Image recognition-based oral risk assessment methods, systems, and media

Abstract

This invention discloses an image recognition-based method, system, and medium for oral risk assessment, relating to the field of image recognition technology. The method includes: segmenting a user's oral cavity image into blocks; activating a lesion recognizer to identify image block features; matching oral diseases and obtaining corresponding bacterial type sets; imaging the target bacterial community using a fluorescence microscope and extracting fluorescence features; and combining the identified disease information and fluorescence features to generate an oral risk decision for the target user. This achieves the technical effect of multi-dimensional image comprehensive analysis, improving the accuracy and efficiency of the assessment.

Description

Technical Field

[0001] This invention relates to the field of image recognition technology, and in particular to a method, system and medium for oral risk assessment based on image recognition. Background Art

[0002] The unique working environment of long-haul operations can alter the microbial flora of personnel in their mouths, respiratory tracts, and gastrointestinal tracts, leading to a series of pathological changes, various diseases, and affecting the health of crew members.

[0003] Traditional methods for diagnosing oral diseases rely primarily on the clinical examination and experience of dentists, which is subjective and limited to some extent, affecting efficiency. Furthermore, the utilization of oral images is insufficient; image analysis and processing mostly remain superficial, failing to delve into the relevant information contained within the images. Summary of the Invention

[0004] This invention provides an image recognition-based oral risk assessment method, system, and medium to address the technical problems of insufficient recognition accuracy and inadequate risk decision-making in existing technologies, thereby achieving multi-dimensional image comprehensive analysis and improving the accuracy and efficiency of assessment.

[0005] In a first aspect, the present invention provides an image recognition-based oral risk assessment method, wherein the image recognition-based oral risk assessment method includes:

[0006] The target oral cavity image of the target user, acquired through medical imaging equipment, is segmented into blocks to obtain a set of target image blocks.

[0007] The lesion recognizer is activated to analyze and recognize the first image features of the first image block to obtain the first lesion recognition result, wherein the first image block refers to any image block in the target image block set.

[0008] If the first lesion identification result meets the predetermined constraints, then a first oral disease is matched, and the first bacterial community type set corresponding to the first oral disease is obtained by traversing the oral disease microbiome database.

[0009] The first bacterial group in the first bacterial group type set was acquired by fluorescence imaging using a fluorescence microscope, and a first fluorescence image was obtained.

[0010] The first fluorescence feature of the first fluorescence image is collected and combined with the first oral disease to form the target oral risk decision for the target user.

[0011] In one feasible implementation, before activating the lesion recognizer to analyze and recognize the first image features of the first image block to obtain the first lesion recognition result, the following steps are included:

[0012] Determine the first oral cavity region corresponding to the first image block.

[0013] A first image sample set is constructed for the first oral cavity region, wherein the first image sample set includes a first healthy image and a first lesion image.

[0014] A training dataset is constructed based on the first healthy image and the first lesion image, and supervised machine learning is performed on the training dataset to obtain the lesion recognizer.

[0015] The training dataset, constructed based on the first healthy image and the first lesion image, includes:

[0016] Obtain the first feature information of the first health image and form the first dataset.

[0017] Second feature information of the first lesion image is obtained, and a second dataset is formed.

[0018] The first dataset and the second dataset are used to construct the training dataset.

[0019] In one feasible implementation, if the first lesion identification result meets predetermined constraints, then a first oral disease is matched, and the first set of microbial types corresponding to the first oral disease is obtained by traversing the oral disease microbiome database, including:

[0020] Obtain a healthy oral cavity sample, wherein the healthy oral cavity sample has a healthy microbial community type set.

[0021] Obtain a first oral sample of the first oral disease, and the first oral sample has a disease microbiota type set.

[0022] By comparing the healthy microbial community type set with the disease microbial community type set, the first microbial community type set of the first oral disease is obtained.

[0023] Based on the correspondence between the first oral disease and the first microbial community type set, the oral disease microbial community database is constructed.

[0024] In one feasible implementation, a first fluorescent image is obtained by acquiring fluorescence imaging data of a first bacterial community within the first bacterial community type set using a fluorescence microscope, including:

[0025] Obtain the first antibody corresponding to the first bacterial group.

[0026] The first antibody was fluorescently labeled to obtain the first labeling result.

[0027] The first fluorescence image is acquired using the fluorescence microscope, wherein the first fluorescence image refers to the combined imaging of the first labeling result and the first bacterial community.

[0028] In one feasible implementation, collecting first fluorescence features from the first fluorescence image and combining them with the first oral disease to form a target oral risk decision for the target user includes:

[0029] The first fluorescence index is obtained by normalizing and weighting the first fluorescence feature.

[0030] The first fluorescence index is used as the first risk index for the first oral disease.

[0031] The target oral risk decision is formed based on the first correspondence between the first oral disease and the first risk index.

[0032] In one feasible implementation, the first fluorescence feature includes at least a first fluorescence coverage and a first fluorescence dispersion.

[0033] In one feasible implementation, after collecting the first fluorescence features of the first fluorescence image and combining them with the first oral disease to form a target oral risk decision for the target user, the method further includes:

[0034] Obtain the first predetermined amplification curve of the first bacterial population.

[0035] Using the first predetermined amplification curve as a constraint, and combining the first fluorescence characteristics, the predicted fluorescence characteristics of the first bacterial community at the target time are obtained.

[0036] The predicted fluorescence features are analyzed to obtain a predicted risk index, which is then used to form a target oral risk prediction for the target user.

[0037] In one feasible implementation, the predicted fluorescence characteristics of the first bacterial community at the target time are obtained by combining the first predetermined amplification curve with the first fluorescence characteristics, including:

[0038] The first predetermined amplification curve is subjected to polynomial fitting to obtain the first fitting formula.

[0039] The target time is input into the first fitting formula to obtain the first fitting value.

[0040] The first fitted value is used as the first predicted amplification coefficient, and the first fluorescence feature is weighted and adjusted to obtain the predicted fluorescence feature.

[0041] Secondly, the present invention also provides an image recognition-based oral risk assessment system, wherein the image recognition-based oral risk assessment system includes:

[0042] The image block segmentation module is used to segment the target oral cavity image of the target user obtained through medical imaging equipment into blocks to obtain a set of target image blocks.

[0043] The lesion recognition module is used to activate the lesion recognizer to analyze and recognize the first image features of the first image block to obtain the first lesion recognition result, wherein the first image block refers to any image block in the target image block set.

[0044] The oral disease matching module is used to match a first oral disease if the first lesion identification result meets the predetermined constraints, and to traverse the oral disease microbiome database to obtain the first microbiome type set corresponding to the first oral disease.

[0045] The fluorescence imaging acquisition module is used to acquire fluorescence images of the first bacterial group in the first bacterial group type set using a fluorescence microscope, and obtain the first fluorescence image.

[0046] The oral risk decision module is used to collect the first fluorescence features of the first fluorescence image and combine them with the first oral disease to form the target oral risk decision for the target user.

[0047] Thirdly, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the image recognition-based oral risk assessment method provided by the present invention.

[0048] This invention discloses an image recognition-based oral risk assessment method, system, and medium, comprising: segmenting target oral images of a target user acquired through medical imaging equipment to construct a target image block set; activating a lesion identifier to extract and identify features from any first image block in the target image block set to obtain a first lesion identification result; if the first lesion identification result meets predetermined identification constraints, matching the corresponding first oral disease and searching for the first bacterial community type set corresponding to the first oral disease in a preset oral disease microbiome database; performing fluorescence imaging on the target first bacterial community in the first bacterial community type set using a fluorescence microscope to obtain a first fluorescence image; extracting the first fluorescence feature of the first fluorescence image and combining it with the first oral disease to perform risk inference, forming a target oral risk decision result for the target user. The image recognition-based oral risk assessment method, system, and medium disclosed in this invention solve the technical problems of insufficient identification accuracy and insufficient comprehensiveness of risk decision, and achieve the technical effects of multi-dimensional image comprehensive analysis and improved assessment accuracy and efficiency. Attached Figure Description

[0049] Figure 1This is a flowchart illustrating the oral risk assessment method based on image recognition of the present invention.

[0050] Figure 2 This is a schematic diagram of the oral risk assessment system based on image recognition of the present invention.

[0051] Figure labeling: Image block segmentation module 11, lesion recognition module 12, oral disease matching module 13, fluorescence imaging acquisition module 14, oral risk decision module 15. Detailed Implementation

[0052] The above technical solutions will now be described in detail with reference to the accompanying drawings and specific embodiments to provide a better understanding of them. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments used only to explain the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. Furthermore, it should be noted that, for ease of description, only the parts related to the present invention are shown in the drawings, not all of them.

[0053] Example 1, as Figure 1 This is a flowchart illustrating the oral risk assessment method based on image recognition of the present invention, wherein the oral risk assessment method based on image recognition includes:

[0054] S100: Perform block segmentation on the target oral cavity image of the target user acquired through medical imaging equipment to obtain a set of target image blocks.

[0055] Specifically, medical imaging equipment refers to medical devices that can acquire images of the oral cavity, such as dental CBCT and digital panoramic dental X-ray machines, which can provide detailed image information of the target user's oral cavity for subsequent analysis.

[0056] Specifically, block segmentation is an image processing technique used to divide regions according to a preset spatial partitioning strategy, thereby obtaining a set of image blocks with local features. This facilitates the individual analysis of different regions of the image, improving the accuracy and efficiency of the analysis.

[0057] Specifically, firstly, images of the target user's oral cavity are acquired using medical imaging equipment to obtain raw image data containing the complete oral cavity structure; then, based on the image size, resolution, and oral cavity structure distribution characteristics, image block segmentation parameters are set, including image block size, overlap rate, and segmentation method.

[0058] Furthermore, during the segmentation process, the target oral cavity image can be spatially segmented using methods such as sliding windows or grid partitioning to generate multiple image blocks, each corresponding to a local region in the oral cavity image; at the same time, the position index of different image blocks in the original image is identified to form a set of target image blocks for subsequent processing modules to call.

[0059] Through the segmentation operation described above, local regions with different features can be extracted from the overall image, which is the set of target image patches, providing a basis for subsequent in-depth analysis of each image patch.

[0060] S200: Activate the lesion recognizer to analyze and recognize the first image features of the first image block to obtain the first lesion recognition result, wherein the first image block refers to any one image block in the target image block set.

[0061] Specifically, the lesion identifier is a pre-defined image analysis module used to automatically identify and classify potential lesion regions in oral cavity image blocks. This lesion identifier integrates an image feature extraction submodule based on a convolutional neural network and a lesion type discrimination submodule, enabling it to model and analyze pathological features in the input image block. The first image block refers to any image block selected from the target image block set, containing local structural information of the oral cavity image, and can be input into the lesion identifier as an independent analysis unit for processing. The first image feature is multi-dimensional image feature data extracted from the first image block, exemplarily including texture features, edge features, color distribution, spatial morphology, etc.

[0062] Through the above process, lesion identification and analysis can be performed on any image block in the target image block set, which helps to improve the automation and intelligence level of lesion detection, reduce the repetitive labor intensity of relevant personnel, and provide reliable data support for the subsequent personalized treatment plan.

[0063] In some embodiments, before activating the lesion recognizer to analyze and recognize the first image features of the first image block to obtain the first lesion recognition result, the process includes:

[0064] Determine the first oral cavity region corresponding to the first image patch; construct a first image sample set for the first oral cavity region, wherein the first image sample set includes a first healthy image and a first lesion image; construct a training dataset based on the first healthy image and the first lesion image, and perform supervised machine learning on the training dataset to obtain the lesion recognizer; wherein constructing the training dataset based on the first healthy image and the first lesion image includes:

[0065] First feature information of the first healthy image is obtained and a first dataset is formed; second feature information of the first lesion image is obtained and a second dataset is formed; the first dataset and the second dataset are used to construct the training dataset.

[0066] Specifically, the first oral cavity region refers to the specific area corresponding to the first image patch in the target oral cavity image, including but not limited to the incisor region, molar region, gingival margin, and lingual region. This region guides the construction of subsequent sample sets, thereby enabling targeted training and enhancement of the model. The first image sample set is a collection and organization of representative images targeting the first oral cavity region. It includes first healthy images and first lesion images acquired from the same or similar locations. The first healthy image is a standard structural image without obvious lesion features, while the first lesion image is a labeled image of a diagnosed lesion area. This set is used to construct a training dataset to improve the lesion identifier's recognition accuracy in this region.

[0067] Specifically, after image patch segmentation, the first oral cavity region corresponding to the first image patch is determined by combining the position information of the image patch in the original image with the oral cavity structure mapping relationship. Then, image samples corresponding to the oral cavity region are retrieved from historical databases or expert-annotated samples to form a first image sample set, which contains multiple first healthy images and first lesion images. Next, image features are extracted from the first healthy images, including first feature information such as texture, edge, color, and shape, to form a first dataset. At the same time, image features are extracted from the first lesion images to obtain second feature information, forming a second dataset. Then, the first dataset and the second dataset are merged to construct a training dataset, and a lesion recognizer based on methods such as convolutional neural networks, support vector machines, or ensemble learning is constructed and trained using supervised machine learning algorithms.

[0068] Through the above process, a region-specific training dataset is constructed based on the oral anatomical location corresponding to the first image patch, and a dedicated recognition model is obtained through supervised machine learning methods. This enables the lesion recognizer to have stronger regional adaptability and classification accuracy, which helps to improve the early recognition accuracy of lesions in specific oral locations.

[0069] S300: If the first lesion identification result meets the predetermined constraints, then match the first oral disease, and traverse the oral disease microbiome database to obtain the first microbiome type set corresponding to the first oral disease.

[0070] Specifically, pre-defined constraints are a set of conditions (parameters or indicators) set in advance to determine whether the lesion identification result is reliable, such as a specific range of feature values, image pattern matching degree, confidence level of the first lesion identification result, etc.

[0071] Specifically, the oral disease microbiota database is a collection of data that stores information on different types of oral diseases and their corresponding microbiota. In other words, the oral disease microbiota database contains detailed information on the types, quantities, and distribution of microbiota in the oral cavity under various oral disease conditions. By traversing and searching through this database, the first microbiota type set corresponding to the first oral disease can be obtained, that is, the relevant microbiota combination specific to the oral disease, which can be used as a reference for further collection and analysis.

[0072] In the above processing flow, on the one hand, by imposing preset constraints on the lesion identification results, unreliable identification results are filtered out, thereby improving the accuracy of subsequent risk assessment; on the other hand, the process of matching oral diseases and obtaining corresponding microbial type sets broadens the dimensions of risk assessment, and the subsequent assessment combined with oral microbial indicators helps to improve the comprehensiveness of the assessment results.

[0073] In some embodiments, if the first lesion identification result meets predetermined constraints, a first oral disease is matched, and a first set of microbial types corresponding to the first oral disease is obtained by traversing the oral disease microbiome database, including:

[0074] Obtain a healthy oral cavity sample, wherein the healthy oral cavity sample has a healthy microbial community type set; obtain a first oral cavity sample of the first oral disease, wherein the first oral cavity sample has a disease microbial community type set; compare the healthy microbial community type set with the disease microbial community type set to obtain the first microbial community type set of the first oral disease; construct the oral disease microbial community database according to the correspondence between the first oral disease and the first microbial community type set.

[0075] Specifically, a healthy oral cavity sample is a sample taken from the oral cavity of a healthy individual, which can represent the composition of oral microorganisms in a normal and healthy state. Among them, the healthy microbial community type set is the combination information of various healthy microorganisms contained in the sample.

[0076] Specifically, the first oral sample for a first oral disease refers to an oral sample from a patient with a specific oral disease (i.e., the first oral disease), and the combination of microorganisms present in the sample constitutes the disease microbiota type set.

[0077] Specifically, firstly, oral samples were collected from individuals without clinical symptoms, i.e., healthy oral samples. After microbiome sequencing or flora analysis, a healthy flora type set was determined, including the composition and abundance distribution of flora under normal microecological conditions. Simultaneously, samples were collected from patients diagnosed with a primary oral disease, i.e., primary oral samples. Similarly, these samples underwent similar microbiome analysis to obtain the corresponding disease flora type set, reflecting the flora imbalance characteristics under that disease state.

[0078] Furthermore, a difference analysis is performed between the healthy microbiota type set and the disease microbiota type set, including but not limited to differences in species diversity, changes in dominant species, and significance tests of characteristic microbiota abundance. Based on statistical methods (such as LEfSe analysis, Wilcoxon rank-sum test, principal component analysis, PCA, etc.), microbiota types with significant differences are extracted and output as the first microbiota type set for the first oral disease. This microbiota type set may include indicators such as a list of characteristic species, abundance change trends, and abnormalities in related metabolic pathways.

[0079] Furthermore, based on the mapping relationship between the first oral disease and the first bacterial community type set, an oral disease bacterial community database is constructed to record different oral diseases (such as dental caries, gingivitis, periodontitis, oral candidiasis, etc.) and their corresponding characteristic bacterial community type sets.

[0080] For example, the structure of an oral disease microbiota database is shown in Table 1 below:

[0081] Table 1. Exemplary oral disease microbiota database

[0082]

[0083] Preferably, the oral disease microbiota database can be further linked with information such as microbiota regulation strategies, probiotic intervention programs, and individualized treatment recommendations, which can facilitate the improvement of response efficiency after risk assessment.

[0084] The above process can effectively link the lesion results of the image recognition layer with the microbial community characteristics of the microecological layer, realizing cross-modal information fusion from lesion detection to microbial community association, and providing data support for subsequent analysis.

[0085] S400: The first bacterial group in the first bacterial group type set is acquired by fluorescence imaging using a fluorescence microscope to obtain the first fluorescence image.

[0086] Specifically, a fluorescence microscope is a precision instrument that uses fluorescence properties to observe and image samples. It works by exciting fluorescent substances in a sample with light of a specific wavelength, thereby revealing the distribution and morphology of specific components within the sample. Observing the first bacterial community using a fluorescence microscope converts the observed fluorescence signal into image data, resulting in a first fluorescence image. This image visually displays the morphology, distribution, and other characteristics of the first bacterial community after fluorescent labeling, providing data support for subsequent quantitative analysis of the community's characteristics and risk assessment.

[0087] For example, firstly, fluorescence imaging is performed on the first bacterial group within the first bacterial group type set using a fluorescence microscope to obtain a first fluorescence image. The first bacterial group is a target strain isolated from a first oral cavity sample, and it has been pre-stained specifically using fluorescence labeling methods (e.g., fluorescence in situ hybridization FISH, fluorescent protein labeling, fluorescent dye staining, etc.) to enhance its detectability during fluorescence imaging. The acquired first fluorescence image records the distribution characteristics, aggregation morphology, and relative abundance information of the target bacterial group in the oral cavity sample.

[0088] In some embodiments, fluorescence imaging is performed on a first bacterial community within the first bacterial community type set using a fluorescence microscope to obtain a first fluorescence image, including:

[0089] Obtain the first antibody corresponding to the first bacterial community; perform fluorescent labeling on the first antibody to obtain a first labeling result; obtain the first fluorescence image through the fluorescence microscope, wherein the first fluorescence image refers to the binding image of the first labeling result and the first bacterial community.

[0090] Specifically, firstly, based on the specific antigens of the target bacterial species in the first bacterial group, a first antibody with highly specific binding ability is obtained. This first antibody can be a monoclonal or polyclonal antibody, preferably designed to target the surface antigens of the target strain to ensure specific binding and accurate imaging. Then, the first antibody is fluorescently labeled to obtain the first labeling result. For example, the fluorescent label can use common fluorescent molecules such as fluorescein isothiocyanate (FITC), rhodamine, or Alexa Fluor series dyes, enabling the antibody to generate a detectable fluorescent signal under specific wavelength excitation.

[0091] Furthermore, the first labeling result is incubated with the first bacterial group to allow the fluorescently labeled antibody to bind to the surface antigen of the target bacterial group and form a stable complex; subsequently, the bound sample area is imaged and acquired using a fluorescence microscope to obtain the first fluorescence image.

[0092] The above process transforms microbiological information into analyzable image data, achieving a significant shift from identifying bacterial community types to visualizing specific bacterial community characteristics. This provides crucial image data for subsequent quantitative analysis and risk decision-making based on fluorescence features.

[0093] S500: Collect the first fluorescence features of the first fluorescence image and combine them with the first oral disease to form a target oral risk decision for the target user.

[0094] Specifically, the first fluorescence feature refers to the image feature information extracted from the first fluorescence image, including but not limited to parameters such as brightness distribution, color channel intensity, fluorescence intensity gradient, and edge morphology, used to characterize the optical properties of the lesion area. There is a specific pattern matching relationship between the first fluorescence feature and the first oral disease. Based on this relationship, the oral risks of the target user can be identified, and a target oral risk decision can be generated.

[0095] In some implementations, the first fluorescence feature includes at least a first fluorescence coverage and a first fluorescence dispersion.

[0096] Specifically, the first fluorescence features of the first fluorescence image include indicators that reflect the growth status of the bacterial community, such as fluorescence coverage and fluorescence dispersion. First fluorescence coverage refers to the proportion of area in the first fluorescence image that is detected as abnormal fluorescence (such as weak or strong fluorescence areas), used to measure the distribution breadth of lesions or plaques in the entire oral cavity image. First fluorescence dispersion refers to the dispersion of the distribution of abnormal fluorescence areas in the image, reflecting whether the lesion area is concentrated in a localized area or widely distributed across multiple locations. For example, it can be quantified by methods such as centroidal distribution, standard deviation, or spatial clustering coefficient, used to assist in judging the severity and development trend of the lesion.

[0097] Specifically, firstly, a fluorescence anomaly region is identified using a fluorescence intensity threshold segmentation algorithm, and its area proportion in the entire image is calculated to obtain the first fluorescence coverage rate. For example, in an image with a resolution of 1024×1024 pixels, if the total area of ​​the abnormal fluorescence region is 50,000 pixels, the first fluorescence coverage rate is approximately 4.77%. Subsequently, the spatial distribution of these abnormal regions is analyzed, their centroid coordinates are calculated, and the standard deviation or mean distance of their distribution is evaluated to obtain the first fluorescence dispersion.

[0098] In some embodiments, collecting first fluorescence features from the first fluorescence image and combining them with the first oral disease to form a target oral risk decision for the target user includes:

[0099] The first fluorescence feature is normalized and weighted to obtain a first fluorescence index; the first fluorescence index is used as a first risk index for the first oral disease; and the target oral risk decision is formed based on the first correspondence between the first oral disease and the first risk index.

[0100] Specifically, after extracting features such as primary fluorescence coverage and primary fluorescence dispersion, normalization is first performed. For example, coverage is mapped from 0%–100% to the 0–1 range, and dispersion is normalized to the same range based on maximum and minimum values. Then, weights are set based on experience or data-driven methods, for example, coverage weight is 0.6 and dispersion weight is 0.4. The primary fluorescence index is calculated as 0.6 × normalized coverage + 0.4 × normalized dispersion. For example, if a user's normalized coverage is 0.7 and dispersion is 0.5, the primary fluorescence index is 0.62. Next, this fluorescence index is used as the primary risk index for the primary oral disease (such as dental caries). Based on a preset correspondence (e.g., 0–0.3 for low risk, 0.3–0.6 for medium risk, and above 0.6 for high risk), the user is classified as high-risk. A target oral risk decision is then output based on this risk level, such as recommending a follow-up visit or improved daily oral hygiene. The correspondence between the first oral disease and the first risk index can be obtained through risk thresholds or risk models derived from statistical analysis of clinical data.

[0101] Through the above process, complex and multidimensional image feature information can be transformed into unified and interpretable risk quantification indicators, which simplifies the subsequent risk assessment and clinical decision-making process, while improving the objectivity and consistency of the assessment results.

[0102] In some embodiments, after collecting the first fluorescence features of the first fluorescence image and combining them with the first oral disease to form a target oral risk decision for the target user, the method further includes:

[0103] Obtain a first predetermined amplification curve of the first bacterial community; using the first predetermined amplification curve as a constraint, combine the first fluorescence characteristics to obtain the predicted fluorescence characteristics of the first bacterial community at the target time; analyze the predicted fluorescence characteristics to obtain a predicted risk index, and form a target oral risk prediction for the target user.

[0104] Specifically, the first predetermined amplification curve refers to a model of the growth trend or fluorescence signal change of a specific oral microbiota (such as Streptococcus mutans, Porphyromonas gingivalis, etc.) under specific environmental conditions. Preferably, this first predetermined amplification curve is obtained by fitting experimental data and can reflect the changes in the reproduction rate or metabolic activity of the microbiota over time. The predicted fluorescence characteristics are based on the current fluorescence image characteristics and combined with the microbiota amplification model to predict the fluorescence image characteristics (such as coverage, dispersion, etc.) at a future time point, used to predict the development trend of lesions.

[0105] Specifically, the predictive risk index is a future disease risk level calculated based on the aforementioned predictive fluorescence characteristics. It is a time extension of the current risk index (i.e., a prospective risk index). By obtaining the predictive risk index, it is possible to further provide information on the trend of oral health changes over a future period based on the current risk assessment, which has prospective and intervention guidance significance.

[0106] Specifically, after completing the risk assessment of the current fluorescence image, a microbial amplification model related to the current user is further invoked. For example, based on saliva samples or historical data, the main pathogenic bacterium for the user is identified as *Porphyromonas gingivalis*, and its fluorescence signal amplification curve under standard conditions is invoked (e.g., fluorescence coverage increases exponentially over time, with an initial value of 0.62 and a growth rate of +0.05 per day). Then, using this amplification curve as a constraint, combined with the first fluorescence feature extracted from the current fluorescence image, the predicted fluorescence coverage and dispersion at several future time points (e.g., 3 days, 7 days later) are calculated to form predicted fluorescence features. Assuming the predicted coverage is 0.77 and the dispersion is 0.58 after 3 days, the predicted fluorescence index is calculated to be 0.70 through normalized weighted calculation.

[0107] Furthermore, based on the mapping relationship between fluorescence index and oral diseases, it is determined that the user will be in a high-risk state after 3 days, and the corresponding target oral risk prediction results are output, such as: the risk of gingivitis is expected to increase after 3 days, and it is recommended to strengthen the use of dental floss and have a follow-up examination.

[0108] Through the above process, the risk of future diseases can be predicted based on the trend of microbial proliferation, and potential health threats can be identified in advance, thereby helping to improve the timeliness and scientific nature of individualized prevention and intervention.

[0109] In some implementations, the predicted fluorescence characteristics of the first bacterial community at the target time are obtained by combining the first predetermined amplification curve with the first fluorescence characteristics, including:

[0110] The first predetermined amplification curve is fitted with a polynomial to obtain a first fitting formula; the target time is input into the first fitting formula to obtain a first fitting value; the first fitting value is used as a first predicted amplification coefficient, and the first fluorescence feature is weighted and adjusted to obtain the predicted fluorescence feature.

[0111] Specifically, the fluorescence response values ​​of the first bacterial swarm (such as Streptococcus mutans) at multiple time points are first extracted from experimental or historical data, and these data are fitted with a polynomial to obtain a fitting function. Then, the target time (such as the next 3 days) is substituted into the fitting formula to obtain the predicted fluorescence response value, and the ratio of the fitted value to the initial value is calculated to obtain the first predicted amplification coefficient.

[0112] Then, the current fluorescence features are weighted and adjusted according to the first predicted amplification coefficient (normalization is performed if the value is exceeded). The adjusted features will be used as the predicted fluorescence features at the target time point for subsequent risk index calculation.

[0113] Through the above process, it is possible to extrapolate changes in image features at future time points based on current fluorescence image features, taking into account the growth dynamics of the microbial community. This enhances the biological rationality and temporal continuity of the prediction model, providing a clearer time window and risk basis for subsequent intervention strategies.

[0114] In summary, the image recognition-based oral risk assessment method provided by this invention has the following technical effects:

[0115] By segmenting the target oral cavity images of the target user acquired by medical imaging equipment into image blocks, a target image block set is constructed. A lesion identifier is activated, and feature extraction and recognition are performed on any first image block in the target image block set to obtain a first lesion recognition result. If the first lesion recognition result meets predetermined recognition constraints, a corresponding first oral disease is matched, and the first bacterial community type set corresponding to the first oral disease is searched in a preset oral disease microbiome database. Fluorescence imaging of the target first bacterial community in the first bacterial community type set is performed using a fluorescence microscope to obtain a first fluorescence image. The first fluorescence feature of the first fluorescence image is extracted and combined with the first oral disease for risk inference to form a target oral risk decision result for the target user. This achieves the technical effect of multi-dimensional image comprehensive analysis and improves the accuracy and efficiency of assessment.

[0116] Example 2, as Figure 2 This is a schematic diagram of the oral risk assessment system based on image recognition of the present invention. For example, Figure 1 The flowchart of the oral risk assessment method based on image recognition of the present invention can be seen as follows: Figure 2 The structure shown is implemented.

[0117] Based on the same concept as the image recognition-based oral risk assessment method in the embodiments described above, the present invention also provides an image recognition-based oral risk assessment system comprising:

[0118] The image block segmentation module 11 is used to segment the target oral cavity image of the target user obtained by medical imaging equipment into blocks to obtain a set of target image blocks.

[0119] The lesion recognition module 12 is used to activate the lesion recognizer to analyze and recognize the first image features of the first image block to obtain the first lesion recognition result, wherein the first image block refers to any image block in the target image block set.

[0120] The oral disease matching module 13 is used to match the first oral disease if the first lesion identification result meets the predetermined constraints, and to traverse the oral disease microbiome database to obtain the first microbiome type set corresponding to the first oral disease.

[0121] The fluorescence imaging acquisition module 14 is used to acquire fluorescence images of the first bacterial group in the first bacterial group type set using a fluorescence microscope, and obtain a first fluorescence image.

[0122] The oral risk decision module 15 is used to collect the first fluorescence features of the first fluorescence image and combine them with the first oral disease to form the target oral risk decision for the target user.

[0123] In some embodiments, the lesion recognition module 12 includes:

[0124] The first oral cavity location determination unit is used to determine the first oral cavity location corresponding to the first image block.

[0125] The first image sample set assembly unit is used to assemble a first image sample set of the first oral cavity, wherein the first image sample set includes a first healthy image and a first lesion image.

[0126] The training dataset construction and lesion recognizer training unit is used to construct a training dataset based on the first healthy image and the first lesion image, and to perform supervised machine learning on the training dataset to obtain the lesion recognizer.

[0127] The first dataset acquisition unit is used to acquire the first feature information of the first health image and form the first dataset.

[0128] The second dataset acquisition unit is used to acquire the second feature information of the first lesion image and form a second dataset.

[0129] A training dataset building unit is used to build the training dataset from the first dataset and the second dataset.

[0130] In some embodiments, the oral disease matching module 13 includes:

[0131] A healthy oral cavity sample acquisition unit is used to acquire a healthy oral cavity sample, wherein the healthy oral cavity sample has a set of healthy microbial community types.

[0132] The first oral sample acquisition unit is used to acquire a first oral sample of the first oral disease, and the first oral sample has a disease microbial community type set.

[0133] The microbial community type set comparison unit is used to compare the healthy microbial community type set with the disease microbial community type set to obtain the first microbial community type set of the first oral disease.

[0134] The oral disease microbiota database construction unit is used to construct the oral disease microbiota database based on the correspondence between the first oral disease and the first microbiota type set.

[0135] In some embodiments, the fluorescence imaging acquisition module 14 includes:

[0136] The first antibody acquisition unit is used to acquire the first antibody corresponding to the first bacterial group.

[0137] A fluorescent labeling unit is used to fluorescently label the first antibody to obtain a first labeling result.

[0138] The first fluorescence image acquisition unit is used to acquire the first fluorescence image through the fluorescence microscope, wherein the first fluorescence image refers to the combined imaging of the first labeling result and the first bacterial community.

[0139] In some embodiments, the oral risk decision module 15 includes:

[0140] The first antibody acquisition unit is used to acquire the first antibody corresponding to the first bacterial group.

[0141] A fluorescent labeling unit is used to fluorescently label the first antibody to obtain a first labeling result.

[0142] The first fluorescence image acquisition unit is used to acquire the first fluorescence image through the fluorescence microscope, wherein the first fluorescence image refers to the combined imaging of the first labeling result and the first bacterial community.

[0143] In some embodiments, the oral risk decision module 15 further includes:

[0144] The first predetermined amplification curve acquisition unit is used to acquire the first predetermined amplification curve of the first bacterial population.

[0145] The predictive fluorescence feature calculation unit is used to obtain the predicted fluorescence features of the first bacterial community at the target time by combining the first predetermined amplification curve with the first fluorescence features.

[0146] The target oral risk prediction forming unit is used to analyze the predicted fluorescence features to obtain the predicted risk index and form the target oral risk prediction for the target user.

[0147] In some implementations, the execution steps of the predictive fluorescence feature calculation unit in the oral risk decision module 15 include:

[0148] The first predetermined amplification curve is subjected to polynomial fitting to obtain the first fitting formula.

[0149] The target time is input into the first fitting formula to obtain the first fitting value.

[0150] The first fitted value is used as the first predicted amplification coefficient, and the first fluorescence feature is weighted and adjusted to obtain the predicted fluorescence feature.

[0151] In some embodiments, the first fluorescence feature includes at least a first fluorescence coverage and a first fluorescence dispersion.

[0152] In embodiment three, this application provides a computer-readable storage medium that can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the image recognition-based oral risk assessment method in this embodiment of the invention. By running the software programs, instructions, and modules stored in the computer-readable storage medium, the above-mentioned image recognition-based oral risk assessment method can be realized.

[0153] It should be understood that the focus of the embodiments mentioned in this specification is their difference from other embodiments. The specific embodiments in the aforementioned Embodiment 1 are also applicable to the image recognition-based oral risk assessment system described in Embodiment 2. For the sake of brevity, they will not be elaborated further here.

[0154] It should be understood that the embodiments disclosed in this invention and the above description enable those skilled in the art to implement this invention. However, this invention is not limited to the embodiments mentioned above. It should be understood that those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this invention, and should all be included within the protection scope of this invention.

Claims

1. An image recognition-based oral risk assessment method, characterized in that, include: The target oral cavity image of the target user acquired by medical imaging equipment is segmented into blocks to obtain a set of target image blocks; The lesion identifier is activated to analyze and identify the first image features of the first image block to obtain the first lesion identification result, wherein the first image block refers to any image block in the target image block set; If the first lesion identification result meets the predetermined constraints, then the first oral disease is matched, and the first bacterial community type set corresponding to the first oral disease is obtained by traversing the oral disease microbiome database. The first bacterial group in the first bacterial group type set was acquired by fluorescence imaging using a fluorescence microscope to obtain the first fluorescence image; Collect the first fluorescence features of the first fluorescence image and combine them with the first oral disease to form a target oral risk decision for the target user; Collecting the first fluorescence features of the first fluorescence image and combining them with the first oral disease to form a target oral risk decision for the target user includes: The first fluorescence index is obtained by performing a normalized weighted calculation on the first fluorescence feature; The first fluorescence index is used as the first risk index for the first oral disease. Based on the first correspondence between the first oral disease and the first risk index, the target oral risk decision is formed; The first fluorescence feature includes at least a first fluorescence coverage and a first fluorescence dispersion; After collecting the first fluorescence features from the first fluorescence image and combining them with the first oral disease to form a target oral risk decision for the target user, the method further includes: Obtain the first predetermined amplification curve of the first bacterial population; Using the first predetermined amplification curve as a constraint, and combining the first fluorescence characteristics, the predicted fluorescence characteristics of the first bacterial community at the target time are obtained. The predicted fluorescence features are analyzed to obtain a predicted risk index, and a target oral risk prediction for the target user is formed. Using the first predetermined amplification curve as a constraint, and combining the first fluorescence characteristics, the predicted fluorescence characteristics of the first bacterial community at the target time are obtained, including: The first predetermined amplification curve is subjected to polynomial fitting to obtain the first fitting formula; Input the target time into the first fitting formula to obtain the first fitting value; The first fitted value is used as the first predicted amplification coefficient, and the first fluorescence feature is weighted and adjusted to obtain the predicted fluorescence feature.

2. The oral risk assessment method based on image recognition as described in claim 1, characterized in that, Before activating the lesion recognizer to analyze and recognize the first image features of the first image block and obtain the first lesion recognition result, the process includes: Determine the first oral cavity region corresponding to the first image block; A first image sample set is constructed for the first oral cavity, wherein the first image sample set includes a first healthy image and a first lesion image; A training dataset is constructed based on the first healthy image and the first lesion image, and supervised machine learning is performed on the training dataset to obtain the lesion recognizer. The training dataset, constructed based on the first healthy image and the first lesion image, includes: Obtain the first feature information of the first health image and form a first dataset; Obtain the second feature information of the first lesion image and form a second dataset; The first dataset and the second dataset are used to construct the training dataset.

3. The oral risk assessment method based on image recognition as described in claim 1, characterized in that, If the first lesion identification result meets the predetermined constraints, then a first oral disease is matched, and the first set of microbial types corresponding to the first oral disease is obtained by traversing the oral disease microbiome database, including: Obtain a healthy oral cavity sample, wherein the healthy oral cavity sample has a set of healthy microbial community types; Obtain a first oral sample of the first oral disease, and the first oral sample has a disease microbiota type set; By comparing the healthy microbiota type set with the disease microbiota type set, the first microbiota type set of the first oral disease is obtained; Based on the correspondence between the first oral disease and the first microbial community type set, the oral disease microbial community database is constructed.

4. The oral risk assessment method based on image recognition as described in claim 1, characterized in that, The first bacterial group in the first bacterial group type set was acquired by fluorescence imaging using a fluorescence microscope, resulting in a first fluorescence image, including: Obtain the first antibody corresponding to the first bacterial group; The first antibody was fluorescently labeled to obtain the first labeling result; The first fluorescence image is acquired using the fluorescence microscope, wherein the first fluorescence image refers to the combined imaging of the first labeling result and the first bacterial community.

5. An image recognition-based oral risk assessment system, characterized in that, To implement the image recognition-based oral risk assessment method according to any one of claims 1 to 4, the method comprises: The image block segmentation module is used to segment the target oral cavity image of the target user acquired by medical imaging equipment into blocks to obtain a set of target image blocks. The lesion recognition module is used to activate the lesion recognizer to analyze and recognize the first image features of the first image block to obtain the first lesion recognition result, wherein the first image block refers to any one image block in the target image block set; The oral disease matching module is used to match the first oral disease if the first lesion identification result meets the predetermined constraints, and to traverse the oral disease microbiome database to obtain the first microbiome type set corresponding to the first oral disease. The fluorescence imaging acquisition module is used to acquire fluorescence images of the first bacterial group in the first bacterial group type set using a fluorescence microscope, and obtain the first fluorescence image. The oral risk decision module is used to collect the first fluorescence features of the first fluorescence image and combine them with the first oral disease to form the target oral risk decision for the target user.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the image recognition-based oral risk assessment method as described in any one of claims 1 to 4.

Images