Oral cavity case association method and system based on image features

Abstract

The invention provides an oral cavity case association method and system based on image features, and the method comprises the steps: collecting and preprocessing an oral cavity image, and obtaining a first image; segmenting the first image by using an image segmentation algorithm based on deep learning to obtain a second image; performing feature extraction on the second image by adopting an optical character recognition technology to obtain character features; and based on the character features, in combination with a preset oral medical term library, through a semantic association method based on natural language processing, mapping the character features and case fields corresponding to the oral image, and realizing association of oral cases. According to the invention, the key information of the oral cavity image is effectively extracted and is associated with the electronic case, so that the intelligent processing of the oral cavity image is realized.

Description

Technical Field

[0001] The present invention belongs to the field of information technology, and in particular relates to an oral case association method and system based on image features. Background Art

[0002] Oral health management is a crucial branch of healthcare, and oral case correlation plays a significant role in improving the efficiency and quality of patient diagnosis and treatment. With the widespread adoption of digital technology, the recording and analysis of oral case data has gradually shifted from traditional paper archives to electronic systems, providing ample room for oral case management. However, many current solutions still have significant shortcomings when processing oral image data, particularly in their ability to automatically extract and correlate information. This results in inefficient oral case management and poor data integration.

[0003] Existing methods, most systems rely on manual data entry or simple image storage, failing to directly extract key information from complex oral images. This approach is not only time-consuming and labor-intensive, but also prone to errors introduced by human operators, compromising the accuracy and completeness of case data. More critically, the lack of intelligent recognition of hidden text within images prevents the effective utilization of a large amount of valuable data.

[0004] A deeper analysis of the challenges facing this field reveals that the core issue lies in the automatic extraction of text from images. Oral images, such as X-rays or CT scans, often contain handwritten or printed text, which varies greatly in font, layout, and background noise, making it difficult for conventional recognition technologies to accurately capture it. This recognition challenge further complicates data association. Even if some text is extracted, it's difficult to accurately match it with the structured data in electronic medical records to form a unified medical record information network. This layered barrier from recognition to association ultimately prevents truly intelligent and efficient medical record management.

[0005] Therefore, how to accurately extract marked text from complex oral images and effectively associate it with electronic case data has become a key issue in improving the level of automation in oral case management. Summary of the Invention

[0006] In order to solve the problems existing in the prior art, the present invention provides an oral case association method and system based on image features, which aims to accurately extract marked text in complex oral images and effectively associate it with electronic case data, thereby effectively improving the level of automation of oral case management.

[0007] To achieve the above object, the present invention provides the following solutions:

[0008] A method for associating oral cases based on image features, the method comprising:

[0009] S1, collecting an oral image and performing preprocessing to obtain a first image;

[0010] S2. Segment the first image using a deep learning-based image segmentation algorithm to obtain a second image;

[0011] S3, extracting features from the second image using optical character recognition technology to obtain text features;

[0012] S4. Based on the text features and in combination with a preset oral medical terminology library, the text features are mapped to the case fields corresponding to the oral images through a semantic association method based on natural language processing to achieve association of oral cases.

[0013] Preferably, the step S1 collects an oral image and performs preprocessing to obtain a first image, including:

[0014] S11, collecting an oral cavity image, removing background noise in the oral cavity image, and obtaining a denoised oral cavity image;

[0015] S12, performing contrast enhancement on the denoised oral image to obtain a contrast-enhanced oral image;

[0016] S13, performing edge enhancement on the contrast-enhanced oral image to obtain an edge-enhanced oral image;

[0017] S14. Based on a preset image quality assessment standard, perform a quality assessment on the edge-enhanced oral image to obtain a quality assessment result;

[0018] S15. Based on the quality assessment result, perform format standardization processing on the edge-enhanced oral image to obtain a first image.

[0019] Preferably, the step S2 uses a deep learning-based image segmentation algorithm to segment the first image to obtain a second image, including:

[0020] S21. Segment the first image using a deep learning-based image segmentation algorithm to obtain a text image, and generate bounding box coordinates of the marked text in the text image;

[0021] S22. Based on the bounding box coordinates, obtain an independent image block of each region where the marked text is located, and perform size normalization processing on the independent image block to obtain a normalized text image set;

[0022] S23. Obtain a second image based on the text image set.

[0023] Preferably, the image segmentation algorithm based on deep learning in S2 includes: adopting an EAST model, wherein the EAST model includes: a feature extraction layer, a feature fusion layer and an output layer;

[0024] Introducing the SENet attention mechanism between the feature extraction layer and the feature fusion layer;

[0025] The ASPP module is used to replace the conv 5 module in the feature extraction layer;

[0026] A BLSTM network is introduced between the feature fusion layer and the output layer.

[0027] Preferably, the step S3 uses optical character recognition technology to extract features from the second image to obtain text features, including:

[0028] S31, scanning the second image using the optical character recognition technology to obtain initial text data;

[0029] S32, extracting font shape features and typesetting features based on the initial text data;

[0030] S33: Obtain text features based on the font shape features and the typesetting features.

[0031] Preferably, the step S4 maps the text features with the case fields corresponding to the oral images based on the text features in combination with a preset oral medical terminology library through a semantic association method based on natural language processing to achieve association of oral cases, including:

[0032] S41. Obtaining a case field corresponding to the oral image based on a preset oral medical terminology library;

[0033] S42. Mapping the text features to the case fields using a semantic association method based on natural language processing;

[0034] S43. Based on the mapping result, a matching relationship between the text feature and the case field is obtained to achieve association of oral cases.

[0035] The present invention also provides an oral case association system based on image features, the system is used to implement the aforementioned oral case association method based on image features, the system includes: an acquisition module, a segmentation module, a feature extraction module and an association module;

[0036] The acquisition module is used to acquire an oral image and perform preprocessing to obtain a first image;

[0037] The segmentation module is configured to segment the first image using a deep learning-based image segmentation algorithm to obtain a second image;

[0038] The feature extraction module is used to extract features from the second image using optical character recognition technology to obtain text features;

[0039] The association module is used to map the text features with the case fields corresponding to the oral images based on the text features in combination with a preset oral medical terminology library through a semantic association method based on natural language processing, thereby realizing the association of oral cases.

[0040] Preferably, the image segmentation algorithm based on deep learning in the segmentation module adopts the EAST model, the EAST model includes: a feature extraction layer, a feature fusion layer and an output layer, and the segmentation module includes: an attention unit, an ASPP unit and a BLSTM unit;

[0041] The attention unit is used to introduce the SENet attention mechanism between the feature extraction layer and the feature fusion layer;

[0042] The ASPP unit is used to replace the conv 5 module in the feature extraction layer with the ASPP module;

[0043] The BLSTM unit is used to introduce a BLSTM network between the feature fusion layer and the output layer.

[0044] The present invention also provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the aforementioned method for associating oral cases based on image features when executing the program.

[0045] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed, the aforementioned oral case association method based on image features is implemented.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] This invention discloses a method and system for associating oral medical records based on image features. This method uses preprocessing to remove noise and enhance acquired oral images. A deep learning-based image segmentation algorithm is then used to segment text regions and annotate them with bounding boxes. Optical character recognition technology is then used to extract text features. These features are then mapped and matched against a pre-built oral medical terminology library to achieve association with oral medical records. This method effectively extracts key information from oral images and associates it with electronic medical records, enabling intelligent processing of oral images. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0049] Figure 1 Schematic diagram of the flow of an oral case association method based on image features according to an embodiment of the present invention;

[0050] Figure 2 This is a schematic diagram of the improved EAST model structure according to an embodiment of the present invention;

[0051] Figure 3 This is a schematic diagram of the ASPP module structure according to an embodiment of the present invention;

[0052] Figure 4 This is a schematic diagram of the BLSTM network structure according to an embodiment of the present invention;

[0053] Figure 5 Schematic diagram of an oral case association system module based on image features according to an embodiment of the present invention;

[0054] Figure 6 The figure is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0055] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0056] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0057] Example 1

[0058] like Figure 1 As shown, the present invention provides an oral case association method based on image features, comprising:

[0059] S1, collecting an oral image and performing preprocessing to obtain a first image;

[0060] S2. Segment the first image using a deep learning-based image segmentation algorithm to obtain a second image;

[0061] S3, extracting features from the second image using optical character recognition technology to obtain text features;

[0062] S4. Based on text features and combined with the preset oral medical terminology library, the text features are mapped to the case fields corresponding to the oral images through a semantic association method based on natural language processing to achieve the association of oral cases.

[0063] Specifically, S1 collects an oral image and performs preprocessing to obtain a first image, including:

[0064] S11, collecting an oral image, removing background noise in the oral image, and obtaining a denoised oral image;

[0065] S12, performing contrast enhancement on the denoised oral image to obtain a contrast-enhanced oral image;

[0066] S13, performing edge enhancement on the contrast-enhanced oral image to obtain an edge-enhanced oral image;

[0067] S14. Based on a preset image quality assessment standard, perform a quality assessment on the edge-enhanced oral image to obtain a quality assessment result;

[0068] S15. Based on the quality assessment result, perform format standardization processing on the edge-enhanced oral image to obtain a first image.

[0069] S2. Segmenting the first image using a deep learning-based image segmentation algorithm to obtain a second image, including:

[0070] S21. Segment the first image using a deep learning-based image segmentation algorithm to obtain a text image, and generate bounding box coordinates of the marked text in the text image;

[0071] S22. Based on the bounding box coordinates, obtain an independent image block of the area where each marked text is located, perform size normalization processing on the independent image block, and obtain a normalized text image set;

[0072] S23. Obtain a second image based on the text image set.

[0073] Among them, the image segmentation algorithm based on deep learning adopts the EAST model, which includes: feature extraction layer, feature fusion layer and output layer. The present invention improves the EAST model. The schematic diagram of the improved EAST model structure is shown in the figure. Figure 2 As shown:

[0074] Introduce the SENet attention mechanism between the feature extraction layer and the feature fusion layer;

[0075] Use the ASPP module to replace the conv 5 module in the feature extraction layer;

[0076] A BLSTM network is introduced between the feature fusion layer and the output layer.

[0077] Specifically, the EAST model has certain limitations in its performance when dealing with short text detection tasks. The main reason is the downsampling operation in the feature extraction process. The model downsamples (i.e., reduces the resolution) layer by layer from Conv1 to Conv5, reducing the image size and information content by merging or discarding pixels. However, this operation inevitably leads to the gradual loss of image detail features. For the lower-resolution feature maps of the lower layers, the originally smaller details (especially the key features of short text) will become blurred or even disappear completely.

[0078] To address this challenge, this paper introduces an attention mechanism in the shallow layers of the model (i.e., layers that undergo less downsampling). These shallow feature maps (such as f2, f3, and f4) retain rich original image detail information. The attention mechanism automatically identifies and enhances these detailed feature channels that are crucial for short text detection, assigning them higher weights.

[0079] Drawing on the U-Net architecture, this paper introduces an attention mechanism before feature fusion, fusing deep semantic information with shallow detail information. Shallow feature maps contain a wealth of undiluted detail. By applying attention to shallow features before fusion, the detailed components can be more effectively highlighted.

[0080] Specifically, this paper uses the SENet (Squeeze-and-Excitation Network) module to implement channel attention. The SENet mechanism is relatively lightweight, requiring only a small number of parameters (primarily a fully connected layer) to automatically generate channel weights through learning. It can adaptively assess and adjust the importance of each channel in the feature map, significantly improving the discriminative power of features, and is particularly beneficial for capturing and preserving the fine structure required for short text.

[0081] like Figure 2 As shown in the figure, the improved model structure with the SENet attention mechanism embeds SENet modules at three key shallow feature map locations, f4, f3, and f2. Because these layers are rich in detailed features (especially subtle clues in short text), SENet is able to effectively focus on them, assigning greater weight to relevant feature channels. This improved strategy is intended to significantly improve the model's detection accuracy and robustness, especially for short text.

[0082] Traditional convolutional neural network models typically rely on downsampling operations to expand the receptive field of neurons, enabling them to capture a wider range of information and extract more abstract features. However, this strategy is accompanied by a continuous decrease in image resolution, causing local detail information to be gradually lost in the process. Although methods such as linear interpolation upsampling are often used to restore resolution, the problem of information loss is still difficult to completely avoid. The backbone networks used by the EAST text detection algorithm (such as VGG16 or ResNet50) also have this limitation: they increase the receptive field through downsampling, but inevitably sacrifice spatial resolution.

[0083] The emergence of atrous convolution technology provides an effective way to resolve this contradiction. It allows the network to expand the receptive field of the convolution kernel equivalently by introducing "holes" (sampling intervals) without actual downsampling, thereby capturing a wider range of contextual information while maintaining the spatial resolution of the feature map.

[0084] This paper improves the backbone network of the EAST algorithm by replacing the conv 5 portion of the backbone network (using VGG16 as an example) with the Atrous Spatial Pyramid Pooling (ASPP) module. The core idea of ​​ASPP is to apply convolution operations with different dilation rates (and global average pooling) to the input feature map in parallel, thereby capturing multi-scale image context information at a single resolution level.

[0085] After the output feature map of conv 4, the ASPP module is connected, such as Figure 3 As shown, the module contains five parallel branches:

[0086] Global average pooling branch (×1): First, global average pooling is performed, then the number of channels is adjusted through 1×1 convolution, and finally bilinear interpolation upsampling is used to restore the original spatial size.

[0087] Dilated convolution branch (×4): Convolution layers with dilation rates of 1 (i.e., standard convolution), 6, 12, and 18 are used respectively (usually combined with 1×1 convolution for channel adjustment).

[0088] The outputs of these five branches are concatenated in the channel dimension, and then passed through a 1×1 convolutional layer for feature fusion and dimension reduction to the required number of channels as input for subsequent processing.

[0089] Furthermore, text detection tasks require not only identifying the position of individual characters but also understanding the continuity between them. Characters themselves possess visual features, while adjacent characters have contextual relationships. Convolutional neural network models excel at learning local spatial features, while sequence modeling (such as LSTM) effectively captures sequential dependencies. Therefore, leveraging both the spatial feature extraction capabilities of convolutional neural network models and the contextual modeling capabilities of sequence models is crucial for improving text sequence detection performance.

[0090] To this end, the present invention introduces a bidirectional long short-term memory (BLSTM) network after the feature merging layer and before the final output layer of the EAST network structure. BLSTM is a special recurrent neural network (RNN) composed of two LSTMs, one forward and one backward.

[0091] The specific structure of BLSTM is as follows Figure 4 As shown in Figure 2, this module receives the feature maps output by the EAST feature merging layer. The core mechanism of BLSTM is that it processes the input sequence (converted from the feature maps) in both the forward (past to future) and reverse (future to past) directions, generating two independent hidden state sequences. These two hidden state sequences capture the "historical" and "future" contextual information of the sequence, respectively. Ultimately, these two sequences are concatenated to form a new feature representation that incorporates bidirectional contextual information, which is then used for the final target output (such as text line detection). This design enables effective serialized association modeling of feature maps, making the generated sequence samples more continuous and reasonable.

[0092] S3. Extract features from the second image using optical character recognition technology to obtain text features, including:

[0093] S31, scanning the second image using optical character recognition technology to obtain initial text data;

[0094] S32, extracting font shape features and typesetting features based on the initial text data;

[0095] S33. Obtain text features based on font shape features and typesetting features.

[0096] S4. Based on text features and a preset oral medical terminology database, the text features are mapped to the case fields corresponding to the oral images through a semantic association method based on natural language processing to achieve association of oral cases, including:

[0097] S41. Based on a preset oral medical terminology library, obtaining case fields corresponding to the oral image;

[0098] S42. Map text features to case fields using a semantic association method based on natural language processing;

[0099] S43. Based on the mapping results, the matching relationship between the text features and the case fields is obtained to achieve the association of oral cases.

[0100] In summary, the present invention discloses an image-feature-based oral medical record association method. This method uses preprocessing to denoise and enhance acquired oral images, then uses a deep learning-based image segmentation algorithm to segment text regions and annotate them with bounding boxes. Optical character recognition technology is then used to extract text features. These features are then mapped and matched against a pre-built oral medical terminology library to achieve oral medical record association. This method effectively extracts key information from oral images and associates it with electronic medical records, enabling intelligent processing of oral images.

[0101] Example 2

[0102] like Figure 5 As shown, the present invention also provides an oral case association system based on image features, which is used to implement the oral case association method based on image features of the above embodiment. The system includes: an acquisition module, a segmentation module, a feature extraction module and an association module;

[0103] An acquisition module, configured to acquire an oral cavity image and perform preprocessing to obtain a first image;

[0104] a segmentation module, configured to segment the first image using a deep learning-based image segmentation algorithm to obtain a second image;

[0105] A feature extraction module, configured to extract features from the second image using optical character recognition technology to obtain text features;

[0106] The association module is used to map text features with case fields corresponding to oral images based on text features and a preset oral medical terminology library through a semantic association method based on natural language processing, thereby realizing the association of oral cases.

[0107] Furthermore, the deep learning-based image segmentation algorithm in the segmentation module adopts the EAST model. The EAST model includes: feature extraction layer, feature fusion layer and output layer. The segmentation module includes: attention unit, ASPP unit and BLSTM unit;

[0108] Attention unit, used to introduce SENet attention mechanism between feature extraction layer and feature fusion layer;

[0109] ASPP unit, used to replace the conv 5 module in the feature extraction layer with the ASPP module;

[0110] The BLSTM unit is used to introduce the BLSTM network between the feature fusion layer and the output layer.

[0111] Example 3

[0112] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor, wherein when the processor executes the program, the oral case association method based on image features described in any of the above embodiments is implemented.

[0113] Figure 6 10 is a schematic diagram showing a more specific hardware structure of an electronic device provided in this embodiment. The device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are communicatively connected to each other within the device via the bus 1050.

[0114] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0115] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage devices, dynamic storage devices, etc. The memory 1020 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0116] The input / output interface 1030 is used to connect input / output modules to implement information input and output. The input / output modules can be configured as components within the device (not shown in the figure) or can be externally connected to the device to provide corresponding functions. Input devices may include a keyboard, mouse, touch screen, microphone, various sensors, etc., and output devices may include a display, speaker, vibrator, indicator light, etc.

[0117] The communication interface 1040 is used to connect to a communication module (not shown) to enable communication between the device and other devices. The communication module can communicate via a wired method (e.g., USB (Universal Serial Bus), network cable, etc.) or a wireless method (e.g., mobile network, WIFI (Wireless Fidelity), Bluetooth, etc.).

[0118] The bus 1050 comprises a path for transmitting information between the various components of the device (eg, the processor 1010 , the memory 1020 , the input / output interface 1030 , and the communication interface 1040 ).

[0119] It should be noted that although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040, and the bus 1050, in a specific implementation, the device may also include other components necessary for normal operation. In addition, it will be understood by those skilled in the art that the above device may only include the components necessary to implement the embodiments of this specification, and does not necessarily include all the components shown in the figure.

[0120] The system of the above embodiment is used to implement the corresponding oral case association method based on image features in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0121] Example 4

[0122] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present disclosure also provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable the computer to execute the oral case association method based on image features as described in any of the above embodiments.

[0123] The computer-readable media of this embodiment include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0124] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the oral case association method based on image features as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0125] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present disclosure (including the claims) is limited to these examples. Within the scope of the present disclosure, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present disclosure as described above, which are not provided in detail for the sake of simplicity.

[0126] In addition, to simplify the description and discussion, and so as not to obscure the embodiments of the present disclosure, known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided figures. In addition, devices may be shown in the form of block diagrams to avoid obscuring the embodiments of the present disclosure, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present disclosure are to be implemented (i.e., these details should be fully within the purview of those skilled in the art). Where specific details (e.g., circuits) are set forth to describe exemplary embodiments of the present disclosure, it will be apparent to those skilled in the art that the embodiments of the present disclosure may be implemented without these specific details or with variations in these specific details. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0127] Although the present disclosure has been described in conjunction with specific embodiments thereof, many alternatives, modifications, and variations of these embodiments will be apparent to those skilled in the art based on the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.

[0128] Therefore, the units of each example described in the embodiments of this application can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0129] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by persons skilled in the art should fall within the scope of protection defined by the claims of the present invention.

Claims

1. A method for associating oral cases based on image features, characterized in that: The method comprises: S1, collecting an oral image and performing preprocessing to obtain a first image; S2. Segment the first image using a deep learning-based image segmentation algorithm to obtain a second image; S3, extracting features from the second image using optical character recognition technology to obtain text features; S4. Based on the text features and in combination with a preset oral medical terminology library, the text features are mapped to the case fields corresponding to the oral images through a semantic association method based on natural language processing to achieve association of oral cases.

2. The oral case association method based on image features according to claim 1, characterized in that: The step S1 collects an oral image and performs preprocessing to obtain a first image, including: S11, collecting an oral cavity image, removing background noise in the oral cavity image, and obtaining a denoised oral cavity image; S12, performing contrast enhancement on the denoised oral image to obtain a contrast-enhanced oral image; S13, performing edge enhancement on the contrast-enhanced oral image to obtain an edge-enhanced oral image; S14. Based on a preset image quality assessment standard, perform a quality assessment on the edge-enhanced oral image to obtain a quality assessment result; S15. Based on the quality assessment result, perform format standardization processing on the edge-enhanced oral image to obtain a first image.

3. The oral case association method based on image features according to claim 1, characterized in that: The step S2 uses a deep learning-based image segmentation algorithm to segment the first image to obtain a second image, including: S21. Segment the first image using a deep learning-based image segmentation algorithm to obtain a text image, and generate bounding box coordinates of the marked text in the text image; S22. Based on the bounding box coordinates, obtain an independent image block of each region where the marked text is located, and perform size normalization processing on the independent image block to obtain a normalized text image set; S23. Obtain a second image based on the text image set.

4. The oral case association method based on image features according to claim 1, characterized in that: The deep learning-based image segmentation algorithm described in S2 includes: adopting an EAST model, wherein the EAST model includes: a feature extraction layer, a feature fusion layer, and an output layer; Introducing the SENet attention mechanism between the feature extraction layer and the feature fusion layer; The ASPP module is used to replace the conv 5 module in the feature extraction layer; A BLSTM network is introduced between the feature fusion layer and the output layer.

5. The oral case association method based on image features according to claim 1, characterized in that: The step S3 uses optical character recognition technology to extract features from the second image to obtain text features, including: S31, scanning the second image using the optical character recognition technology to obtain initial text data; S32, extracting font shape features and typesetting features based on the initial text data; S33: Obtain text features based on the font shape features and the typesetting features.

6. The oral case association method based on image features according to claim 1, characterized in that: The S4 maps the text features with the case fields corresponding to the oral images based on the text features and a preset oral medical terminology library through a semantic association method based on natural language processing to achieve association of oral cases, including: S41. Obtaining a case field corresponding to the oral image based on a preset oral medical terminology library; S42. Mapping the text features to the case fields using a semantic association method based on natural language processing; S43. Based on the mapping result, a matching relationship between the text feature and the case field is obtained to achieve association of oral cases.

7. An oral case association system based on image features, the system being used to implement the oral case association method based on image features according to any one of claims 1 to 6, characterized in that: The system includes: an acquisition module, a segmentation module, a feature extraction module and an association module; The acquisition module is used to acquire an oral image and perform preprocessing to obtain a first image; The segmentation module is configured to segment the first image using a deep learning-based image segmentation algorithm to obtain a second image; The feature extraction module is used to extract features from the second image using optical character recognition technology to obtain text features; The association module is used to map the text features with the case fields corresponding to the oral images based on the text features in combination with a preset oral medical terminology library through a semantic association method based on natural language processing, thereby realizing the association of oral cases.

8. The oral case association system based on image features according to claim 7, characterized in that: The image segmentation algorithm based on deep learning in the segmentation module adopts the EAST model, the EAST model includes: a feature extraction layer, a feature fusion layer and an output layer, and the segmentation module includes: an attention unit, an ASPP unit and a BLSTM unit; The attention unit is used to introduce the SENet attention mechanism between the feature extraction layer and the feature fusion layer; The ASPP unit is used to replace the conv 5 module in the feature extraction layer with the ASPP module; The BLSTM unit is used to introduce a BLSTM network between the feature fusion layer and the output layer.

9. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for associating oral cases based on image features according to any one of claims 1 to 6 is implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed, the oral case association method based on image features according to any one of claims 1 to 6 is implemented.

Images