Intelligent auxiliary identification method and system for developmental aging abnormity based on LCR image

Abstract

The invention discloses an intelligent auxiliary identification method and system for developmental aging abnormity based on an LCR image, and the method comprises the following steps: extracting universal aging features from an X-ray skull positioning side image through an Officient-B0 age deduction model, and generating an aging saliency map; calculating average aging saliency by combining all aging saliency maps of the same age to obtain an aging-related dynamic attention value, wherein the aging-related dynamic attention value represents common age-related development and aging modes of head and neck bone and tooth tissues of people of different age stages; matching each anatomical region of different genders in the to-be-analyzed image with an aging-related dynamic attention value by using an average aging significance quantification result; and based on a result after matching, generating a difference value between the aging-related dynamic attention value and the aging saliency of the to-be-analyzed skull positioning side position film, and if the difference value exceeds a set threshold value, prompting that developmental aging abnormity exists in an area in the skull positioning side position film corresponding to the difference value exceeding the set threshold value.

Description

Technical Field

[0001] The present invention belongs to the field of image processing, and in particular relates to an intelligent auxiliary recognition method and system for developmental aging abnormalities based on LCR images. Background Art

[0002] Medical imaging is of indispensable importance in the assessment of development and aging. It not only provides non-invasive and accurate assessment methods, but also provides doctors with more comprehensive and in-depth information, which helps to detect and deal with problems in the development and aging process at an early stage. In the field of developmental assessment, medical imaging technologies, such as ultrasound and MRI, play an irreplaceable role. These technologies can observe the internal structure of the fetus and children in a non-invasive and in-depth manner, providing doctors with valuable information about their growth and development. Ultrasound examination can clearly show the development of the fetus' organs. Once abnormalities are found, it can provide a strong basis for early intervention. For children and adolescents, due to the rapid morphological changes of various tissues in the development process, traditional methods such as anthropometry can infer whether growth and development are normal by measuring growth characteristics, but medical imaging technology provides a more accurate and in-depth assessment method. For example, in the inference of bone age and dental age, in addition to traditional anthropometry methods, the application of medical imaging technology is becoming more and more popular. By analyzing X-rays of wrist bones, the Greulich and Pyle (GP) method or the Tanner and Whitehouse (TW) method can be used to accurately infer bone age. The GP method compares the wrist bone development map classified by age with the individual bone development stage to accurately determine the physiological age; while the TW method classifies and weights each wrist bone development stage, and finally infers the bone age by the sum of the weight values. In addition, medical imaging examinations of the cervical spine and other parts also provide doctors with more detailed information about the development, providing a more comprehensive basis for evaluation.

[0003] Medical imaging also plays a pivotal role in aging assessment. As people age, their organs and tissues gradually undergo degenerative changes, which can be intuitively displayed in medical images. Through examinations using advanced technologies such as CT and MRI, doctors can accurately assess changes in the brain structure of the elderly, take wrist bone films to diagnose aging phenomena such as osteoporosis, and use digital model registration to assess changes in 3D mandibular teeth with age, attempting to understand or characterize human aging from different perspectives to provide an important basis for the prevention and treatment of related diseases.

[0004] The above development and aging assessment methods mainly measure the aging-related changes in human tissue morphology, which mostly rely on manual measurement and subjective judgment. The assessment accuracy of the development and aging stage needs to be improved, especially the assessment of the aging stage. The morphological changes of human tissues with aging are very small and difficult to measure. The morphological characteristics of aging of human tissues are difficult to obtain and quantify through traditional methods. Therefore, finding a more ideal development and aging assessment method is a problem that scholars are concerned about and urgently need to solve. The emergence of artificial intelligence technology provides a new perspective and means for this.

[0005] After rapid development in recent years, artificial intelligence technology has achieved remarkable success and has had a profound and extensive impact on various fields, including the medical field. With the rapid growth of data scale and computing power, deep learning technology has been widely used in the field of medical health. In many medical image analysis tasks, Convolutional Neural Network (CNN) has achieved good performance, including classification, detection, segmentation, registration and general feature description of medical images.

[0006] The current developmental aging assessment methods rely heavily on professional knowledge, and the results are easily affected by subjective factors and take a long time. The automated developmental aging abnormality intelligent auxiliary diagnosis system can reduce the workload, save precious time, and improve the accuracy of the assessment. For example, there is a skeletal development assessment system based on deep learning. By training the model to identify skeletal features, the system can assess the individual's bone age and determine whether its development is normal. There is also an automatic skin aging analysis system: Skin aging is one of the most intuitive manifestations of the aging process. Through deep learning technology, the system can automatically analyze the texture, wrinkles, spots and other features of the facial skin, and then assess the individual's skin aging degree; because brain aging is closely related to cognitive decline. Through deep learning analysis of brain CT or magnetic resonance imaging (MRI) images, the brain aging assessment system based on medical imaging can automatically detect abnormalities such as brain atrophy and lesions, thereby assessing the individual's brain aging degree and providing relatively objective brain aging assessment results.

[0007] These automated intelligent auxiliary diagnosis systems for developmental aging abnormalities not only improve the accuracy of assessments, but also reduce the workload of doctors and save valuable time. However, one of the challenges facing the current intelligent auxiliary diagnosis systems for developmental aging abnormalities is the lack of developmental aging stage assessments covering the entire age span and the lack of quantitative indicators.

[0008] In computer vision, saliency can be used to highlight representative areas, and it is difficult to continuously track, sample, and characterize age-related features of the complete development and aging process of bones, teeth, and other tissues of a single individual in the population. However, large samples of people at different age stages can be used to characterize the aging characteristics of the human development and aging process. Features based on large samples of people are more typical and universal than those obtained from a single individual. An automatic analysis process and method for the development and aging of the human head, face, and neck based on deep learning can be used to study the quantitative dynamic distribution of ARDA in lateral cephalometric radiographs (LCR) images of the northern Chinese population. By identifying, comparing, and mining a large sample data set of lateral cephalometric radiographs of the Han Chinese population in northern my country with continuous age distribution, the development and aging characteristics of the northern Chinese population in the aging process and the relationship between developmental aging and age are discovered.

[0009] Since there are significant gender differences in the developmental process of males and females, ignoring the close connection between age and gender will have an adverse effect on the accuracy of age estimation and the stability of gender classification, thereby reducing the accuracy of the overall assessment. Summary of the invention

[0010] In the medical field, the evaluation of development and aging processes mainly relies on imaging data, but the existing technology has defects such as subjectivity and insufficient accuracy. To this end, the present invention proposes an auxiliary recognition system based on deep learning and ARDA, which realizes auxiliary recognition of developmental aging abnormalities by quantifying the aging significance of LCR images.

[0011] In order to achieve the above object, the technical solution adopted by the present invention is: an intelligent auxiliary identification method for developmental aging abnormalities based on LCR images, comprising the following steps:

[0012] Use the Efficient-B0 age inference model to extract common aging features from X-ray head positioning lateral images and generate aging saliency maps;

[0013] The aging-related dynamic attention value is obtained by merging all aging-saliency maps of the same age and calculating the average aging saliency. The aging-related dynamic attention value represents the common age-related development and aging patterns of bone and tooth tissues in the head and neck of people of different age stages.

[0014] Using the average age-related saliency quantification results, the dynamic attention values ​​related to aging are matched to each anatomical region of different genders in the analyzed images.

[0015] Based on the matching result, the difference between the aging-related dynamic attention value and the aging significance of the head positioning lateral film to be analyzed is generated. If the difference exceeds the set threshold, it indicates that the area in the head positioning lateral film corresponding to the difference exceeding the set threshold has developmental aging abnormalities.

[0016] Further, obtaining X-ray head positioning lateral image includes:

[0017] Acquire an X-ray head positioning lateral image to be analyzed, perform histogram equalization, short-side zero padding, and normalization processing on the X-ray head positioning lateral image to be analyzed, and obtain an X-ray head positioning lateral image to be analyzed that meets the analysis requirements.

[0018] Furthermore, the Efficient-B0 age inference model is used to extract common aging features from the X-ray head positioning lateral image to generate an aging saliency map including:

[0019] The global average of the feature map gradients in the trained Efficient-B0 model is defined as its weight, which is calculated by inputting the weight of the kth feature map into the fully connected layer;

[0020] Nonlinear mapping, resizing, and color mapping operations are performed, and the weighted sum of all feature maps is calculated to obtain an aging saliency map, namely, an aging saliency map.

[0021] Furthermore, when merging all aging saliency maps of the same age to calculate the average aging saliency to obtain the aging-related dynamic attention value, the indicator function To exclude areas where the aging significance is lower than the threshold t.

[0022] Furthermore, when matching each anatomical region of different genders with the aging-related dynamic attention value in the analyzed image, the developmental aging status of different anatomical regions is evaluated by comparing the aging salience of the anatomical region with the aging-related dynamic attention value.

[0023] Furthermore, the aging significance of each anatomical region was calculated. The absolute difference D between the corresponding mean aging significance (ARDA) and the threshold value is set as the standard deviation b of different structural regions according to the knowledge and experience in the medical field, and the calculated D of each anatomical region is compared with the threshold value; if D≤b, it is judged that the anatomical region has normal development and aging; if D>b, it is judged that the anatomical region has abnormal development and aging;

[0024] Calculate the abnormality score S for the anatomical region with abnormal development and aging

[0025]

[0026] D is the aging significance of each anatomical region ARDA is the absolute difference between the corresponding average aging significance and the average aging significance of each anatomical region.

[0027] With the same concept as the above method, the present invention provides an intelligent auxiliary recognition system for developmental aging abnormalities based on age-related dynamic attention ARDA, including an aging saliency map generation module, a calculation module, a matching module, and a recognition module;

[0028] The aging saliency map generation module uses the Efficient-B0 age inference model to extract common aging features from the X-ray head positioning lateral image to generate an aging saliency map;

[0029] The calculation module calculates the average aging saliency by merging all aging saliency maps of the same age to obtain the aging-related dynamic attention value, which represents the common age-related development and aging patterns of bone and tooth tissues in the head and neck of people at different age stages.

[0030] The matching module uses the average age-related saliency quantification results to match the aging-related dynamic attention values ​​of each anatomical region of different genders in the analyzed image;

[0031] Based on the matching results, the recognition module generates a difference between the aging-related dynamic attention value and the aging significance of the head positioning lateral film to be analyzed. If the difference exceeds a set threshold, it indicates that there is developmental aging abnormality in the area of ​​the head positioning lateral film corresponding to the difference exceeding the set threshold.

[0032] The application of the intelligent auxiliary identification method of developmental aging abnormalities based on LCR images of the present invention is used to identify development and aging assessment.

[0033] The present invention can also provide a computer device, including a processor and a memory, the memory is used to store a computer executable program, the processor reads the computer executable program from the memory and executes it, and when the processor executes the computer executable program, it can implement the intelligent auxiliary identification method of developmental aging abnormalities based on LCR images described in the present invention.

[0034] At the same time, a computer-readable storage medium is provided, in which a computer program is stored. When the computer program is executed by a processor, the intelligent auxiliary identification method for developmental aging abnormalities based on LCR images of the present invention can be implemented.

[0035] Compared with the prior art, the present invention has at least the following beneficial effects:

[0036] The present invention uses deep learning technology and ARDA quantitative indicators to accurately identify anatomical regional developmental aging abnormalities in LCR images. Compared with the prior art, the present invention has the advantages of greater objectivity, higher accuracy, and providing quantitative indicators and large sample average value references, which helps to improve the accuracy and reliability of medical diagnosis.

[0037] During the operation, the doctor only needs to input the LCR image to be diagnosed into the system, and the system will automatically perform quantitative analysis of aging significance and give the diagnosis results and corresponding quantitative indicators. This will greatly improve the doctor's work efficiency and also help improve the accuracy and reliability of diagnosis;

[0038] Through the implementation of the present invention, the accuracy and reliability of the diagnosis of developmental and aging abnormalities can be effectively improved, providing a powerful auxiliary tool for medical diagnosis. At the same time, the present invention can also be applied to other similar fields, such as disease prediction, health assessment, etc., and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 Display the anatomical structure marking of LCR images;

[0040] Figure 2 Schematic diagram of LCR aging saliency map generation module.

[0041] Figure 3 This is the LCR of a 28-year-old female patient randomly selected in the embodiment of the present invention. DETAILED DESCRIPTION

[0042] In the present invention, aging saliency is used to orderly express the degree of correlation between regions and aging in LCR images, and the severity of aging in regions in LCR images can be intuitively seen. In the present invention, aging saliency is quantified, and aging saliency is used to orderly characterize the degree of correlation between regions and aging in images, and the severity of aging in regions in LCR images can be intuitively visualized; the general aging attention obtained by calculating the average aging saliency by merging all aging saliency maps of the same age is called aging-related dynamic attention (ARDA), and the value of ARDA is the average aging saliency. The ARDA value represents the common age-related development and aging patterns of bone and tooth tissues in the head and neck of a large sample population at different age stages, and shows the distribution of the average aging saliency and aging regions in LCR images and their dynamic changes.

[0043] ARDA calculates the average aging significance by merging all aging significance maps of the same age. ARDA can reflect the common age-related development and aging patterns of the head and neck bone and tooth tissues in a large sample population. This averaging effect helps to eliminate individual differences and reveal more general aging laws. In addition, its advantage in the evaluation of developmental and aging abnormalities is that it can quantify the degree of correlation between the region in the image and aging and visualize it, so that doctors can intuitively see the severity of aging in different regions of the LCR image, facilitating rapid identification. ARDA not only reflects the average aging significance, but also can show the distribution of aging areas and their dynamic changes, which helps doctors to dynamically monitor the development and aging process of individuals and detect abnormal changes in a timely manner.

[0044] The present invention proposes an auxiliary recognition method based on deep learning and ARDA, which realizes the auxiliary recognition of male and female developmental aging abnormalities by quantifying the aging significance of LCR images of different genders. Although ARDA reveals the general aging law through the averaging effect, it can be used as an auxiliary recognition method.

[0045] The aging features of LCR images are extracted through the pre-trained CNN network for age estimation. The common aging features are extracted from the LCR images through the trained Efficient-B0 age inference model to generate an aging saliency map. ARDA is obtained by merging all aging saliency maps of the same age. Without being affected by abnormal samples, the overall distribution of aging features on the LCR image can be displayed. The ARDA generated for each age can show the dynamic changes of aging saliency with age in the LCR image. In order to quantitatively analyze the aging saliency (correlation with age) of each anatomical structure in the LCR image. The LCR image is marked according to the anatomical structure as follows: Figure 2 According to the ARDA value, the LCR aging distribution module and each anatomical structure were quantitatively analyzed to obtain the quantitative ARDA of each anatomical structure of the craniofacial and cervical spine regions of the LCR image for different genders and ages (as shown in Tables 1 and 2):

[0046] By quantifying the aging significance of the LCR image to be analyzed, and comparing it with the aging significance of each anatomical region calculated by the ARDA calculation module, it is determined whether there is developmental aging abnormality. The specific steps are as follows:

[0047] Step 1) Image preprocessing module: This module is responsible for receiving the LCR image to be analyzed and performing necessary preprocessing operations on it, such as size normalization, denoising, contrast enhancement, etc., to ensure that the image quality meets the requirements of subsequent analysis.

[0048] Step 1a) performing histogram equalization on the X-ray head positioning lateral image to be analyzed. The histogram equalization method used in this embodiment is a contrast-limited histogram equalization method;

[0049] Step 1b) padding the short side of the X-ray head positioning lateral image after data enhancement with zeros, and scaling the size of the square X-ray head positioning lateral image after zero padding to 1000×1000;

[0050] Step 1c) In order to eliminate the influence of the dimension of the X-ray head positioning lateral image on the network training, the X-ray head positioning lateral image after the size transformation is normalized, that is, the value of each pixel of the image is divided by the difference between the maximum and minimum values ​​of all pixels of the image to obtain the normalized X-ray head positioning lateral image;

[0051] Step 2) Age estimation module: Due to the good performance of the deep learning model Efficentnet-B0 in the LCR age inference research. The present invention uses the Efficentnet-B0 model to perform age inference on the preprocessed LCR image as input. For the newly input image, the age estimation module can automatically estimate its corresponding age. The convolutional layer in the Efficentnet-B0 model extracts image features from the input LCR image and inputs them into the fully connected layer in the Efficentnet-B0 model, and then the fully connected layer outputs the estimated value of the age of the input LCR image.

[0052] Step 3) Aging saliency map generation module: Aging saliency is used to orderly express the correlation between the region and aging in the LCR image, and can intuitively visualize the severity of aging in the region in the LCR image. The schematic diagram of the LCR aging saliency map generation module is shown in Figure 3 As shown,

[0053] Step 3a) The global average of the feature map gradients in the trained Efficient-B0 model is defined as its weight, which is calculated by inputting the weight of the kth feature map into the fully connected layer

[0054]

[0055] In the formula, Z is the number of pixels in the feature map F, is the output of the network, is the value of pixel (i, j) in the kth feature map.

[0056] Step 3b) Then perform nonlinear mapping, resizing, color mapping and other operations, and calculate the weighted sum of all feature maps to obtain the aging saliency map:

[0057]

[0058] Step 4) Quantitative analysis of developmental and aging significance:

[0059] Step 4a) After visualizing the significant bone and tooth tissues with aging changes in the LCR image to form a significance map, further quantitatively analyze the significance of the development and aging of bone and tooth tissues in the head and neck. In the present invention, the significance of a certain area in the LCR image at the age of a is defined as follows:

[0060]

[0061] In the formula, M is the coordinate point set of the area.

[0062] Step 4b) Use the indicator function To exclude regions with aging significance lower than the threshold t, so as to eliminate the influence of regional area on its aging significance.

[0063]

[0064] Step 5) ARDA matching judgment: Using the average aging significance quantitative results obtained in the previous experiments: quantitative ARDA of different ages of various anatomical structures in the craniofacial region of female LCR images (as shown in Table 1), quantitative ARDA of different ages of cervical spine C1-C4 vertebrae and vertebral bodies in female LCR images (as shown in Table 2), quantitative ARDA of different ages of various anatomical structures in the craniofacial region of male LCR images (as shown in Table 3), quantitative ARDA of different ages of cervical spine C1-C4 vertebrae and vertebral bodies in male LCR images (as shown in Table 4), each anatomical region of different genders in the image to be analyzed is matched with ARDA. By comparing the aging significance of the anatomical region with ARDA, the developmental aging status of different anatomical regions can be evaluated.

[0065] Table 1 Quantified ARDA (a±b) of various anatomical structures of craniofacial region in LCR images of females at different ages (a: mean; b: standard deviation)

[0066]

[0067]

[0068] Table 2 Quantified ARDA (a±b) of cervical spine C1-C4 vertebral processes and vertebral bodies at different ages in LCR images of women (a: mean; b: standard deviation)

[0069]

[0070]

[0071]

[0072] Table 3 Quantified ARDA (a±b) of various anatomical structures of craniofacial region in male LCR images at different ages (a: mean; b: standard deviation)

[0073]

[0074]

[0075] Table 4 Quantified ARDA (a±b) of cervical spine C1-C4 vertebral processes and vertebral bodies at different ages in male LCR images (a: mean; b: standard deviation)

[0076]

[0077]

[0078] Step 5) Abnormal diagnosis module: Based on the matching results, the difference between the aging significance of ARDA and the target analysis LCR is generated as the output result of the abnormal diagnosis module, and a certain threshold is set to determine whether there is developmental aging abnormality in the anatomical region. If the difference between the aging significance of a certain anatomical region and ARDA exceeds the threshold, it indicates that there is developmental aging abnormality in the region;

[0079] Step 5a) Calculate the difference: First, calculate the aging significance of each anatomical region The absolute difference D between the corresponding mean aging significance (ARDA). The absolute value is used here to ensure that no matter (a) is greater than or less than A positive difference can be obtained, which is convenient for subsequent comparison:

[0080]

[0081] Step 5b) Setting a threshold: Based on medical knowledge and experience, a suitable threshold (standard deviation b of different structural regions) is set. This threshold is used to distinguish between normal aging and abnormal aging.

[0082] Step 5c) Abnormality judgment: Compare the calculated D of each anatomical region with the threshold (standard deviation b).

[0083] If D≤b, then the anatomical region is considered to be developing and aging normally;

[0084] If D>b, it is judged that this anatomical region has developmental and aging abnormalities;

[0085] Step 5D) If it is determined that this anatomical region has developmental and aging abnormalities, calculate the abnormality score S. The abnormality score, as a quantitative indicator, can provide more specific and detailed information on the degree of abnormality. The higher the abnormality score, the more serious the developmental and aging abnormalities in the anatomical region, which may require closer attention and more active treatment measures. On the contrary, if the abnormality score is low, it may mean that the degree of abnormality is mild, but it still needs to be paid attention to and regularly checked. The present invention defines S:

[0086] S=D / ARDA

[0087] Step 6) Report generation module: According to the results of the abnormality diagnosis module, a detailed diagnosis report is automatically generated. The report contains the basic information of the image to be analyzed, the aging significance of each anatomical region, the comparison results with ARDA, and the final diagnosis conclusion of developmental aging abnormality. The output of the abnormality diagnosis module is the existence of developmental aging abnormality and the abnormality score S in each anatomical region.

[0088] Take the LCR of a 28-year-old female patient randomly selected as an example, refer to Figure 3 , the aging significance of different anatomical structures generated As shown in Table 5:

[0089] Table 5 Aging significance of different anatomical structures obtained by using LCR of a randomly selected 28-year-old female patient as input

[0090]

[0091] The absolute difference D between the aging significance of each anatomical region and the corresponding average aging significance (ARDA) was calculated to obtain the results, refer to Table 6 and Table 7:

[0092] Table 6. Quantified ARDA (a±b) of different anatomical structures in a 28-year-old female patient (a: mean; b: standard deviation)

[0093]

[0094] Table 7. Absolute difference D between the aging significance of each anatomical region and the corresponding average aging significance (ARDA) of the 28-year-old female

[0095]

[0096] The calculated D of each anatomical region was compared with the threshold value (standard deviation b) (see Table 6), and it was found that D≤b in all the above anatomical regions except the temporal bone. It was judged that the aging of each anatomical region of the patient was basically within the normal range.

[0097] Temporal bone region: Abnormal score S = 0.039 / 0.092 × 100% = 42.4%, indicating that the degree of abnormal aging in the temporal bone anatomical region is 42.4%.

[0098] The present invention provides an intelligent auxiliary recognition system for developmental aging abnormalities based on age-related dynamic attention (ARDA), including an aging saliency map generation module, a calculation module, a matching module and a recognition module;

[0099] The aging saliency map generation module uses the Efficient-B0 age inference model to extract common aging features from the X-ray head positioning lateral image to generate an aging saliency map;

[0100] The calculation module calculates the average aging saliency by merging all aging saliency maps of the same age to obtain the aging-related dynamic attention value, which represents the common age-related development and aging patterns of bone and tooth tissues in the head and neck of people at different age stages.

[0101] The matching module uses the average age-related saliency quantification results to match the aging-related dynamic attention values ​​of each anatomical region of different genders in the analyzed image;

[0102] Based on the matching results, the recognition module generates a difference between the aging-related dynamic attention value and the aging significance of the head positioning lateral film to be analyzed. If the difference exceeds a set threshold, it indicates that there is developmental aging abnormality in the area of ​​the head positioning lateral film corresponding to the difference exceeding the set threshold.

[0103] On the other hand, the present invention provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the intelligent auxiliary identification method for developmental aging abnormalities based on LCR images described in the present invention can be implemented.

[0104] The present invention can also provide a computer device, including a processor and a memory, the memory is used to store a computer executable program, the processor reads the computer executable program from the memory and executes it, and when the processor executes the computer executable program, the intelligent auxiliary identification method for developmental aging abnormalities based on LCR images described in the present invention can be implemented.

[0105] The computer device may be a laptop computer, a desktop computer or a workstation.

[0106] The processor may be a central processing unit (CPU), a digital signal processor (DSP), an application specific integrated circuit (ASIC), or an off-the-shelf field programmable gate array (FPGA).

[0107] The memory of the present invention may be an internal storage unit of a laptop computer, a desktop computer or a workstation, such as a memory or a hard disk; or an external storage unit, such as a mobile hard disk or a flash memory card.

[0108] Computer-readable storage media may include computer storage media and communication media. Computer storage media include volatile and non-volatile, removable and non-removable media implemented by any method or technology for storing information such as computer-readable instructions, data structures, program modules or other data. Computer-readable storage media may include: read-only memory (ROM), random access memory (RAM), solid-state drive (SSD) or optical disk, etc. Among them, random access memory may include resistance random access memory (ReRAM) and dynamic random access memory (DRAM).

[0109] The above description is only a specific example of the present invention and does not constitute any limitation to the present invention. It is obvious that for professionals in this field, after understanding the content and principles of the present invention, it is possible to make various modifications and changes in form and details without departing from the principles and structures of the present invention. However, these modifications and changes based on the ideas of the present invention are still within the scope of protection of the claims of the present invention.

Claims

1. An intelligent auxiliary identification method for developmental aging abnormalities based on LCR images, characterized in that: The following steps are involved: Use the Efficient-B0 age inference model to extract common aging features from X-ray head positioning lateral images and generate aging saliency maps; The aging-related dynamic attention value is obtained by merging all aging-saliency maps of the same age and calculating the average aging saliency. The aging-related dynamic attention value represents the common age-related development and aging patterns of bone and tooth tissues in the head and neck of people of different age stages. Using the average age-related saliency quantification results, the dynamic attention values ​​related to aging are matched to each anatomical region of different genders in the analyzed images. Based on the matching result, the difference between the aging-related dynamic attention value and the aging significance of the head positioning lateral film to be analyzed is generated. If the difference exceeds the set threshold, it indicates that the area in the head positioning lateral film corresponding to the difference exceeding the set threshold has developmental aging abnormalities.

2. The intelligent auxiliary identification method for developmental aging abnormalities based on LCR images according to claim 1 is characterized in that: Obtaining X-ray lateral head positioning images includes: Acquire an X-ray head positioning lateral image to be analyzed, perform histogram equalization, short-side zero padding, and normalization processing on the X-ray head positioning lateral image to be analyzed, and obtain an X-ray head positioning lateral image to be analyzed that meets the analysis requirements.

3. The intelligent auxiliary identification method for developmental aging abnormalities based on LCR images according to claim 1 is characterized in that: The Efficient-B0 age inference model is used to extract common aging features from X-ray head positioning lateral images to generate aging saliency maps including: The global average of the feature map gradients in the trained Efficient-B0 model is defined as its weight, which is calculated by inputting the weight of the kth feature map into the fully connected layer; Nonlinear mapping, resizing, and color mapping operations are performed, and the weighted sum of all feature maps is calculated to obtain an aging saliency map, namely, an aging saliency map.

4. The intelligent auxiliary identification method for developmental aging abnormalities based on LCR images according to claim 1 is characterized in that: When calculating the average aging saliency by merging all aging saliency maps of the same age to obtain the aging-related dynamic attention value, the indicator function To exclude areas where the aging significance is lower than the threshold t.

5. The intelligent auxiliary identification method for developmental aging abnormalities based on LCR images according to claim 1 is characterized in that: When matching each anatomical region of different genders with the aging-related dynamic attention value in the analyzed image, the developmental aging status of different anatomical regions is evaluated by comparing the aging salience of the anatomical region with the aging-related dynamic attention value.

6. The intelligent auxiliary identification method for developmental aging abnormalities based on LCR images according to claim 1 is characterized in that: Calculate the aging significance for each anatomical region The absolute difference D between the corresponding mean aging significance (ARDA) and the threshold value is set as the standard deviation b of different structural regions according to the knowledge and experience in the medical field, and the calculated D of each anatomical region is compared with the threshold value; if D≤b, it is judged that the anatomical region has normal development and aging; if D>b, it is judged that the anatomical region has abnormal development and aging; Calculate the abnormality score S for the anatomical region with abnormal development and aging D is the aging significance of each anatomical region ARDA is the absolute difference between the corresponding average aging significance and the average aging significance of each anatomical region.

7. An intelligent auxiliary recognition system for developmental aging abnormalities based on age-related dynamic attention ARDA, characterized in that: It includes an aging saliency map generation module, a calculation module, a matching module and a recognition module; The aging saliency map generation module uses the Efficient-B0 age inference model to extract common aging features from the X-ray head positioning lateral image to generate an aging saliency map; The calculation module calculates the average aging saliency by merging all aging saliency maps of the same age to obtain the aging-related dynamic attention value, which represents the common age-related development and aging patterns of bone and tooth tissues in the head and neck of people at different age stages. The matching module uses the average age-related saliency quantification results to match the aging-related dynamic attention values ​​of each anatomical region of different genders in the analyzed image; Based on the matching results, the recognition module generates a difference between the aging-related dynamic attention value and the aging significance of the head positioning lateral film to be analyzed. If the difference exceeds a set threshold, it indicates that there is developmental aging abnormality in the area of ​​the head positioning lateral film corresponding to the difference exceeding the set threshold.

8. A computer device, characterized in that: The invention comprises a processor and a memory, wherein the memory is used to store a computer executable program, and the processor reads part or all of the computer executable program from the memory and executes it. When the processor executes part or all of the computer executable program, the intelligent auxiliary identification method for developmental aging abnormalities based on LCR images as described in any one of claims 1 to 6 can be implemented.

9. A computer-readable storage medium, characterized in that: A computer program is stored in a computer-readable storage medium, and when the computer program is executed by a processor, it can implement the intelligent auxiliary identification method for developmental aging abnormalities based on LCR images as described in any one of claims 1-6.

10. The application of the intelligent auxiliary identification method for developmental aging abnormalities based on LCR images according to any one of claims 1 to 6, characterized in that: Used to identify development and aging assessments.

Images