Tooth image analysis method for oral surgery

Abstract

The invention relates to the technical field of tooth image analysis equipment, and particularly discloses a tooth image analysis method for oral surgery. Through the steps of periodontal condition assessment and intelligent adjustment of image shooting positions, image data acquisition and preprocessing, intelligent segmentation of teeth and surrounding tissue structures, three-dimensional reconstruction and feature quantitative extraction, intelligent risk assessment and surgical scheme recommendation, analysis report generation and interactive optimization and the like, accurate analysis and surgical assistance of tooth images are realized; various advanced algorithms and models are adopted, the accuracy and efficiency of image analysis are improved, and a scientific and reliable reference basis is provided for oral surgery.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image analysis equipment, and specifically discloses an analysis method for tooth images in oral surgery. BACKGROUND

[0002] In current oral surgery, tooth image analysis is a key link in formulating a surgical plan. However, the traditional analysis method relies on the manual operation and experience judgment of doctors, and has many shortcomings. The image shooting position and parameter adjustment rely on manual operation, it is difficult to accurately capture the risk area, and may result in missed shooting or poor image quality; the image preprocessing effect is poor, and noise and blur problems affect the accuracy of subsequent analysis, such as metal artifacts, quantum noise, etc. which are difficult to effectively remove; the segmentation of tooth and surrounding tissue structure is not comprehensive and accurate, and often only part of the key structure can be segmented, and the segmentation effect of the periodontal pocket and other subtle structures is poor; the risk assessment lacks scientific model support, and relies on the experience of doctors, and the accuracy is not high, and it is difficult to generate a personalized surgical plan. These problems may increase the risk of surgery and affect the treatment effect.

[0003] Therefore, the present application provides an analysis method for tooth images in oral surgery in order to solve the above problems. SUMMARY

[0004] The present application aims to solve the problems of relying on manual operation, low efficiency, poor accuracy, incomplete segmentation, and inaccurate risk assessment in the analysis of tooth images in existing oral surgery, and provides an intelligent analysis method for tooth images in oral surgery.

[0005] In order to achieve the above purpose, the basic scheme of the present application provides an analysis method for tooth images in oral surgery, comprising the following steps:

[0006] S001: Periodontal condition assessment and intelligent adjustment of image shooting position: collect periodontal data and locate the risk area, and adjust the position, angle and parameters of the shooting equipment through a mechanical arm;

[0007] S002: Image data acquisition and preprocessing: based on the adjusted position, acquire multi-modal images, and sequentially perform noise processing, blur correction, image enhancement and alignment;

[0008] S003: Intelligent segmentation of tooth and surrounding tissue structure: use a segmentation model to segment the key structures of tooth, alveolar bone, periodontal membrane, mandibular nerve canal, maxillary sinus bottom and periodontal pocket;

[0009] S004: Three-dimensional reconstruction and feature quantization extraction: based on the segmentation model result, perform three-dimensional reconstruction, and extract the quantized features of tooth, bone tissue, periodontal tissue and dangerous structure;

[0010] S005: Intelligent Risk Assessment and Surgical Plan Recommendation: Outputs surgical risk level and risk points through risk assessment model, and generates safe surgical plan and alternative surgical plan;

[0011] S006: Analysis Report Generation and Interaction Optimization: Generate comprehensive analysis reports, supporting doctors to manually adjust and update risk assessment results in real time.

[0012] Furthermore, the noise processing includes: using a MAR-Net network to remove metal artifacts, using an adaptive nonlocal mean filter to suppress quantum noise, and using a wavelet transform threshold denoising method to filter electronic noise.

[0013] Furthermore, the blur correction includes: processing motion blur using Radon transform and Lucy-Richardson algorithm; correcting device jitter blur based on Fourier transform and Wiener filtering; and processing defocus blur using Laplace pyramid decomposition method.

[0014] Furthermore, the segmentation model includes an attention mechanism module, which comprises several labeled image training samples that can accurately segment periodontal pocket structures.

[0015] Furthermore, the quantitative features include periodontal pocket volume, distance between the gingival margin and cementoenamel junction, alveolar ridge deformation displacement under occlusal load, and periodontal ligament stress distribution.

[0016] Furthermore, the risk assessment model includes a random forest algorithm, which, combined with a clinical database of several periodontal-related surgical cases, can assess the risk of poor periodontal tissue healing.

[0017] Furthermore, the comprehensive analysis report includes a periodontal risk heatmap, before-and-after image processing comparison images, and two-dimensional and three-dimensional annotations of periodontal problem areas.

[0018] Furthermore, the attention mechanism also includes spatial attention for focusing on irregular areas of the periodontal pocket wall and channel attention for enhancing the contrast between the periodontal ligament and alveolar bone.

[0019] The principle and effect of this basic scheme are as follows:

[0020] 1. It realizes intelligent adjustment of image shooting position, reduces manual operation, improves the accuracy and efficiency of image shooting, and can accurately capture images of risk areas.

[0021] 2. By employing a variety of advanced noise processing and blur correction methods, various noises and blurs in the images are effectively removed, improving image quality and providing a reliable basis for subsequent analysis.

[0022] 3. The segmentation model achieves accurate segmentation of multiple structures of teeth and surrounding tissues by setting an attention mechanism module, especially significantly improving the segmentation effect of fine structures such as periodontal pockets.

[0023] 4. Based on three-dimensional reconstruction and quantitative feature extraction, it is possible to gain a more comprehensive and intuitive understanding of the condition of teeth and surrounding tissues, providing a quantitative basis for risk assessment.

[0024] 5. The risk assessment model is based on the random forest algorithm and a large number of clinical cases, making risk assessment more scientific and accurate. The generated surgical plans are personalized and targeted, reducing surgical risks.

[0025] 6. The analysis report is comprehensive and supports manual adjustments by doctors, achieving optimized human-computer interaction and improving the report's practicality and reliability. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This illustration shows a step-by-step diagram of a method for analyzing dental images for oral surgery according to an embodiment of this application;

[0028] Figure 2 This illustration shows a schematic diagram of image data acquisition and preprocessing for a method of analyzing dental images for oral surgery proposed in an embodiment of this application;

[0029] Figure 3 A schematic diagram illustrating noise processing of a method for analyzing dental images for oral surgery, as proposed in an embodiment of this application, is shown.

[0030] Figure 4 A schematic diagram illustrating fuzz correction of a method for analyzing dental images for oral surgery proposed in an embodiment of this application is shown.

[0031] Figure 5 A schematic diagram illustrating the attention mechanism of a method for analyzing dental images for oral surgery proposed in an embodiment of this application is shown.

[0032] Figure 6 This illustration shows a schematic diagram of a comprehensive analysis report of a method for analyzing dental images for oral surgery proposed in an embodiment of this application. Detailed Implementation

[0033] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0034] This invention provides an intelligent analysis method for dental images used in oral surgery, implementing, for example... Figure 1 As shown:

[0035] Includes the following steps:

[0036] First, a periodontal assessment and intelligent adjustment of the imaging position are performed: periodontal data is collected and risk areas are located. A robotic arm then adjusts the position, angle, and parameters of the imaging device. Initially, basic periodontal data, such as periodontal pocket depth and gingival recession, is collected using intraoral sensors. Combined with preliminary imaging data, risk areas where periodontal problems may exist are located. Then, based on the identified risk areas, the robotic arm automatically calculates and adjusts the optimal position, angle, and exposure parameters of the imaging device to ensure clear imaging of the risk areas.

[0037] like Figure 2 , 3 The process involves image data acquisition and preprocessing: Multimodal images are acquired based on the adjusted position, followed by noise processing, blur correction, image enhancement, and alignment. Multimodal images include X-ray images and CT images, which can reflect the condition of teeth and surrounding tissues from different angles and levels. Noise processing: A MAR-Net network is used to remove metal artifacts. This network, trained on a large number of image samples containing metal artifacts, can accurately identify and remove them. Adaptive nonlocal mean filtering is used to suppress quantum noise. The filtering parameters are adaptively adjusted according to the noise characteristics of different regions in the image, suppressing noise while preserving image details. Wavelet transform threshold denoising is used to filter electronic noise. By performing wavelet decomposition on the image, different thresholds are applied to coefficients of different frequencies, effectively filtering electronic noise.

[0038] like Figure 4The process also includes blur correction: Motion blur is addressed using Radon transform and the Lucy-Richardson algorithm. First, Radon transform is used to estimate the direction and distance of the motion blur, then the Lucy-Richardson algorithm is used for iterative deconvolution to eliminate motion blur. Equipment jitter blur is corrected based on Fourier transform and Wiener filtering. Fourier transform is performed on the blurred image to analyze its spectral characteristics, and then Wiener filtering is used for correction. Defocus blur is addressed using the Laplacian pyramid decomposition method. The image is decomposed into pyramid layers of different resolutions, each layer is sharpened, and then the image is reconstructed to improve defocus blur. After noise processing and blur correction, the image is enhanced to improve contrast and sharpness. Finally, image alignment technology is used to accurately match multimodal images, ensuring accurate correspondence between different images.

[0039] Intelligent segmentation of teeth and surrounding tissues: Using a segmentation model, key structures such as teeth, alveolar bone, periodontal ligament, mandibular nerve canal, maxillary sinus floor, and periodontal pockets are segmented. For example... Figure 5 The segmentation model incorporates an attention mechanism module, which includes tens of thousands of labeled image training samples capable of accurately segmenting periodontal pocket structures. The attention mechanism also includes spatial attention for focusing on irregular areas of the periodontal pocket walls and channel attention for enhancing the contrast between the periodontal ligament and alveolar bone. Spatial attention allows the model to pay closer attention to irregular areas of the periodontal pocket walls, improving segmentation accuracy; channel attention enhances the contrast between the periodontal ligament and alveolar bone on feature channels, making it easier for the model to distinguish between these two tissues.

[0040] 3D Reconstruction and Feature Quantization Extraction: Based on the segmentation model results, 3D reconstruction is performed to extract quantitative features of teeth, bone tissue, periodontal tissue, and critical structures. Using the segmented 2D images of each tissue structure, a 3D model of the teeth and surrounding tissues is constructed through a 3D reconstruction algorithm, providing a more intuitive representation of the spatial relationships between these structures. Quantitative features include periodontal pocket volume, distance between the gingival margin and cementoenamel junction, alveolar kurtosis displacement under occlusal load, and periodontal ligament stress distribution. These features quantify the physiological and pathological state of the teeth and surrounding tissues.

[0041] Intelligent Risk Assessment and Surgical Plan Recommendation: The risk assessment model outputs the surgical risk level and risk points, generating both a safe surgical plan and a backup plan. The risk assessment model is built based on the random forest algorithm, which, combined with a clinical database of tens of thousands of periodontal-related surgical cases, can assess the risk of poor periodontal tissue healing. The risk assessment model first analyzes the extracted quantitative features, combining them with cases from the clinical database to identify potential surgical risk points and classify the risk levels according to their severity. Then, based on the risk level and risk points, a safe surgical plan is generated for the patient, while a backup surgical plan is generated considering possible unforeseen circumstances.

[0042] Analysis report generation and interaction optimization: such as Figure 6 The system generates a comprehensive analysis report, allowing doctors to manually adjust and update the risk assessment results in real time. This report includes a periodontal risk heatmap, before-and-after image processing comparisons, and two-dimensional and three-dimensional annotations of periodontal problem areas. Doctors can manually adjust the risk assessment results and surgical plans in the report based on their experience and judgment. The system updates the risk assessment results in real time after adjustments, ensuring the accuracy and applicability of the report.

[0043] To further illustrate the technical solution of the present invention, the following detailed description is provided in conjunction with specific embodiments: Periodontal condition assessment and intelligent adjustment of imaging position: 100 oral surgery patients were selected, and their periodontal data were collected using intraoral sensors. Risk areas were located using data processing software. Then, a robotic arm was activated, and the robotic arm automatically adjusted the position, angle, and parameters of the imaging device based on the positioning results to capture images of each patient.

[0044] Image Data Acquisition and Preprocessing: X-ray and CT images of each patient were acquired as multimodal images. Noise Reduction: The MAR-Net network was used to process images containing metal artifacts, achieving a metal artifact removal rate of over 90%. Adaptive nonlocal mean filtering was used to process quantum noise, improving the signal-to-noise ratio by an average of 20 dB. Wavelet transform threshold denoising was used to process electronic noise, resulting in significant noise removal while preserving image details. Blur Correction: For images with motion blur, the blur direction and distance were estimated using Radon transform, followed by 10 iterations of the Lucy-Richardson algorithm, effectively eliminating motion blur. For images blurred by equipment jitter, Fourier transform and Wiener filtering reduced blur and improved sharpness. For defocused images, Laplacian pyramid decomposition was used, improving edge sharpness. Enhancement processing was applied to the processed images. Histogram equalization was used to improve image contrast, and then an image alignment algorithm was used to align X-ray and CT images, reducing alignment errors.

[0045] Intelligent segmentation of teeth and surrounding tissues: Preprocessed images are input into the segmentation model, which includes an attention mechanism module with 500 labeled training samples. Spatial attention focuses on irregular areas of the periodontal pocket wall, while channel attention enhances the contrast between the periodontal ligament and alveolar bone. This segmentation effectively segments teeth, alveolar bone, periodontal ligament, mandibular canal, maxillary sinus floor, and periodontal pockets, achieving an accuracy rate exceeding 95%. 3D reconstruction and feature quantification extraction: Based on the segmentation results, 3D reconstruction software is used to construct a 3D model of the teeth and surrounding tissues. Quantitative features such as periodontal pocket volume, distance between the gingival margin and cementoenamel junction, alveolar kurtosis displacement under occlusal load, and periodontal ligament stress distribution are extracted and recorded for each patient.

[0046] Intelligent Risk Assessment and Surgical Plan Recommendation: Extracted quantitative features are input into a risk assessment model based on a random forest algorithm, combined with a clinical database of 1000 periodontal-related surgical cases. The model outputs the surgical risk level and risk points for each patient, categorizing them into 20 high-risk patients, 50 medium-risk patients, and 30 low-risk patients. For each patient, a safe surgical plan and a backup surgical plan are generated, including surgical steps, key operational points, potential risks, and corresponding countermeasures. Analysis Report Generation and Interaction Optimization: A comprehensive analysis report is generated for each patient, including a periodontal risk heatmap, before-and-after image processing comparisons, and 2D and 3D annotations of periodontal problem areas.

[0047] Doctors manually adjust the risk assessment results and surgical plans in the report based on their clinical experience. The system updates the risk assessment results in real time after the adjustments are made, and the adjusted report is verified by the doctor.

[0048] The present invention has the following advantages:

[0049] 1. Precise positioning: The robotic arm's intelligent adjustment improves the capture rate of high-risk areas;

[0050] 2. High-quality images: Multi-level noise processing and blur correction ensure image quality;

[0051] 3. Fine segmentation: Dual attention mechanism enables fine segmentation of periodontal pockets;

[0052] 4. Quantitative decision-making: Risk assessment model based on biomechanical characteristics;

[0053] 5. Human-machine collaboration: Interactive reporting supports clinical decision optimization.

[0054] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for analyzing dental images for oral surgery, characterized in that, Includes the following steps: S001: Periodontal condition assessment and intelligent adjustment of imaging position: Collect periodontal data and locate risk areas, and adjust the position, angle and parameters of the imaging equipment through a robotic arm; S002: Image data acquisition and preprocessing: Based on the adjusted position, multimodal images are acquired and noise processing, blur correction, image enhancement and alignment are performed in sequence; S003: Intelligent segmentation of teeth and surrounding tissues: Using a segmentation model, key structures such as teeth, alveolar bone, periodontal ligament, mandibular nerve canal, maxillary sinus floor, and periodontal pockets are segmented. S004: 3D Reconstruction and Feature Quantization Extraction: Based on the segmentation model results, perform 3D reconstruction and extract quantitative features of teeth, bone tissue, periodontal tissue and dangerous structures; S005: Intelligent Risk Assessment and Surgical Plan Recommendation: Outputs surgical risk level and risk points through risk assessment model, and generates safe surgical plan and alternative surgical plan; S006: Analysis Report Generation and Interaction Optimization: Generate comprehensive analysis reports, supporting doctors to manually adjust and update risk assessment results in real time.

2. The method for analyzing dental images for oral surgery according to claim 1, characterized in that, The noise processing includes: using a MAR-Net network to remove metal artifacts, using an adaptive nonlocal mean filter to suppress quantum noise, and using a wavelet transform threshold denoising method to filter electronic noise.

3. The method for analyzing dental images for oral surgery according to claim 1, characterized in that, The blur correction includes: processing motion blur using Radon transform and Lucy-Richardson algorithm; correcting device jitter blur based on Fourier transform and Wiener filtering; and processing defocus blur using Laplace pyramid decomposition method.

4. The method for analyzing dental images for oral surgery according to claim 1, characterized in that, The segmentation model includes an attention mechanism module, which comprises several labeled image training samples that can accurately segment periodontal pocket structures.

5. The method for analyzing dental images for oral surgery according to claim 1, characterized in that, The quantitative characteristics include periodontal pocket volume, distance between the gingival margin and cementoenamel junction, alveolar ridge deformation displacement under occlusal load, and periodontal ligament stress distribution.

6. The method for analyzing dental images for oral surgery according to claim 1, characterized in that, The risk assessment model includes a random forest algorithm, which, combined with a clinical database of several periodontal-related surgical cases, can assess the risk of poor periodontal tissue healing.

7. The method for analyzing dental images for oral surgery according to claim 1, characterized in that, The comprehensive analysis report includes a periodontal risk heat map, before-and-after image processing comparisons, and two-dimensional and three-dimensional annotations of periodontal problem areas.

8. The method for analyzing dental images for oral surgery according to claim 4, characterized in that, The attention mechanism also includes spatial attention for focusing on irregular areas of the periodontal pocket wall and channel attention for enhancing the contrast between the periodontal ligament and alveolar bone.

Images