CN-122000037-A - Multi-parameter analysis-based maxillary first molar extraction type decision method and system

CN122000037ACN 122000037 ACN122000037 ACN 122000037ACN-122000037-A

Abstract

The invention discloses a multi-parameter analysis-based maxillary first molar extraction type decision method and system. The method comprises the steps of firstly obtaining age and full view pictures of a patient, then segmenting the full view pictures through a trained DeepLabV3+ semantic segmentation model, extracting the outer edges of the near-middle cheek roots, the inner edges of the near-middle cheek roots, the outer edges of the far-middle cheek roots, the inner edges of the far-middle cheek roots and root bifurcation points, calculating the minimum curvature radius R of the tooth roots, the bifurcation included angle theta and the local standardized bone density index D based on the extracted outline, finally entering a preset hierarchical decision engine according to the age of the patient, automatically outputting an operation mode of suggesting the whole extraction or the divided extraction by combining the quantized parameters, and generating a decision basis and a quantized parameter report. The invention combines deep learning feature extraction with a clinical rule engine, realizes objective quantitative decision and personalized recommendation of tooth extraction, and improves the accuracy and the interpretability of the decision.

Inventors

HUANG XIN
Zhang Qinruo
ZHANG TINGTING
ZHANG YAJING
HAN JIANMIN

Assignees

天津医科大学口腔医院

Dates

Publication Date: 20260508
Application Date: 20260409

Claims (10)

1. A multi-parameter analysis-based maxillary first molar extraction decision method, comprising the steps of: Step S1, acquiring the age A of a patient and an oral cavity full-view film I; S2, inputting the oral cavity full view slice I into a semantic segmentation model, and extracting a near-middle cheek root outer edge, a near-middle cheek root inner edge, a far-middle cheek root outer edge, a far-middle cheek root inner edge and root bifurcation points; Step S3, calculating a minimum curvature radius R of the tooth root based on the extracted outer edges of the near and far cheek roots, calculating a bifurcation angle theta based on the extracted inner edges of the near and far cheek roots and bifurcation points, and calculating a local standardized bone density index D based on the extracted outer edges of the near and far cheek roots; And S4, comparing the age A of the patient with a preset age threshold, selecting a corresponding decision sub-module according to a comparison result, calling the curvature radius R, the included angle theta and the bone mineral density index D in the decision sub-module, and outputting a main decision D_primary through a preset decision tree logic or a weighted scoring model.
2. The multi-parameter analysis based first molar extraction decision method according to claim 1, wherein the semantic segmentation model in step S2 is a DeepLabV3+ model trained to perform six categories of semantic segmentation tasks, background, mesial-facial root lateral margin, mesial-facial root medial margin, distal mesial-facial root lateral margin, distal facial root medial margin, and root bifurcation, respectively.
3. The method according to claim 1, wherein calculating the minimum curvature radius R in the step S3 comprises smoothing the extracted lateral edges of the mesial and distal cheek roots, calculating the instantaneous curvature k of each point on the contour line, and taking the point with the largest curvature absolute value among all points, wherein the maximum instantaneous curvature is k_max, and the curvature radius r=1/|k_max|, and converting the maximum curvature radius into a physical unit in combination with the pixel-millimeter scale of the full scene.
4. The method according to claim 1, wherein calculating the bifurcation angle θ in step S3 includes positioning four points 2mm and 4mm from the coronal side of the bifurcation point on the medial edge of the mesial cheek root and the medial edge of the distal cheek root, respectively, connecting the points 2mm and 4mm from the coronal side to obtain vectors v_mb and v_db representing the two central trends, and calculating the angle between the two vectors as bifurcation angle θ.
5. The method according to claim 1, wherein the calculating the normalized bone density index D in the step S3 specifically includes generating a region of interest by expanding the region of interest to the outside with the medial lateral edge of the cheek root and the lateral edge of the cheek root as a base line, selecting a reference region on the panoramic sheet, and calculating a normalized Z-score of a pixel mean value in the region of interest relative to a pixel mean value in the reference region as the bone density index D.
6. The multi-parameter analysis based maxillary first molar extraction decision method according to claim 1, wherein step S4 specifically comprises: The first-stage flow distribution comprises the steps of directly outputting a proposal root division pulling out if A is more than or equal to 60, entering a middle-young layer comprehensive decision sub-module if A is more than or equal to 35 and entering a young layer form decision sub-module if A is less than or equal to 60; In the middle-young layer comprehensive decision sub-module, if R is less than or equal to 8mm, outputting a 'suggested root-splitting plucking', if R is more than 8mm and theta is more than 20 degrees, starting a weighted scoring model S=w1X (theta-20) +w2xD+w3X (A-35), if S is more than or equal to a threshold value, outputting a 'suggested root-splitting plucking', and otherwise outputting a 'tendency overall plucking', wherein w1, w2 and w3 are weights obtained through historical clinical data training; in the young layer morphology decision sub-module, if R is less than or equal to 8mm or theta is more than 20 degrees, outputting a 'suggested root-splitting plucking', otherwise outputting a 'suggested whole plucking'.
7. The multi-parameter analysis based maxillary first molar extraction decision method according to claim 5 wherein the reference area is the mandibular lower margin compact cortical bone area when calculating the normalized bone density index D in step S3.
8. The multi-parameter analysis based first molar extraction decision method according to claim 1, further comprising the step of Savitzky-Golay filter smoothing the extracted contour lines before calculating the minimum radius of curvature R in step S3.
9. The multi-parameter analysis based maxillary first molar extraction decision method of claim 6, wherein the weights w1, w2, w3 in the weighted scoring model are obtained by logistic regression training on historical clinical data.
10. A multi-parameter analysis-based maxillary first molar extraction decision system comprising: the input module is used for acquiring the age A of the patient and the oral cavity full-view film I; The feature extraction module is used for extracting the tooth root outline and the root bifurcation point from the oral cavity full view slice I based on the semantic segmentation model, and calculating the curvature radius R, the included angle theta and the bone density index D; the decision engine module is used for selecting a corresponding decision path according to the age A of the patient, combining the R, the theta and the D, and outputting a main decision D_primary according to a preset decision logic; and the output module is used for outputting the main decision D_primary, the decision basis and the quantized parameter report.

Description

Multi-parameter analysis-based maxillary first molar extraction type decision method and system Technical Field The invention relates to the technical field of artificial intelligence auxiliary diagnosis in stomatology, in particular to a first molar extraction type decision method and system based on multi-parameter analysis. Background The first molar of the upper jaw is one of the permanent teeth that erupts earliest in the mouth, has the highest caries rate and is most at risk of loss due to periodontal disease, and extraction is one of the most common operations in oromaxillofacial surgery. The maxillary first molar typically has three roots, namely a mesial, distal and palatine root, where mesial and distal buccal roots are commonly referred to as "buccal bicuspids". The anatomy of the buccal double root is complex and variable, and the bending degree, bifurcation angle and peri-osseous condition are key factors for determining the extraction operation. At present, clinically, for the extraction of the palate root and the buccal double root of the first molar of the upper jaw, two operation modes mainly exist, namely, one operation mode is that the whole teeth are extracted directly after the gap is increased and loosened, and the other operation mode is that the teeth are extracted in a root-dividing mode, namely, the palate root and the buccal double root are firstly cut and separated in a T-shaped mode by a drill needle, and then extracted one by one. What kind of surgery is chosen is directly related to the size of the surgical wound, the time of surgery, the risk of postoperative complications and the patient's post-healing experience. Improper selection may lead to root breakage, alveolar bone injury, post-operative infection and even serious complications to the upper jaw Dou Chuantong. In the prior art, doctors rely mainly on preoperatively photographed oral full-length shots and combine with individual clinical experience to make decisions. And a doctor can subjectively judge the curvature, bifurcation angle and peri-radicular bone density of the tooth root by visually observing the full-sight lens. However, there are a number of limitations to this conventional decision approach. Firstly, the naked eye evaluation lacks a unified quantification standard, and the interpretation results of the same image by different doctors can have larger difference, even the interpretation of the same doctor at different times can be inconsistent, so that the subjectivity of the decision is stronger, and the standardization and homogenization of treatment are difficult to ensure. Second, it is difficult for the unaided human eye to precisely quantify the fine curvature and bifurcation angle of the tooth root, and the judgment of critical conditions is prone to bias, e.g., the definition of "slight curvature" tends to vary from person to person. Furthermore, the evaluation of bone mineral density is highly dependent on personal perception of doctors on two-dimensional full-view pictures, and lacks objective quantitative indexes to eliminate image differences caused by different equipment and different exposure parameters. In addition, the age of the patient is an important biological factor, the bone of the young patient is harder, the tooth root is complete, the tooth extraction resistance is higher, the bone of the old patient is relatively loose, the periodontal ligament gap is widened, the tooth extraction resistance is lower, the healing capacity is reduced, and the old patient is more sensitive to surgical wounds. The existing decision method cannot effectively conduct fusion analysis on the age factors and the anatomical form parameters. Searching the prior patent literature and non-patent literature found that although studies on root anatomy morphology measurement, such as measuring bifurcation angle, root height, etc. by ex vivo teeth, these studies have remained largely at the level of basic anatomy description and have not been converted into clinically useful auxiliary decision tools. There are some patent applications related to intelligent analysis of oral cavity images, but most focus on caries detection, periodontal disease diagnosis or implant planning, and no method and system for performing intelligent decision by fusing age stratification and panoramic film multiple parameters specifically aiming at the first molar extraction of the upper jaw are known. Therefore, there is a need for a method that can objectively, quantitatively, and individually provide surgical decision support for first molar extraction of the upper jaw, so as to overcome the drawbacks of the prior art that the subjective experience is relied on, and the unified standard and the quantitative basis are lacking, reduce the risk of surgery, and improve the success rate of treatment. Disclosure of Invention In order to solve the technical problems, the invention provides a method and a system for determining a first molar