such as a network card, a modem, etc.). Such communication can be carried out through the input / output (I / O) interface 411. Moreover, the electronic device 40 can also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through the network adapter 412. As shown in the figure, the network adapter 412 communicates with other modules of the electronic device 40 through the bus 403. It should be understood that although Figure 5 not shown in the figure, other hardware and / or software modules can be used in combination with the electronic device 40, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0112] The processing unit 401 executes various functional applications and data processing by running programs stored in the system memory 402, for example, implementing the method for segmenting the mandibular nerve canal provided by the embodiments of the present invention.

[0113] Embodiment Five

[0114] Embodiment Five of the present invention also provides a storage medium containing computer-executable instructions, and the computer-executable instructions are used to execute a method for segmenting the mandibular nerve canal when executed by a computer processor. The method includes:

[0115] Obtain the original oral image of the target object, and perform preprocessing on the original oral image to obtain the target oral image;

[0116] Determine the target mandibular bone segmentation image of the target object according to the target oral image and the pre-established mandibular bone rough segmentation model;

[0117] Determine the target mental foramen positioning image of the target object according to the target mandibular bone image and the pre-established mental foramen detection model;

[0118] Determine the segmentation result of the mandibular nerve canal of the target object according to the target mental foramen positioning image and the pre-established nerve canal segmentation model.

[0119] The computer storage medium of an embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, the computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device.

[0120] The computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, in which computer-readable program code is carried. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, and this computer-readable medium can send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device.

[0121] The program code contained on the computer-readable medium can be transmitted by any suitable medium, including but not limited to wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0122] The computer program code for performing the operations of the embodiment of the present invention can be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, by using an Internet service provider to connect through the Internet).

[0123] Note that the above is only the preferred embodiment of the present invention and the technical principles applied. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein. Various obvious changes, re-adjustments, and substitutions can be made by those skilled in the art without departing from the protection scope of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments only. Without departing from the concept of the present invention, more other equivalent embodiments can be included, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for segmenting the mandibular nerve canal, characterized in that, Including: Obtain the original oral image of the target object, preprocess the original oral image to obtain the target oral image; Determine the target mandible image of the target object according to the target oral image and the pre-established rough mandible segmentation model; Determine the target mental foramen localization image of the target object according to the target mandible image and the pre-established mental foramen detection model; Determine the segmentation result of the mandibular nerve canal of the target object according to the target mental foramen localization image and the pre-established nerve canal segmentation model; Wherein, the rough mandible segmentation model is a neural network model that has been trained and is used for rough segmentation of the oral mandible; the rough mandible segmentation model is composed of at least one of a fully convolutional neural network, a recurrent neural network or a deep neural network structure; the target mandible image is an image representing the position of the mandible in the target oral image obtained after being segmented by the rough mandible segmentation model; the mental foramen detection model is a neural network model that has been trained and is used for locating the position of the mental foramen; the mental foramen detection model includes a Gaussian heat map regression model; The determining the target mandible image of the target object according to the target oral image and the pre-established rough mandible segmentation model includes: inputting the target oral image into the established rough mandible segmentation model to obtain a mandible mask image; cropping the target oral image based on the mandible mask image to obtain the target mandible image; Wherein, the target mental foramen localization image includes a left mental foramen localization image and a right mental foramen localization image; the determining the target mental foramen localization image of the target object according to the target mandible image and the pre-established mental foramen detection model includes: inputting the target mandible image into the pre-established mental foramen detection model to obtain a preliminary mental foramen localization image, wherein the preliminary mental foramen localization image includes the first mental foramen position information of the left mental foramen and the second mental foramen position information of the right mental foramen; cropping the target mandible image according to the first mental foramen position information and the second mental foramen position information to respectively obtain the left mental foramen localization image corresponding to the left mandibular nerve canal of the target object and the right mental foramen localization image corresponding to the right mandibular nerve canal of the target object.

2. The method according to claim 1, wherein The mental foramen detection model includes a Gaussian heat map regression model, and the loss function of the mental foramen detection model is weighted Adaptive Wing loss.

3. The method according to claim 1, wherein The determining the segmentation result of the mandibular nerve canal of the target object according to the target mental foramen localization image and the pre-established nerve canal segmentation model includes: Determine the left nerve canal segmentation image and the right nerve canal segmentation image of the target object according to the left mental foramen localization image, the right mental foramen localization image and the pre-established nerve canal segmentation model; Determine the segmentation result of the mandibular nerve canal of the target object according to the left nerve canal segmentation image and the right nerve canal segmentation image.

4. The method according to claim 3, wherein Determining the left and right neural tube segmentation images of the target object according to the left mental foramen positioning image, the right mental foramen positioning image, and a pre-established neural tube segmentation model includes: Inputting the left mental foramen positioning image into the pre-established neural tube segmentation model to obtain the left neural tube segmentation image of the target object; Inputting the right mental foramen positioning image into the pre-established neural tube segmentation model to obtain the right neural tube segmentation image of the target object.

5. The method according to claim 3, wherein Determining the segmentation result of the mandibular neural tube of the target object according to the left neural tube segmentation image and the right neural tube segmentation image includes: Performing image fusion on the left neural tube segmentation image, the right neural tube segmentation image, and the original oral cavity image to obtain the segmentation result of the mandibular neural tube of the target object.

6. A segmentation device for the mandibular nerve canal, characterized in that, Including: A preprocessing module, configured to obtain the original oral cavity image of the target object, and perform preprocessing on the original oral cavity image to obtain a target oral cavity image; A segmentation image determination module, configured to determine the target mandible image of the target object according to the target oral cavity image and a pre-established mandible rough segmentation model; A positioning image determination module, configured to determine the target mental foramen positioning image of the target object according to the target mandible image and a pre-established mental foramen detection model; A segmentation result determination module, configured to determine the segmentation result of the mandibular neural tube of the target object according to the target mental foramen positioning image and a pre-established neural tube segmentation model; Wherein, the mandible rough segmentation model is a neural network model that has been trained and is used for rough segmentation of the oral mandible; the mandible rough segmentation model is composed of at least one of a fully convolutional neural network, a recurrent neural network, or a deep neural network structure; the target mandible image is an image representing the position of the mandible in the target oral cavity image obtained after being segmented by the mandible rough segmentation model; the mental foramen detection model is a neural network model that has been trained and is used for positioning the mental foramen position; the mental foramen detection model includes a Gaussian heatmap regression model; The segmentation image determination module is specifically configured to input the target oral cavity image into the established mandible rough segmentation model to obtain a mandible mask image; and crop the target oral cavity image based on the mandible mask image to obtain the target mandible image; The target mental foramen positioning image includes a left mental foramen positioning image and a right mental foramen positioning image; The positioning image determination module is specifically configured to input the target mandible image into the pre-established mental foramen detection model to obtain a preliminary mental foramen positioning image, where the preliminary mental foramen positioning image includes the first mental foramen position information of the left mental foramen and the second mental foramen position information of the right mental foramen; and crop the target mandible image according to the first mental foramen position information and the second mental foramen position information to respectively obtain the left mental foramen positioning image corresponding to the left mandibular neural tube of the target object and the right mental foramen positioning image corresponding to the right mandibular neural tube.

7. An electronic device, characterized in that, The electronic device includes: One or more processors; A storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method for segmenting the mandibular nerve canal as described in any one of claims 1-5.

8. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to execute the method for segmenting the mandibular nerve canal as described in any one of claims 1-5.

Images