CN114004970A - Tooth area detection method, device, equipment and storage medium - Google Patents
Tooth area detection method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN114004970A CN114004970A CN202111320584.7A CN202111320584A CN114004970A CN 114004970 A CN114004970 A CN 114004970A CN 202111320584 A CN202111320584 A CN 202111320584A CN 114004970 A CN114004970 A CN 114004970A
- Authority
- CN
- China
- Prior art keywords
- medical image
- tooth
- oral
- model
- oral medical
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
Abstract
The application discloses a tooth region detection method, a device, equipment and a storage medium, wherein the method comprises the following steps: acquiring an oral medical image dataset; the data set includes an oral medical image and annotated oral medical image data; preprocessing the data set, and taking the preprocessed data set as a training sample; constructing an oral tooth prediction model; the prediction model comprises a semantic segmentation model and a target detection model for parallel processing of images, and a screening module for combining tooth region results output by the two models and screening out error regions; training the prediction model by using a training sample; and inputting the oral medical image to be detected into the trained prediction model for recognition and segmentation, outputting the tooth area of the oral medical image and performing coding representation. Therefore, the introduction of a parallel processing frame based on a semantic segmentation model and a target detection model enables the image segmentation process to be more accurate, and tooth areas which are accurately identified and clearly segmented are effectively obtained.
Description
Technical Field
The invention relates to the technical field of oral medical imaging, in particular to a tooth region detection method, a tooth region detection device, tooth region detection equipment and a storage medium.
Background
Along with the development and progress of the society, the living standard of people is greatly improved, the living style is diversified more and more, the problems of oral cavities caused by a plurality of bad living habits are generated, and people pay more and more attention to the tooth conditions of the oral cavities.
The whole structure of the oral cavity can be observed according to the oral cavity image, and further, the case data can be recorded. The dentist firstly carries out visual judgment on the teeth, observes whether the teeth have damage and color change under the light source through eyes, then detects the condition of the teeth through a special oral examination instrument, and carries out comprehensive analysis by combining an oral panoramic film. Although intraoral imaging is characterized by high prevalence and low cost, it also requires a doctor to determine teeth, gums and other oral areas based on a panoramic view of the mouth. Since human film reading is affected by emotion and fatigue, long-time film reading can cause the reduction of recognition efficiency of doctors. Meanwhile, the oral cavity image is a gray image, and the naked eye cannot distinguish the area with the unobvious target edge contour in the gray image. Thus, imaging the entire mouth does not allow a physician to quickly identify the teeth and other areas of the mouth.
Disclosure of Invention
In view of the above, the present invention provides a tooth region detection method, apparatus, device and storage medium, which can effectively obtain tooth regions with accurate identification and clear segmentation. The specific scheme is as follows:
a method of dental region detection, comprising:
acquiring an oral medical image dataset; the oral medical image data set comprises an oral medical image and oral medical image data labeled on the oral medical image;
preprocessing the oral medical image data set, and taking the preprocessed oral medical image data set as a training sample;
constructing an oral tooth prediction model; the oral cavity tooth prediction model comprises a semantic segmentation model and an object detection model which are used for carrying out parallel processing on images, and further comprises a screening module which is used for combining tooth region results output by the semantic segmentation model and the object detection model and screening error regions;
training the oral tooth prediction model by using the training sample;
and inputting the oral medical image to be detected into the trained oral tooth prediction model for recognition and segmentation, outputting the tooth area of the oral medical image to be detected, and performing coding representation.
Preferably, in the above tooth region detection method provided by the embodiment of the present invention, the preprocessing the oral medical image data set includes:
and performing data enhancement processing on the oral medical image data set.
Preferably, in the above tooth region detection method provided by the embodiment of the present invention, the preprocessing the oral medical image data set further includes:
resizing the images in the oral medical image dataset to a uniform size using an interpolation method;
processing the image after the size adjustment by using a high contrast retention algorithm to obtain a high contrast retention characteristic;
and normalizing the processed image.
Preferably, in the tooth region detection method provided in the embodiment of the present invention, the screening module is specifically configured to combine the tooth region output by the semantic segmentation model and the tooth region detection frame and tooth region coordinates output by the target detection model, screen out an error region from a combination result to obtain a pseudo label of the training sample, and add unlabeled oral medical image data; and taking other tooth areas except the error area in the combination result as the screened tooth area.
Preferably, in the above tooth region detection method provided by the embodiment of the present invention, the oral tooth prediction model further includes: and the correction module is used for correcting the tooth area screened by the screening module and updating the labeled oral medical image data.
Preferably, in the tooth region detection method provided in the embodiment of the present invention, the semantic segmentation model is a deep learning neural network model including four convolutional layers, four pooling layers, four activation layers, four pooling layers, four upsampling layers, and an output layer; the semantic segmentation model uses a fully connected conditional random field;
the target detection model is a deep learning neural network model which comprises thirteen convolutional layers, fifteen active layers, four pooling layers, two full-connection layers and an output layer.
Preferably, in the tooth region detection method provided in the embodiment of the present invention, the method further includes:
and predicting and analyzing the oral medical image which is not marked by using the trained oral tooth prediction model, analyzing an independent tooth area from a prediction result according to a preset rule, and marking the independent tooth area.
An embodiment of the present invention further provides a tooth area detection device, including:
the data set acquisition module is used for acquiring an oral medical image data set; the oral medical image data set comprises an oral medical image and oral medical image data labeled on the oral medical image;
the data set preprocessing module is used for preprocessing the oral medical image data set and taking the preprocessed oral medical image data set as a training sample;
the model construction module is used for constructing an oral tooth prediction model; the oral cavity tooth prediction model comprises a semantic segmentation model and an object detection model which are used for carrying out parallel processing on images, and further comprises a screening module which is used for combining tooth region results output by the semantic segmentation model and the object detection model and screening error regions;
the model training module is used for training the oral tooth prediction model by using the training sample;
and the tooth area identification module is used for inputting the oral cavity medical image to be detected into the trained oral cavity tooth prediction model for identification and segmentation, outputting the tooth area of the oral cavity medical image to be detected and carrying out coding representation.
The embodiment of the present invention further provides a tooth region detection device, which includes a processor and a memory, wherein the processor implements the tooth region detection method provided by the embodiment of the present invention when executing the computer program stored in the memory.
Embodiments of the present invention further provide a computer-readable storage medium for storing a computer program, wherein the computer program is executed by a processor to implement the tooth region detection method according to the embodiments of the present invention.
According to the technical scheme, the tooth area detection method provided by the invention comprises the following steps: acquiring an oral medical image dataset; the oral medical image data set comprises an oral medical image and oral medical image data marked on the oral medical image; preprocessing an oral medical image data set, and taking the preprocessed oral medical image data set as a training sample; constructing an oral tooth prediction model; the oral cavity tooth prediction model comprises a semantic segmentation model and a target detection model which are used for carrying out parallel processing on images, and further comprises a screening module which is used for combining tooth region results output by the semantic segmentation model and the target detection model and screening error regions; training the oral tooth prediction model by using the training sample; and inputting the oral medical image to be detected into the trained oral tooth prediction model for recognition and segmentation, outputting the tooth area of the oral medical image to be detected, and performing coding representation.
The invention collects the oral medical image training semantic segmentation model and the target detection model, integrates the semantic segmentation model and the target detection model into a frame, processes the oral medical image by adopting a parallel idea, reads and processes the image simultaneously for the input oral medical image, and the semantic segmentation and target detection model combines the output results of the two models, thereby effectively obtaining the tooth area with accurate identification and clear segmentation after removing the error area. The introduction of the parallel processing frame based on the semantic segmentation model and the target detection model enables the image segmentation process to be more accurate, makes up for the defect of inaccurate single-model segmentation positioning, reduces the human interference factors and the dependence limit on professional literacy of doctors in the film reading process, and further improves the accuracy, consistency and reliability.
In addition, the invention also provides a corresponding device, equipment and a computer readable storage medium for the tooth region detection method, so that the method is further more practical, and the device, the equipment and the computer readable storage medium have corresponding advantages.
Drawings
In order to more clearly illustrate the embodiments of the present invention or technical solutions in related arts, the drawings used in the description of the embodiments or related arts will be briefly introduced below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a flow chart of a method for dental region detection according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a tooth region detection device according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention provides a tooth region detection method, as shown in fig. 1, comprising the following steps:
s101, acquiring an oral medical image data set; the oral medical image data set comprises an oral medical image and oral medical image data marked on the oral medical image;
this step is the raw image acquisition phase. The oral medical image data can be labeled by a professional doctor, and the oral medical image can be an oral tooth panoramic image and contains complete oral tooth structure information. Preferably, the oral medical image is selected according to randomness, including subjects of different sexes and ages.
S102, preprocessing an oral medical image data set, and taking the preprocessed oral medical image data set as a training sample;
this step is a pre-treatment stage. The pretreatment method is various and can be selected according to actual conditions. Specifically, each oral medical image and the labeled oral medical image form a training sample, and a plurality of training samples form a data set; the data set is divided into a training set, a verification set and a test set; the training set is used for model training, the verification set is used for model adjustment, and the test set is model testing; the ratio of the training set to the test set is 4: 1. For example, the training set may have 1000 samples and the test set may have 200 samples.
S103, constructing an oral tooth prediction model; the oral cavity tooth prediction model comprises a semantic segmentation model and a target detection model which are used for carrying out parallel processing on images, and further comprises a screening module which is used for combining tooth region results output by the semantic segmentation model and the target detection model and screening error regions;
the step is a model construction stage, and mainly adopts an image parallel processing framework as a classifier structure. The semantic segmentation is to classify each pixel point in the image and determine the category of each pixel point, for example, the pixel point belongs to teeth, gum or other parts, so as to realize the division of different regions. Compared with an example segmentation model, the image semantic segmentation model only distinguishes the types of the pixel points and does not distinguish tooth examples to which the pixel points belong. The object detection is to classify objects in the image and identify the positions of the objects, such as the positions of teeth, gums or other objects, so as to identify different regions.
S104, training the oral tooth prediction model by using the training sample;
this step is the training phase. The training method of the semantic segmentation model comprises the following steps: the training samples are input into a semantic segmentation model to obtain the tooth region. The training method of the target detection model comprises the following steps: and inputting the training sample into the target detection model to obtain the tooth area detection frame and the tooth area coordinates.
S105, inputting the oral medical image to be detected into the trained oral tooth prediction model for recognition and segmentation, outputting the tooth area of the oral medical image to be detected, and performing coding representation;
the step is a prediction stage, the trained oral tooth prediction model is used for recognizing and segmenting oral medical images, and all tooth areas are coded and expressed.
In the tooth area detection method provided by the embodiment of the invention, the semantic segmentation model and the target detection model are trained by acquiring the oral medical image, the semantic segmentation model and the target detection model are integrated into a frame, the oral medical image is processed by adopting a parallel thought, the semantic segmentation and the target detection model read and process the input oral medical image at the same time, the output results of the two models are combined by the screening module, and the tooth area which is accurately identified and clearly segmented can be effectively obtained after the error area is removed. The introduction of the parallel processing frame based on the semantic segmentation model and the target detection model enables the image segmentation process to be more accurate, makes up for the defect of inaccurate single-model segmentation positioning, reduces the human interference factors and the dependence limit on professional literacy of doctors in the film reading process, and further improves the accuracy, consistency and reliability.
In a specific implementation, in the tooth region detection method provided in the embodiment of the present invention, the step S101 of preprocessing the oral medical image data set may include: carry out data enhancement to oral cavity medical image data set, for example horizontal upset and vertical upset etc. can enrich oral cavity medical image training set like this, reduce the overfitting phenomenon of model, strengthen the generalization ability of model.
Further, the step S101 is to pre-process the oral medical image data set, and may further include: the size of the image in the oral medical image data set is adjusted to be uniform by using an interpolation method, so that the image characteristics can be better extracted, and the model is prevented from being over-fitted. Specifically, the oral medical image is usually large in size, and the labeled oral medical image is scaled to 512 × 512, which is preset. The overfitting phenomenon of the model is reduced, and the generalization capability of the model is enhanced.
In practical application, the image and the annotation graph can be randomly cut, and when the cutting size is larger than the original graph, the filling operation is performed to fill the original graph to the size needing to be cut.
Further, the step S101 is to pre-process the oral medical image data set, and may further include: and processing the image after the size adjustment by using a high contrast retention algorithm to obtain a high contrast retention characteristic. Specifically, a Gaussian filter is used for smoothing the image, the enhanced edge value is obtained by subtracting the image after Gaussian blur from the original image, and the enhanced edge value multiplied by a correlation coefficient is added on the basis of the blurred image, so that the image with enhanced effect is obtained, the image noise is reduced, and the edge detail characteristics are reserved. Specifically, loss of image information due to the fact that most pixel values exceed the effective range is avoided, and the high-contrast retention feature is obtained by adding 127 to an enhanced edge value obtained by subtracting the image after Gaussian blur from the original image. Wherein the N-dimensional spatial gaussian distribution equation is:
where r denotes the blur radius, σ denotes the standard deviation of the normal distribution, and N denotes the dimension of the sample.
Wherein the two-dimensional space Gaussian distribution equation is as follows:
where u represents the x-coordinate of the sample and v represents the y-coordinate of the sample.
Further, the step S101 is to pre-process the oral medical image data set, and may further include: and normalizing the processed image. Therefore, affine invariance can be kept, the calculated amount of the model is simplified, and the convergence of the training neural network is accelerated. Specifically, the polar difference transformation method is used for changing the pixel value of the oral medical image into a decimal between 0 and 1, and the influence of dimension and data value range is eliminated. And converting the dimensional expression into a dimensionless expression to become a pure quantity. Wherein the normalization formula is as follows:
where x represents the entire sample set, xiRepresents the ith sample and norm represents the normalized sample set.
In specific implementation, in the tooth region detection method provided in the embodiment of the present invention, the screening module is specifically configured to combine the tooth region output by the semantic segmentation model and the tooth region detection frame and tooth region coordinates output by the target detection model, screen out an error region from the combination result to obtain a pseudo label of the training sample, and add the unlabeled dental medical image data; and taking other tooth areas except the error areas in the combination result as the screened tooth areas.
In a specific implementation manner, in the tooth region detection method provided in an embodiment of the present invention, the oral tooth prediction model further includes: and the correction module is used for correcting the tooth area screened by the screening module and updating the labeled oral medical image data. The pixel points and coordinates of the identified area are corrected to achieve the pixel-level distinguishing effect
In particular, in the tooth region detection method provided by the embodiment of the invention, the semantic segmentation model may include a convolutional layer, an activation layer, a pooling layer, an upsampling layer, and a softmax layer. Preferably, the semantic segmentation model may be set as a deep learning neural network model including four convolutional layers, four pooling layers, four activation layers, four pooling layers, four upsampling layers, and one output layer. The semantic segmentation model uses fully connected conditional random fields. The conditional random field used by the semantic segmentation model conforms to Gibbs distribution:
in the formula, E (X | I) denotes an energy function, z (I) denotes a normalization factor, and P (X ═ X | I) denotes a conditional probability.
In the formula, the unary potential function comes from the output of the former fully-connected network, i and j respectively represent,
representing a unitary potential, only in relation to the value of point i,
representing binary potential energy, the relation between each node, and E (x) representing a potential energy function.
In a specific implementation manner, in the tooth region detection method provided by the embodiment of the invention, the target detection model may include a convolutional layer, an active layer, a pooling layer, a full link layer, and a softmax layer. Preferably, the target detection model may be set as a deep learning neural network model including thirteen convolutional layers, fifteen activation layers, four pooling layers, two fully-connected layers, and one output layer. The neural network parameter updating formula is as follows:
wherein, W represents weight, b represents bias, z represents output result of each layer of the neural network, and l represents the l-th layer of the neural network;
the gradient of the weight W of the ith layer is represented, the partial derivative of the function to W is lost, and the gradient represents that the function changes most quickly along the gradient direction at one point;
a gradient representing the bias b of the ith layer, a partial derivative of the loss function to b;
representing the partial derivative of the loss function to the ith output layer function z;
representing the partial derivative of the output function of the l layer to the weight W of the l layer;
representing the partial derivative of the l-th layer output function to the l-th layer bias b.
Further, in a specific implementation, the tooth region detection method provided in the embodiment of the present invention may further include: and predicting and analyzing the oral medical image which is not marked by using the trained oral tooth prediction model, analyzing an independent tooth area from a prediction result according to a preset rule, and marking the independent tooth area.
Specifically, the preset marking rule is set according to a dental position representation method of a dental standard. Wherein, the tooth position representation method is a method for numbering and representing each human tooth; the upper and lower dentition are divided into four regions, upper, lower, left and right, by cross, the upper right region is also called region A, the upper left region is also called region B, the lower right region is also called region C, and the lower left region is also called region D. Each tooth is marked with two digits, the first representing the quadrant in which the tooth is located: the upper right, upper left, lower left and lower right of the patient are 1, 2, 3 and 4 in permanent teeth and 5, 6, 7 and 8 in deciduous teeth; the second bit represents the position of the tooth: 1-8 from the central incisors to the third molars.
The following describes the specific steps of the tooth region detection method provided by the embodiment of the present invention by way of an example:
step one, acquiring data. 2834 samples are collected in an oral hospital, and each sample comprises original oral medical image data and marked oral medical image data corresponding to the original oral medical image data.
And step two, preprocessing the image. The image is subjected to operations such as turning, scaling, Gaussian blur, normalization and the like, so that the oral medical image training set is enriched, the overfitting phenomenon of the model is reduced, and the generalization capability of the model is enhanced.
And step three, dividing a sample set. According to the concentration gradient method, the ratio of 4: scale of 1, 2834 experimental samples were divided into a training set of 2267 samples and a testing set of 567 samples.
And step four, establishing a model. The parallel processing framework mainly comprises a semantic segmentation model and a target detection model. The preset semantic segmentation model comprises four convolution layers, four pooling layers, four activation layers, four pooling layers, four up-sampling layers and an output layer. The preset target detection model comprises thirteen convolutional layers, fifteen active layers, four pooling layers, two full-connection layers and an output layer.
And step five, testing the model. And predicting the modeling set and the prediction set, recording the model with the threshold value exceeding 0.9 in the output result as a positive sample, and obtaining satisfactory prediction precision by the selected model with the precision of 0.96 on the test set.
By executing the steps from one step to five, the tooth area in the exit cavity image can be effectively identified, and the human interference factors and the dependence limit on professional literacy of doctors in the radiographing process are reduced.
Based on the same inventive concept, the embodiment of the present invention further provides a tooth region detection apparatus, and since the principle of the apparatus for solving the problem is similar to that of the aforementioned tooth region detection method, the implementation of the apparatus can refer to the implementation of the tooth region detection method, and repeated details are not repeated.
In practical implementation, the tooth region detecting device provided in the embodiment of the present invention, as shown in fig. 2, specifically includes:
a data set acquisition module 11, configured to acquire an oral medical image data set; the oral medical image data set comprises an oral medical image and oral medical image data marked on the oral medical image;
the data set preprocessing module 12 is configured to preprocess the oral medical image data set, and use the preprocessed oral medical image data set as a training sample;
the model construction module 13 is used for constructing an oral tooth prediction model; the oral cavity tooth prediction model comprises a semantic segmentation model and a target detection model which are used for carrying out parallel processing on images, and further comprises a screening module which is used for combining tooth region results output by the semantic segmentation model and the target detection model and screening error regions;
a model training module 14, configured to train the oral tooth prediction model by using a training sample;
and the tooth area identification module 15 is used for inputting the oral medical image to be detected into the trained oral tooth prediction model for identification and segmentation, outputting the tooth area of the oral medical image to be detected and performing coding representation.
In the tooth area detection device provided by the embodiment of the invention, through the interaction of the five modules, the input oral medical image is read and processed by semantic segmentation and a target detection model at the same time, the screening module combines the output results of the two models, and after an error area is removed, a tooth area which is accurately identified and clearly segmented is effectively obtained, the defect of inaccurate segmentation and positioning of a single model is overcome, the artificial interference factor in the process of reading the film and the dependence limit on professional literacy of doctors are reduced, and the accuracy, the consistency and the reliability are further improved.
In a specific implementation, the tooth region detection method provided in the embodiment of the present invention may further include: and the independent tooth area marking module is used for predicting and analyzing the oral medical image which is not marked by utilizing the trained oral tooth prediction model, analyzing the independent tooth area from the prediction result according to a preset rule and marking the independent tooth area.
For more specific working processes of the modules, reference may be made to corresponding contents disclosed in the foregoing embodiments, and details are not repeated here.
Correspondingly, the embodiment of the invention also discloses tooth area detection equipment, which comprises a processor and a memory; wherein the processor, when executing the computer program stored in the memory, implements the tooth region detection method disclosed in the foregoing embodiments.
For more specific processes of the above method, reference may be made to corresponding contents disclosed in the foregoing embodiments, and details are not repeated here.
Further, the present invention also discloses a computer readable storage medium for storing a computer program; the computer program, when executed by a processor, implements the tooth region detection method disclosed above.
For more specific processes of the above method, reference may be made to corresponding contents disclosed in the foregoing embodiments, and details are not repeated here.
The embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same or similar parts among the embodiments are referred to each other. The device, the equipment and the storage medium disclosed by the embodiment correspond to the method disclosed by the embodiment, so that the description is relatively simple, and the relevant points can be referred to the method part for description.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in Random Access Memory (RAM), memory, Read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
To sum up, the tooth area detection method provided by the embodiment of the invention comprises the following steps: acquiring an oral medical image dataset; the oral medical image data set comprises an oral medical image and oral medical image data marked on the oral medical image; preprocessing an oral medical image data set, and taking the preprocessed oral medical image data set as a training sample; constructing an oral tooth prediction model; the oral cavity tooth prediction model comprises a semantic segmentation model and a target detection model which are used for carrying out parallel processing on images, and further comprises a screening module which is used for combining tooth region results output by the semantic segmentation model and the target detection model and screening error regions; training the oral tooth prediction model by using the training sample; and inputting the oral medical image to be detected into the trained oral tooth prediction model for recognition and segmentation, outputting the tooth area of the oral medical image to be detected, and performing coding representation. The method comprises the steps of collecting an oral medical image training semantic segmentation model and a target detection model, integrating the semantic segmentation model and the target detection model into a framework, processing the oral medical image by adopting a parallel idea, reading and processing the input oral medical image by the semantic segmentation and target detection model, combining output results of the two models by a screening module, and effectively obtaining a tooth area which is accurate in identification and clear in segmentation after an error area is removed. The introduction of the parallel processing frame based on the semantic segmentation model and the target detection model enables the image segmentation process to be more accurate, makes up for the defect of inaccurate single-model segmentation positioning, reduces the human interference factors and the dependence limit on professional literacy of doctors in the film reading process, and further improves the accuracy, consistency and reliability. In addition, the invention also provides a corresponding device, equipment and a computer readable storage medium for the tooth region detection method, so that the method is further more practical, and the device, the equipment and the computer readable storage medium have corresponding advantages.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The tooth region detection method, apparatus, device and storage medium provided by the present invention are described in detail above, and the principles and embodiments of the present invention are explained herein by using specific examples, which are merely used to help understand the method and core ideas of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (10)
1. A method of detecting a tooth region, comprising:
acquiring an oral medical image dataset; the oral medical image data set comprises an oral medical image and oral medical image data labeled on the oral medical image;
preprocessing the oral medical image data set, and taking the preprocessed oral medical image data set as a training sample;
constructing an oral tooth prediction model; the oral cavity tooth prediction model comprises a semantic segmentation model and an object detection model which are used for carrying out parallel processing on images, and further comprises a screening module which is used for combining tooth region results output by the semantic segmentation model and the object detection model and screening error regions;
training the oral tooth prediction model by using the training sample;
and inputting the oral medical image to be detected into the trained oral tooth prediction model for recognition and segmentation, outputting the tooth area of the oral medical image to be detected, and performing coding representation.
2. The dental region detection method of claim 1, wherein the preprocessing the set of dental medical image data comprises:
and performing data enhancement processing on the oral medical image data set.
3. The method of claim 2, wherein the preprocessing the set of dental image data further comprises:
resizing the images in the oral medical image dataset to a uniform size using an interpolation method;
processing the image after the size adjustment by using a high contrast retention algorithm to obtain a high contrast retention characteristic;
and normalizing the processed image.
4. The tooth region detection method according to claim 1, wherein the screening module is specifically configured to combine the tooth region output by the semantic segmentation model and the tooth region detection frame and tooth region coordinates output by the target detection model, screen out an error region from a combination result to obtain a pseudo label of the training sample, and add unlabeled oral medical image data; and taking other tooth areas except the error area in the combination result as the screened tooth area.
5. The method of claim 4, wherein the oral teeth prediction model further comprises: and the correction module is used for correcting the tooth area screened by the screening module and updating the labeled oral medical image data.
6. The tooth region detection method according to claim 1, wherein the semantic segmentation model is a deep learning neural network model including four convolutional layers, four pooling layers, four activation layers, four pooling layers, four upsampling layers, and one output layer; the semantic segmentation model uses a fully connected conditional random field;
the target detection model is a deep learning neural network model which comprises thirteen convolutional layers, fifteen active layers, four pooling layers, two full-connection layers and an output layer.
7. The method of detecting a tooth region according to claim 1, further comprising:
and predicting and analyzing the oral medical image which is not marked by using the trained oral tooth prediction model, analyzing an independent tooth area from a prediction result according to a preset rule, and marking the independent tooth area.
8. A dental region detection apparatus, comprising:
the data set acquisition module is used for acquiring an oral medical image data set; the oral medical image data set comprises an oral medical image and oral medical image data labeled on the oral medical image;
the data set preprocessing module is used for preprocessing the oral medical image data set and taking the preprocessed oral medical image data set as a training sample;
the model construction module is used for constructing an oral tooth prediction model; the oral cavity tooth prediction model comprises a semantic segmentation model and an object detection model which are used for carrying out parallel processing on images, and further comprises a screening module which is used for combining tooth region results output by the semantic segmentation model and the object detection model and screening error regions;
the model training module is used for training the oral tooth prediction model by using the training sample;
and the tooth area identification module is used for inputting the oral cavity medical image to be detected into the trained oral cavity tooth prediction model for identification and segmentation, outputting the tooth area of the oral cavity medical image to be detected and carrying out coding representation.
9. A tooth region detection apparatus comprising a processor and a memory, wherein the processor implements the tooth region detection method according to any one of claims 1 to 7 when executing a computer program stored in the memory.
10. A computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements a tooth region detection method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111320584.7A CN114004970A (en) | 2021-11-09 | 2021-11-09 | Tooth area detection method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111320584.7A CN114004970A (en) | 2021-11-09 | 2021-11-09 | Tooth area detection method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114004970A true CN114004970A (en) | 2022-02-01 |
Family
ID=79928329
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111320584.7A Pending CN114004970A (en) | 2021-11-09 | 2021-11-09 | Tooth area detection method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114004970A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116993702A (en) * | 2023-08-09 | 2023-11-03 | 深圳云甲科技有限公司 | Method and related device for realizing single frame data gum separation by deep learning |
| WO2023246462A1 (en) * | 2022-06-20 | 2023-12-28 | 杭州朝厚信息科技有限公司 | Method for segmenting permanent teeth and deciduous teeth and determining tooth numbers in panoramic dental radiograph |
| WO2024108803A1 (en) * | 2022-11-25 | 2024-05-30 | 漳州松霖智能家居有限公司 | Oral cavity examination method, apparatus, system, and related device |
-
2021
- 2021-11-09 CN CN202111320584.7A patent/CN114004970A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023246462A1 (en) * | 2022-06-20 | 2023-12-28 | 杭州朝厚信息科技有限公司 | Method for segmenting permanent teeth and deciduous teeth and determining tooth numbers in panoramic dental radiograph |
| WO2024108803A1 (en) * | 2022-11-25 | 2024-05-30 | 漳州松霖智能家居有限公司 | Oral cavity examination method, apparatus, system, and related device |
| CN116993702A (en) * | 2023-08-09 | 2023-11-03 | 深圳云甲科技有限公司 | Method and related device for realizing single frame data gum separation by deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Panetta et al. | Tufts dental database: a multimodal panoramic x-ray dataset for benchmarking diagnostic systems | |
| CN111539912B (en) | Health index evaluation method and equipment based on face structure positioning and storage medium | |
| CN114004970A (en) | Tooth area detection method, device, equipment and storage medium | |
| CN108830155B (en) | Heart coronary artery segmentation and identification method based on deep learning | |
| CN109146872B (en) | Heart coronary artery image segmentation and identification method based on deep learning and optical flow method | |
| CN113516784B (en) | Tooth segmentation modeling method and device | |
| KR20130136519A (en) | Diagnosis assitance system utilizing panoramic radiographs, and diagnosis assistance program utilizing panoramic radiographs | |
| Chen et al. | MSLPNet: multi-scale location perception network for dental panoramic X-ray image segmentation | |
| Kong et al. | Automated maxillofacial segmentation in panoramic dental x-ray images using an efficient encoder-decoder network | |
| CN115205469A (en) | Tooth and alveolar bone reconstruction method, equipment and medium based on CBCT | |
| Lin et al. | Tooth numbering and condition recognition on dental panoramic radiograph images using CNNs | |
| CN115880266B (en) | Intestinal polyp detection system and method based on deep learning | |
| CN113436734A (en) | Tooth health assessment method and device based on face structure positioning and storage medium | |
| CN112989954A (en) | Three-dimensional tooth point cloud model data classification method and system based on deep learning | |
| CN113516639A (en) | Panoramic X-ray film-based oral cavity anomaly detection model training method and device | |
| CN113160151A (en) | Panoramic film dental caries depth identification method based on deep learning and attention mechanism | |
| CN116797731A (en) | Artificial intelligence-based oral cavity CBCT image section generation method | |
| CN115439409A (en) | Tooth type identification method and device | |
| CN116205925A (en) | Tooth occlusion wing tooth caries segmentation method based on improved U-Net network | |
| Brahmi et al. | Exploring the role of Convolutional Neural Networks (CNN) in dental radiography segmentation: A comprehensive Systematic Literature Review | |
| CN115410032A (en) | OCTA image classification structure training method based on self-supervision learning | |
| Hsu et al. | DeepOPG: Improving Orthopantomogram Finding Summarization with Weak Supervision | |
| Ghafoor et al. | Multiclass Segmentation using Teeth Attention Modules for Dental X-ray Images | |
| CN112150422A (en) | Modeling method of oral health self-detection model based on multitask learning | |
| CN113792773B (en) | Dental root canal treatment history detection model training method, image processing method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |