CN115690045A - Quantitative assessment method for bone increment before and after periodontitis treatment based on curved surface fault slice - Google Patents
Quantitative assessment method for bone increment before and after periodontitis treatment based on curved surface fault slice Download PDFInfo
- Publication number
- CN115690045A CN115690045A CN202211347246.7A CN202211347246A CN115690045A CN 115690045 A CN115690045 A CN 115690045A CN 202211347246 A CN202211347246 A CN 202211347246A CN 115690045 A CN115690045 A CN 115690045A
- Authority
- CN
- China
- Prior art keywords
- curved surface
- treatment
- alveolar bone
- surface fault
- slice
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
The invention relates to a quantitative assessment method for bone increment before and after periodontitis treatment based on a curved surface fault slice, which comprises the steps of firstly extracting tooth feature point pairs of a part to be assessed in an oral cavity curved surface fault slice obtained twice before and after treatment, automatically extracting feature points of edge points, and matching the feature points of the edge points in two images; constructing a registration model to transform the treated oral cavity curved surface fault layer; and carrying out local slicing processing on the registered treated oral cavity curved surface fracture layer according to the position of the characteristic point pair to obtain a registered slice, carrying out image gray segmentation on the registered slice to obtain a plurality of communicated areas with consistency, automatically selecting the alveolar bone change target position, and calculating the alveolar bone increment area by utilizing the area pixel number and the real resolution of the curved surface fracture layer. The method can accurately convert the picture after treatment to the position locally aligned with the picture before treatment, and automatically extract the change area, thereby obtaining the accurate alveolar bone incremental area.
Description
Technical Field
The invention relates to a quantitative assessment method for bone increment before and after periodontitis treatment based on curved surface fault slices, and belongs to the technical field of novel imaging of alveolar bone increment assessment.
Background
The change of the alveolar bone height is an important clinical index influencing the stage grading of periodontitis and is also one of important clinical evaluation indexes for evaluating the treatment effect. The loss of alveolar bone in the anterior dental area can cause the sector displacement of teeth, which affects the beauty and the cutting and occluding functions; the loss of teeth in the posterior dental area due to alveolar bone resorption and the like seriously affects the chewing function and efficiency. The bone absorption to a certain extent leads to the formation of deep periodontal pockets, the bacteria and toxin in the pockets cause the continuous development of periodontitis, the level of inflammation is increased, the alveolar bone is further absorbed, and the vicious circle is formed. Studies have shown that periodontal treatment improves the level of periodontal inflammation, promoting alveolar bone repair and regeneration to varying degrees.
The imaging evaluation is an indispensable important component for periodontal clinical evaluation, can provide information on the past destruction of periodontal hard tissues, and has important value for longitudinally measuring progressive bone loss and alveolar bone regeneration. In the process of periodontitis treatment, in order to observe alveolar bone absorption and regeneration more intuitively, oral curved surface broken-line slices are often required to be shot to carry out comparative observation on the reconstruction process of the alveolar bone at multiple parts of the whole mouth. The oral curved surface fracture sheet not only gives the doctor and the patient the overall cognition of the condition of the full-mouth alveolar bone, but also has important significance for judging alveolar bone reconstruction before and after periodontal treatment, tooth retention probability, prognosis evaluation, drawing up a later-period periodontal treatment scheme and the like. In addition, the shooting of the oral cavity curved surface broken layer is simple and convenient, the price is easy to accept, and the accuracy and repeatability are better, and the advantages are also one of important factors for wide application. In addition, compared with 3D images such as CBCT (cone beam computed tomography) and the like, the oral curved surface fault layer has the advantages of simple imaging and low cost, and can be used as an important auxiliary means for periodontitis monitoring to detect the change of structures such as alveolar bones and the like more easily, so that the method has important significance for detecting and researching the change of the oral curved surface fault layer.
Registration is a prerequisite for multi-temporal image sequence processing. The concept of registration is defined as an image processing method in which coordinate transformation is performed on coordinate systems of different images with reference to one of a standard image or an existing image so that objective positions of pixels in the image are as consistent as possible. In the field of medical image processing, image registration can be divided into single/multi-modality registration, single-patient registration, multi-patient registration, and the like according to the attributes of input images; according to the dimensionality of an input image, the method can be divided into 3D-3D registration, 2D-2D registration, 2D-3D registration and the like; according to the image registration transformation model, the method can be divided into rigid model registration, affine transformation model registration, deformable model registration and the like; image registration has been applied in a number of medical research fields, such as registering a brain MRI/CT image with a standard brain model to obtain various regional divisions of an actual image to assist in diagnosis and observation of therapeutic effects of brain diseases; and (3) registering the CT lung images before and after treatment to track and observe lung tissues in the respiratory process so as to cooperate with treatment of lung diseases. However, at present, there are few registration methods specially designed for oral cavity images, and since the oral cavity curved surface slice contains a moving mandible and the shooting mode is central rotation imaging, it is often difficult to directly apply the general image registration method.
After the images are completely registered, the method of image segmentation is required to accurately extract the changed area of the alveolar bone. The extraction of the image region of interest is always an important research direction in image processing, the current image region of interest extraction method generally adopts an image segmentation means, and the image segmentation method can be divided into two categories, namely a traditional segmentation method and a segmentation method based on deep learning. The traditional segmentation method generally uses texture features or gray features to perform segmentation by utilizing the uniformity and homogeneity of the same region of an image, has the advantages of convenience and rapidness, does not need or only needs less marked images as priori knowledge, but has extraction effect depending on feature effectiveness and accuracy of manual design and is often low in universality; the segmentation method based on deep learning utilizes a deep neural network to automatically extract and learn the features, can get rid of the constraint of artificial design features, but has the premise that more marked samples are needed to be used as training samples, the current neural network has relatively large mass, the training network needs longer time, and the marking cost is higher. At present, methods for extracting alveolar bone change regions for oral curved surface fault slices are still rare, and because alveolar bone changes often exist only in regions with severe alveolar bone recession and the regions often have high discontinuity, the extraction method for global interested regions is generally difficult to directly apply.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a method which can convert a curve fracture sheet after treatment to a position which is locally and accurately aligned with the curve fracture sheet before treatment so as to obtain accurate alveolar bone augmentation quantification.
In order to achieve the purpose, the technical scheme provided by the invention is as follows: a quantitative assessment method for bone increment before and after periodontitis treatment based on curved surface fault slices comprises the following steps:
step one, preprocessing oral cavity curved surface broken slice before and after treatment;
step two, extracting M tooth characteristic point pairs of an alveolar bone variation evaluation area in the oral cavity curved surface fault layer sheet as basic characteristic point pairs, and extracting an edge point setN is the total number of the edge points;
step three, extracting edge point setMatching the characteristic points of the edge points of the two curved surface fault pieces before and after treatment;
combining the obtained feature point matching result of the edge point with the basic feature point pair to generate a rough matching point pair set;
establishing the following initial registration model:
wherein (x) r Yr) is the pixel coordinates in the pre-treatment tomographic image, and (x, y) is the pixel coordinates in the post-treatment tomographic image;
randomly selecting a plurality of groups of characteristic point pairs from the coarse matching point pair set each time, and calculating a transformation matrix H; then, the number of all matching points satisfying the following formula is calculated:
wherein t is a self-defined threshold;
substituting the transformation matrix H corresponding to the maximum matching point number satisfying the formula into the initial registration model to obtain a final registration model;
step five, transforming the post-treatment oral curved surface fault layer sheet by using the final registration model to obtain a post-treatment oral curved surface fault layer sheet aligned with the pre-treatment oral curved surface fault layer sheet;
sixthly, local slicing is carried out on the registered treated oral cavity curved surface fault layer sheet according to the position of the basic characteristic point pair to obtain a registered slice, and image gray segmentation is carried out on the registered slice;
seventhly, obtaining a plurality of communicated areas with consistency after image gray segmentation, selecting a target area with alveolar bone change, and obtaining the pixel number p of the area with alveolar bone change;
calculating the alveolar bone incremental area S by using the number p of the area pixels and the real resolution R of the curved surface fault layer:
S=p×R。
the technical scheme is further designed as follows: the preprocessing in the first step comprises histogram normalization and median filtering. The characteristic point pairs selected in the step two comprise tooth cusps in an alveolar bone change evaluation area, high points of the external forms of the near and far middle crowns, enamel cementum boundary, root cusps, root bifurcation areas and other tooth anatomical forms which are easy to recognize.
Extracting the edge point set by using a scale invariant feature transform descriptor in the third stepMatching the characteristic points of the edge points of the two curved surface fault layers before and after treatment by using a two-nearest neighbor method under the following constraint condition;
1) The ratio of Euclidean distance from the matched edge point to the nearest neighbor point to the Euclidean distance from the edge point to the next neighbor point is smaller than a threshold lambda;
2) Calculating the distances between the description feature vectors of the edge points and all the basic feature points, and matching the nearest neighbor basic feature points corresponding to the edge points which are matched with each other;
3) The euclidean distance of the matched edge point from the nearest base feature point should be less than the threshold value l.
In the sixth step, a generalized Ostu method is adopted to carry out image gray segmentation, and the specific method is as follows:
wherein k is 1 ,k 2 ,…,k N For N division thresholds, N i Is the number of pixels contained in the ith segment under the current segmentation, N is the total number of pixels, mu i And μ is the mean value of the gray values of all pixels in the ith segment under the current segmentation.
The invention has the beneficial effects that:
the registration part of the oral cavity curved surface fault layer sheet in the method can ensure that the local registration precision error is within 1 pixel, so that the picture after treatment can be ensured to be converted to the position locally and accurately aligned with the picture before treatment. And then, the image segmentation can extract more accurate alveolar bone change regions and provide corresponding specific areas.
The method is based on the traditional image automatic segmentation method, so that pre-training of a model is not needed, the segmentation and registration can give a result within 1s, and an image processing result can be output in real time according to requirements. The obtained quantitative evaluation result can comparatively accurately and intuitively evaluate the loss and increase of the alveolar bone mass in a contrast way, and a diagnosis and treatment basis is provided for a long-term treatment scheme and prognosis of periodontal treatment.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a slice of a patient's pre-treatment oral cavity surface profile;
FIG. 3 is a slice of a curved surface of the mouth after treatment of a patient;
FIG. 4 shows the result of the superposition of the oral cavity curved surface slices before and after the treatment of a patient;
FIG. 5 shows the superimposed results after registration of the positions of the second premolar and the first molar in the posterior dental area of the patient;
fig. 6 shows the result of alveolar bone increment detection between the second premolar and the first molar in the posterior dental area of the patient.
Detailed Description
The invention is described in detail below with reference to the drawings and specific embodiments.
Examples
The method for quantifying alveolar bone increment before and after periodontal treatment based on the oral curved surface fault slice comprises the following specific steps as shown in figure 1,
two images are shown in fig. 2 and 3, taking as an example two oral cavity curved surface fault slices three years apart before and after periodontal treatment of a patient. Through the research and judgment of doctors, obvious bone mass increase exists between the second anterior molar and the first molar in the posterior dental area, and the treatment effectiveness is prompted. However, as shown in fig. 4, the result of the superposition of the two images shows that the two images cannot be directly aligned with each other tooth position, and it is difficult to quantitatively evaluate the increment of the alveolar bone. Taking the evaluation of the alveolar bone variation between the second anterior molar and the first molar in the posterior dental area as an example, the specific scheme of the method is as follows:
the method comprises the following steps of firstly, preprocessing two curved surface fault layers to improve the visual effect of an image and align gray distribution, wherein the preprocessing comprises histogram standardization and median filtering.
And step two, manually extracting 8 tooth feature point pairs expected to enter an alveolar bone variation evaluation area, wherein the 8 tooth feature point pairs comprise a cementum enamel boundary, a root tip and a first molar root bifurcation area of two teeth. In addition, an edge point set is extracted using an edge extraction methodN is the total number of edge points.
Step three, taking the feature point pairs obtained by manual extraction as basic feature point pairs, and in order to improve the registration accuracy, performing constraint pairingAnd matching the point pairs. Extracting feature points of each edge point by using a Scale-invariant feature transform (SIFT) descriptor, and matching by using a two-neighbor method, but simultaneously carrying out the following constraints:
1) The ratio of the Euclidean distance from the matched edge point to the nearest neighbor point to the Euclidean distance from the point to the next neighbor point is less than a threshold value lambda =0.4;
2) Calculating the distance between the description feature vector of the edge point and all the basic feature points, wherein the nearest neighbor basic feature points corresponding to the edge point are matched with the nearest neighbor basic feature points corresponding to the matching point;
3) The euclidean distance of the matched edge point in the graph from the nearest manually extracted feature point should be less than the threshold value l =20.
And step four, combining the obtained edge point matching result with the basic characteristic point pair to generate a rough matching point pair set. And further screening the matching point pairs and calculating a registration model by adopting a random sample consensus (RANSAC) mode. Because the oral cavity curved surface fault layer is shot by the lower jaw position bone structure, and the local part can be regarded as rigid body perspective transformation, the initial registration model is set as a perspective model:
wherein (x) r ,y r ) The coordinates of the pixel points in the pre-treatment image and the coordinates of the pixel points in the post-treatment image (x, y).
The specific process of the RANSAC method is to randomly select 4 groups of feature points from a set of coarse matching point pairs each time, calculate a transformation matrix H by using a Singular Value Decomposition (SVD) method, and calculate the number of all matching points satisfying the following formula:
where t is a self-defined threshold, let t =1 in this example. Equal in number to pairs of characteristic points. And after multiple calculations, selecting a transformation matrix H corresponding to the maximum point number satisfying the formula, and substituting the transformation matrix H into the initial registration model to obtain a final registration model.
And step five, transforming the post-treatment oral curved surface fault layer slice by using the final registration model to obtain a registered post-treatment oral curved surface fault layer slice which is completely aligned with the pre-treatment image, as shown in fig. 5. It can be seen that precise alignment is achieved near the treatment site (within the box) of the posterior dental zone. Through the operation, the image after the treatment and the image before the treatment are registered, and the local registration precision error is within 1 pixel, so that the image is transformed to a position which is locally and accurately aligned with the image before the treatment.
And sixthly, local slicing processing of the curved surface fault slice is carried out on the position of the base characteristic point, and for the slice after registration, a generalized Ostu method is adopted to carry out image gray segmentation:
wherein k is 1 ,k 2 ,…,k N N =5 in this example for N division thresholds. N is a radical of hydrogen i Is the number of pixels contained in the ith segment under the current segmentation, N is the total number of pixels, mu i And μ is the mean value of the gray values of all pixels in the ith segment under the current segmentation.
And seventhly, obtaining a plurality of connected areas with consistency after image segmentation, and obtaining the concerned alveolar bone change area information including the area outline and the area pixel number by interactively selecting the target position of the alveolar bone change by a user, so as to mark in the original image, wherein the obtained alveolar bone change outline is shown in fig. 6. It can be seen that the embodiment can accurately detect the contour of the alveolar bone change position, thereby measuring and obtaining information such as area.
R (cm) is measured by using the number p of area pixels and the real resolution of the curved surface fault layer sheet (a ruler can be added during shooting) 2 Per pixel) calculating alveolar bone incremental area S:
S=p×R
in this example, S = p × R =709 × 1.33 × 10 -4 =0.0942cm 2 。
The embodiment can extract the more accurate alveolar bone change area, give the corresponding specific area, basically accord with the actual alveolar bone increment, and has small error.
The technical solutions of the present invention are not limited to the above embodiments, and all technical solutions obtained by using equivalent substitution modes fall within the scope of the present invention.
Claims (5)
1. A quantitative assessment method for bone increment before and after periodontitis treatment based on curved surface fault slices is characterized by comprising the following steps: the method comprises the following steps:
step one, preprocessing the oral cavity curved surface fault slice before and after treatment;
step two, extracting M tooth characteristic point pairs of an alveolar bone variation evaluation area in the oral cavity curved surface fault layer sheet as basic characteristic point pairs, and extracting an edge point setN is the total number of the edge points;
step three, extracting edge point setMatching the characteristic points of the edge points of the two curved surface fault pieces before and after treatment;
combining the obtained feature point matching result of the edge point with the basic feature point pair to generate a rough matching point pair set;
establishing the following initial registration model:
wherein (x) r ,y r ) The coordinates of the pixel points in the image of the pre-treatment tomographic slice, and the coordinates of the pixel points in the image of the post-treatment tomographic slice (x, y);
randomly selecting a plurality of groups of characteristic point pairs from the coarse matching point pair set each time, and calculating a transformation matrix H; then, the number of all matching points satisfying the following formula is calculated:
wherein t is a self-defined threshold;
substituting the transformation matrix H corresponding to the maximum number of matching points meeting the formula into the initial registration model to obtain a final registration model;
step five, transforming the post-treatment oral curved surface fault layer sheet by using the final registration model to obtain a post-treatment oral curved surface fault layer sheet aligned with the pre-treatment oral curved surface fault layer sheet;
sixthly, local slicing is carried out on the registered treated oral cavity curved surface fault layer sheet according to the position of the basic characteristic point pair to obtain a registered slice, and image gray segmentation is carried out on the registered slice;
seventhly, obtaining a plurality of communicated areas with consistency after image gray segmentation, selecting a target area with alveolar bone change, and obtaining the pixel number p of the area with alveolar bone change;
calculating the alveolar bone incremental area S by using the number p of the area pixels and the real resolution R of the curved surface fault layer:
S=p×R。
2. the method for quantifying alveolar bone increment before and after periodontal treatment based on the oral curved surface fault slice according to claim 1, wherein: the preprocessing in the first step comprises histogram normalization and median filtering.
3. The method for quantifying alveolar bone increment before and after periodontal treatment based on the oral curved surface fault slice as claimed in claim 1, wherein the method comprises the following steps: and the characteristic point pairs selected in the step two comprise tooth cusps in alveolar bone change evaluation areas, high points of the external shapes of the near and far middle crowns, enamel cementum boundaries, root cusps, root bifurcation areas and other tooth anatomical shapes which are easy to identify.
4. The method for quantifying alveolar bone increment before and after periodontal treatment based on the oral curved surface fault slice as claimed in claim 1, wherein the method comprises the following steps: extracting the edge point set by using a scale invariant feature transform descriptor in the third stepMatching the feature points of the edge points of the two curved surface fault pieces before and after treatment by using a two-neighbor method under the following constraint condition;
1) The ratio of Euclidean distance from the matched edge point to the nearest neighbor point to the Euclidean distance from the edge point to the next neighbor point is smaller than a threshold lambda;
2) Calculating the distances between the description feature vectors of the edge points and all basic feature points, wherein nearest neighbor basic feature points corresponding to the edge points which are matched with each other;
3) The euclidean distance of the matched edge point from the nearest base feature point should be less than the threshold value l.
5. The method for quantifying alveolar bone increment before and after periodontal treatment based on the oral curved surface fault slice as claimed in claim 1, wherein the method comprises the following steps: in the sixth step, a generalized Ostu method is adopted to carry out image gray segmentation, and the specific method is as follows:
wherein k is 1 ,k 2 ,…,k N For N division thresholds, N i Is the number of pixels contained in the i-th segment under the current segmentation, N is the total number of pixels, mu i And μ is the mean value of the gray values of all pixels in the ith segment under the current segmentation.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211347246.7A CN115690045A (en) | 2022-10-31 | 2022-10-31 | Quantitative assessment method for bone increment before and after periodontitis treatment based on curved surface fault slice |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211347246.7A CN115690045A (en) | 2022-10-31 | 2022-10-31 | Quantitative assessment method for bone increment before and after periodontitis treatment based on curved surface fault slice |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115690045A true CN115690045A (en) | 2023-02-03 |
Family
ID=85046933
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211347246.7A Pending CN115690045A (en) | 2022-10-31 | 2022-10-31 | Quantitative assessment method for bone increment before and after periodontitis treatment based on curved surface fault slice |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115690045A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117152238A (en) * | 2023-10-25 | 2023-12-01 | 天津医科大学口腔医院 | Automatic anterior dental zone alveolar bone area measuring method and system based on deep learning |
-
2022
- 2022-10-31 CN CN202211347246.7A patent/CN115690045A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117152238A (en) * | 2023-10-25 | 2023-12-01 | 天津医科大学口腔医院 | Automatic anterior dental zone alveolar bone area measuring method and system based on deep learning |
CN117152238B (en) * | 2023-10-25 | 2024-02-09 | 天津医科大学口腔医院 | Automatic anterior dental zone alveolar bone area measuring method and system based on deep learning |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Claes et al. | Computerized craniofacial reconstruction: conceptual framework and review | |
Shui et al. | Densely calculated facial soft tissue thickness for craniofacial reconstruction in Chinese adults | |
Tilotta et al. | Construction and analysis of a head CT-scan database for craniofacial reconstruction | |
Tian et al. | DCPR-GAN: dental crown prosthesis restoration using two-stage generative adversarial networks | |
CN110946652B (en) | Method and device for planning screw path of bone screw | |
CN106447764B (en) | A kind of human pelvis parameter auto-testing method of three-dimensional visualization | |
Dai et al. | Locating anatomical landmarks on 2D lateral cephalograms through adversarial encoder-decoder networks | |
CN110811653A (en) | Method for measuring tooth three-dimensional movement in orthodontic process | |
CN109767841A (en) | A kind of scale model search method and device based on cranium jaw face three-dimensional configuration database | |
CN113052902A (en) | Dental treatment monitoring method | |
CN115690045A (en) | Quantitative assessment method for bone increment before and after periodontitis treatment based on curved surface fault slice | |
CN107240128A (en) | A kind of X-ray film and photochrome method for registering based on contour feature | |
CN113065552A (en) | Method for automatically positioning head shadow measurement mark point | |
CN105748173B (en) | The preparation method and device of orbital reconstruction material | |
Tian et al. | Efficient tooth gingival margin line reconstruction via adversarial learning | |
CN116823729A (en) | Alveolar bone absorption judging method based on SegFormer and oral cavity curved surface broken sheet | |
CN115886863A (en) | Tooth and facial bone three-dimensional overlapping measurement method and device with total skull base as datum plane | |
CN115996668A (en) | Non-contact four-dimensional imaging method and system based on four-dimensional body surface breathing signals | |
CN110858412A (en) | Image registration-based heart coronary artery CTA model establishing method | |
CN108154532A (en) | A kind of method of aided assessment SPECT image thyroid volumes | |
CN114708404A (en) | End-to-end operation puncture path automatic planning method and system based on machine learning | |
El-Fegh et al. | Automated 2-D cephalometric analysis of X-ray by image registration approach based on least square approximator | |
TWI461178B (en) | Method for motion correction and tissue classification of nodules in lung | |
Muad et al. | Tooth segmentation using dynamic programming-gradient inverse coefficient of variation | |
Abdolali et al. | Fully automated detection of the mandibular canal in cone beam CT images using Lie group based statistical shape models |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |