A method for constructing a bone image data model based on CT images

Abstract

This invention provides a method for constructing a skeletal image data model based on CT images, belonging to the field of medical image processing technology. The method includes acquiring a dataset of human skeletal CT images of different age groups; labeling and analyzing the ROI regions of the bones in the CT images to obtain ROI region conformance values ​​for each age group; after preliminary optimization, further analysis of the CT images, fitting of HU-Histogram curves, extraction of Histogram feature conformance values, and secondary optimization based on these features. Finally, skeletal image data models for different age groups are constructed, and their performance is analyzed to obtain skeletal conformance characterization values. This improves the accuracy of skeletal structure analysis and solves the problem in existing technologies that only construct skeletal image data models based on individual CT images without analyzing and constructing skeletal image data models based on CT images of different age groups.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, and in particular to a method for constructing a skeletal image data model based on CT images. Background Technology

[0002] Constructing a skeletal imaging data model from CT images is a complex but precise process that relies on advanced medical imaging technology. Since its inception, CT (computed tomography) technology has undergone rapid development. Its high resolution and rapid imaging capabilities have led to its widespread application in medical diagnosis, particularly in the detection of bone diseases, providing clearer and more detailed images to help doctors accurately determine disease types. CT images can provide high-resolution cross-sectional images, displaying fine internal structures and lesions, aiding in the accurate diagnosis of various diseases, especially in detecting tumors, bleeding, infections, fractures, and other conditions. In orthopedic surgery, CT-based skeletal imaging data models can help doctors conduct detailed preoperative assessments, understanding the location of lesions and their relationship to surrounding tissues. During surgery, they provide real-time navigation, improving surgical accuracy and safety. The application prospects of CT-based skeletal imaging data model construction are broad. Clinically, it provides intuitive data for patient diagnosis; in scientific research, it offers rich data resources for studying disease pathogenesis; and in medical education, it can be used as a teaching tool to help students better understand disease types.

[0003] Existing methods primarily perform binary segmentation on CT images of the human skeleton in different directions to obtain three-dimensional structural feature points in a two-dimensional plane, and then obtain a three-dimensional model of the human skeleton based on these feature points. However, in implementing the technical solution of this invention, the present invention has discovered that the above-mentioned techniques have at least the following technical problems:

[0004] In existing technologies, bone image data models are typically built based on individual CT images only. There is no analysis to build bone image data models based on CT images of different age groups, and no assessment of the accuracy of the built model data and the model itself, nor are corresponding adjustments made. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a method for constructing a skeletal image data model based on CT images. This method solves the problem in existing technologies where skeletal image data models are constructed using only individual CT images, resulting in low accuracy of skeletal image data models for CT images across different age groups. This invention improves the accuracy of skeletal image data models based on CT images.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] A method for constructing a bone imaging data model based on CT images includes the following steps:

[0008] A human skeletal CT image dataset is obtained, comprising CT images of the human body at various age stages. Regions of interest (ROIs) are marked from the CT images of each age stage, and these ROIs are analyzed to obtain conformity values. Preliminary optimization is then performed on the CT images. The preliminarily optimized CT images are analyzed to fit HU-Histogram curves for each age stage, and histogram feature conformity values ​​are obtained. Based on these histogram feature conformity values, a secondary optimization is performed on the CT images. The secondary optimized CT images are extracted to obtain skeletal image data model construction data. A skeletal image data model is constructed for each age stage, yielding skeletal conformity representation values. Performance analysis and suggestions are then provided for the skeletal image data model.

[0009] Optionally, the specific process of obtaining the skeletal ROI regions of the human body at different age stages is as follows: grayscale processing is performed on CT images of the human body at different age stages to obtain grayscale images of CT images of the human body at different age stages; noise is removed from the grayscale images of CT images of the human body at different age stages, and skeletal contours are extracted from the grayscale images of CT images of the human body at different age stages; the skeletal ROI regions of the human body at different age stages are obtained based on the skeletal contours from the grayscale images of CT images of the human body at different age stages.

[0010] Optionally, the specific process for obtaining the ROI region conformity values ​​for different age stages of the human body is as follows: Traverse the voxels in the skeletal ROI regions of different age stages of the human body, extract and collect the feature information of each voxel, and obtain the HU value of each voxel in the skeletal ROI region of the CT images of different age stages of the human body; extract the bone volume covered by the ROI region, the peak value, valley value, and edge length of the HU value from the HU value of each voxel in the skeletal ROI region of the CT images of different age stages of the human body, and process them to obtain the ROI region conformity values ​​for different age stages of the human body. The ROI region conformity values ​​for different age stages of the human body are used to evaluate the degree of conformity of the ROI regions of different age stages of the human body; compare the ROI region conformity values ​​for different age stages of the human body with the ROI region conformity value threshold for different age stages of the human body. If the ROI region conformity values ​​for different age stages of the human body are higher than or equal to the ROI region conformity value threshold for different age stages of the human body, then directly preprocess the CT images of different age stages of the human body. If the ROI region conformity values ​​for different age stages of the human body are lower than the ROI region conformity value threshold for different age stages of the human body, then sharpen and remove artifacts from the CT images.

[0011] Optionally, the specific process of analyzing the preliminarily optimized CT images of the human body at different ages and fitting the HU-Histogram curves of the human body at different ages is as follows: the HU values ​​in the skeletal ROI region of the human body at different ages are divided into intervals to obtain the HU value intervals of the skeletal ROI region of the human body at different ages, and the number of voxels of HU values ​​in each HU value interval of the skeletal ROI region of the human body at different ages is counted, thereby constructing the skeletal HU-Histogram curves of the human body at different ages.

[0012] Optionally, the specific process for obtaining the Histogram feature conformity values ​​of the human body at different age stages is as follows:

[0013] The width and mean HU value of the HU-Histogram curves of human bones at different ages are extracted from the HU-Histogram curves of human bones at different ages. The standard HU-Histogram curves of human bones at different ages stored in the database are extracted and compared with the standard HU-Histogram curves of human bones at different ages to obtain the peak position offset of the standard HU-Histogram curves of human bones at different ages. The Histogram feature conformity value of human bones at different ages is obtained after processing. The Histogram feature conformity value of human bones at different ages is used to evaluate the degree of difference in bone density of human bones at different ages in CT images.

[0014] The analysis process for the histogram characteristics of the human body at different age stages is as follows:

[0015]

[0016] In the formula, ρ i X represents the Histogram feature conformance value of the i-th age stage of the human body, i = 1, 2, ..., n, where i represents the age stage number and n represents the total number of age stages. i Y represents the width of the Histogram curve for the i-th age stage of the human body. i Z represents the mean HU value of the Histogram curve for the i-th age stage of the human body. i ΔX represents the peak position shift of the HU-Histogram curve of the human skeleton at the i-th age stage. i The standard value of the width of the Histogram curve for the i-th age stage of the human body, ΔY i ΔZ represents the standard mean value of the HU value of the Histogram curve for the i-th age stage of the human body. i α1 represents the standard value of the peak position offset of the HU-Histogram curve for the i-th age stage of the human body, α2 represents the correction factor corresponding to the width of the preset Histogram curve, α3 represents the correction factor corresponding to the mean of the HU value of the preset Histogram curve, and α4 represents the correction factor corresponding to the peak position offset of the preset HU-Histogram curve.

[0017] Optionally, the specific process of performing secondary optimization on CT images of different age groups based on the Histogram feature conformance values ​​of different age groups is as follows: extract the Histogram feature conformance values ​​of different age groups and compare them with the Histogram feature conformance thresholds of different age groups stored in the database. If the Histogram feature conformance values ​​of different age groups are higher than or equal to the Histogram feature conformance thresholds of different age groups, then enhance the contrast and brightness of the skeletal ROI regions of different age groups. If the Histogram feature conformance values ​​of different age groups are lower than the Histogram feature conformance thresholds of different age groups, then enhance and segment the CT images of different age groups.

[0018] Optionally, the specific process of extracting the CT images of the human body at different ages after secondary optimization to obtain the skeletal image data model construction data of the CT images of the human body at different ages is as follows: extracting the grayscale images of the CT images of the human body at different ages after secondary optimization and segmenting them to obtain each segmented grayscale image of the CT images of the human body at different ages after secondary optimization, and obtaining the geometric and structural features of the bones based on each segmented grayscale image of the CT images of the human body at different ages after secondary optimization; and combining the geometric and structural features of the bones as the skeletal image data model construction data of the CT images of the human body at different ages.

[0019] Optionally, the specific process of constructing the skeletal image data model of CT images of the human body at different ages is as follows: construct the skeletal image data model of CT images of the human body at different ages based on the skeletal image data model of CT images of the human body at different ages.

[0020] Optionally, the specific process of obtaining the skeletal conformity representation values ​​for each age stage of the human body is as follows: obtain preset bone density values, preset bone weight values, and preset bone toughness values ​​from the database, and combine them with the bone density values, bone weight values, and bone toughness values ​​output by the skeletal image data model of CT images of each age stage of the human body to obtain the skeletal conformity representation values ​​for each age stage of the human body. The skeletal conformity representation values ​​for each age stage of the human body are used to determine the accuracy of the skeletal image data model construction of CT images of each age stage of the human body.

[0021] Optionally, the specific process of performing performance analysis and prompting on the skeletal image data model of CT images of human bodies at different ages is as follows: extract the skeletal conformance values ​​of human bodies at different ages and compare them with the preset skeletal conformance value threshold. If the skeletal conformance values ​​of human bodies at different ages are higher than or equal to the preset skeletal conformance value threshold, it indicates that the model performance is good. If the skeletal conformance values ​​of human bodies at different ages are lower than the preset skeletal conformance value threshold, more CT images of human bodies at different ages are obtained to increase the training data of the model.

[0022] One or more technical solutions provided in the embodiments of the present invention have at least the following technical effects.

[0023] Or advantages:

[0024] 1. Obtain human skeletal CT image datasets for different age groups. Label and analyze the ROI regions in the CT images to obtain ROI region conformance values ​​for each age group. After initial optimization, further analyze the CT images, fit HU-Histogram curves, extract Histogram feature conformance values, and perform secondary optimization based on these features. Finally, construct skeletal image data models for different age groups, analyze their performance, and derive skeletal conformance characterization values, which can improve the accuracy of skeletal structure analysis.

[0025] 2. Grayscale processing is performed on CT images to generate grayscale images, removing noise and extracting bone contours to identify and extract the Regions of Interest (ROIs) of the bones. Within the extracted bone ROIs, voxels are traversed to collect feature information, obtaining the Hub (HU) value for each voxel. By analyzing these HU values, features such as bone volume, peak and trough values ​​of the HU values, and edge length are extracted to calculate the conformity value of the bone ROI, evaluating its conformity. Finally, the conformity value is compared with a threshold to determine whether to preprocess or sharpen and remove artifacts from the CT images, effectively improving the quality of CT images and making subsequent analysis more accurate.

[0026] 3. Extract the width and mean HU value of the HU-Histogram curves for bones of each age group, and compare them with the standard curves in the database to calculate the peak position offset, thereby obtaining the histogram feature conformance value. By comparing these conformance values ​​with stored thresholds, it is determined whether to enhance the contrast and brightness of the skeletal ROI region, or to enhance and segment the image, providing a data foundation for subsequent model construction.

[0027] 4. Perform grayscale image segmentation on CT images of the human body at different ages, extract the grayscale images of bones and their geometric and structural features as input, construct a bone image data model, evaluate the quality of model construction by comparing the bone conformity representation value with the preset threshold, and decide whether to add training data to optimize the model, flexibly adjust and optimize the model to ensure the quality of the model. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention, 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 the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a flowchart of a method for constructing a skeletal image data model based on CT images according to the present invention.

[0030] Figure 2 This is a statistical chart showing the changes in the skeletal conformity characterization values ​​of the present invention. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0032] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms “first,” “second,” and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, the terms “an,” “a,” or “the,” and similar terms do not indicate a quantity limitation, but rather indicate the presence of at least one. The terms “comprising,” “including,” or “including,” and similar terms mean that the element or object preceding the word encompasses the element or object listed following the word and its equivalents, without excluding other elements or objects. The terms “connected,” “linked,” or “connected,” and similar terms are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect.

[0033] It should be noted that the terms "up", "down", "left", "right", "front", and "back" used in this invention are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0034] This invention provides a method for constructing a skeletal image data model based on CT images. This method solves the problem that existing technologies only construct skeletal image data models based on individual CT images, resulting in low accuracy of CT image data models. By acquiring skeletal CT image datasets of people at various age stages, marking skeletal ROI regions from CT images of people at different age stages, fitting HU-Histogram curves and extracting features, optimizing the fitting process, constructing a skeletal image data model, and evaluating model performance, the accuracy of modeling is improved.

[0035] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0036] like Figure 1As shown in the figure, this invention provides a flowchart of a method for constructing a skeletal image data model based on CT images, including: obtaining a human skeletal CT image dataset, wherein the human skeletal CT image dataset includes CT images of the human body at various age stages; marking skeletal ROI regions from the CT images of the human body at various age stages to obtain skeletal ROI regions of the human body at various age stages; analyzing the skeletal ROI regions of the human body at various age stages to obtain ROI region conformity values ​​of the human body at various age stages, and performing preliminary optimization processing on the CT images of the human body at various age stages; analyzing the preliminary optimized CT images of the human body at various age stages, fitting HU-Histogram curves of the human body at various age stages, processing to obtain Histogram feature conformity values ​​of the human body at various age stages, and performing secondary optimization on the CT images of the human body at various age stages based on the Histogram feature conformity values ​​of the human body at various age stages; extracting the secondary optimized CT images of the human body at various age stages to obtain skeletal image data model construction data of the human body at various age stages, constructing skeletal image data models of the human body at various age stages, obtaining skeletal conformity characterization values ​​of the human body at various age stages, and performing performance analysis and suggestions on the skeletal image data models of the human body at various age stages.

[0037] In this embodiment, a dataset of skeletal CT (Computed Tomography) images of individuals at various age stages is acquired. Regions of Interest (ROIs) are marked in each image. By processing these ROIs, ROI values ​​for each age stage are obtained, and the CT images are optimized. Based on the optimization results, a HU-Histogram (Hounsfield Unit Histogram) is fitted. Histogram feature extraction and processing are performed on the HU values ​​to obtain histogram feature values ​​for each age stage, resulting in a secondary optimization of the CT images. Finally, a skeletal image data model for each age stage is constructed, and its performance is analyzed. This addresses the limitation of existing technologies that only construct skeletal image data models based on individual CT images, thereby improving the model's accuracy.

[0038] This process involves grayscale processing of CT images of the human body at different ages to obtain grayscale images of CT images of the human body at different ages, removing noise from the grayscale images of CT images of the human body at different ages, and extracting the skeletal contours on the grayscale images of CT images of the human body at different ages; and obtaining the skeletal ROI regions of the human body at different ages based on the skeletal contours on the grayscale images of CT images of the human body at different ages.

[0039] The dataset of human skeletal CT images was obtained from the database. Based on different age groups, the dataset was divided into CT images of different age groups, which refer to different growth stages of the human body. These stages are mainly divided into: childhood (0-12 years old), adolescence (12-18 years old), adulthood (18-65 years old), and old age (65 years old and above). The skeletal attributes of the human body are different in different age groups. Classifying the CT images according to different age groups and analyzing them separately helps to improve the accuracy of the skeletal image data model of CT images of the human body at different age stages.

[0040] Specifically, grayscale processing of CT images of the human body at different ages is performed to highlight the edges and contours in the images, facilitating feature extraction. The value of each pixel in the CT images of the human body at different ages is directly used as the grayscale value of the grayscale image, and the CT images are converted to grayscale. Canny edge detection is used to identify and extract the skeletal contours of the grayscale images of CT images of the human body at different ages. Canny edge detection sets two thresholds: a high threshold and a low threshold. If the gradient value of a pixel is greater than or equal to the high threshold, it is considered a strong edge point; if the gradient value is less than or equal to the low threshold, it is considered a non-edge point; if the gradient value is between the two, it is considered a weak edge point. For weak edge points, it is further checked whether there are strong edge points in its neighborhood. If there are, it is considered an edge point; if not, it is considered a non-edge point, in order to preserve the true weak edges. The edge points are connected to obtain the skeletal contours of the grayscale images of CT images of the human body at different ages.

[0041] In this embodiment, the ROI region should closely surround the skeletal outline to ensure that it contains complete skeletal information. Therefore, the skeletal outline of the grayscale image of the CT image of the human body at different ages can be the ROI region of the grayscale image of the CT image of the human body at different ages. According to the location of the human body captured by the CT image (such as the spine, ribs, and limb bones), the ROI regions of the grayscale image of the CT image of the human body at different ages are grouped to obtain the skeletal ROI regions of the human body at different ages.

[0042] Optionally, voxels in the skeletal ROI regions of different age groups of the human body are traversed, and feature information of each voxel is extracted and collected to obtain the HU value of each voxel in the skeletal ROI region of the CT images of different age groups of the human body. The volume of the bone covered by the ROI region, the peak value, the valley value, and the edge length of the HU value are extracted from the HU value of each voxel in the skeletal ROI region of the CT images of different age groups of the human body. The ROI region conformity value of different age groups of the human body is processed to obtain the conformity value of the ROI region of different age groups of the human body. The ROI region conformity value of different age groups of the human body is compared with the ROI region conformity value threshold of different age groups of the human body. If the ROI region conformity value of different age groups of the human body is higher than or equal to the ROI region conformity value threshold of different age groups of the human body, the CT images of different age groups of the human body are directly preprocessed. If the ROI region conformity value of different age groups of the human body is lower than the ROI region conformity value threshold of different age groups of the human body, the CT images are sharpened and artifact removal operations are performed.

[0043] In this embodiment, the number of voxels in the skeletal ROI region of CT images of the human body at different ages is counted, and the volume of the ROI region is obtained by summing the voxel volumes. The maximum and minimum HU values ​​of each voxel in the skeletal ROI region of CT images of the human body at different ages are extracted to obtain the peak and valley values ​​of the HU values ​​of each voxel in the skeletal ROI region of CT images of the human body at different ages. The edge length of the ROI region is obtained by image processing software.

[0044] In this embodiment, the preprocessing of CT images of the human body at different ages includes standardizing the HU values ​​in the CT images. Standardization is to adjust the HU values ​​of different CT scans to a uniform range, usually between 0 and 1 or a standard normal distribution (mean of 0 and standard deviation of 1).

[0045] In this embodiment, sharpening is performed on CT images of the human body at different ages to enhance edges and details, making the images appear clearer. Sobel gradient processing is used to sharpen the CT images; the Sobel operator is a first-order differential operator capable of detecting edges in an image. Using the Sobel operator to process the image can highlight bone boundaries, further enhancing the sharpening effect. Artifacts may interfere with the analysis of bone morphology and structure, leading to inaccurate contour recognition. An adaptive Gaussian filter can be used to reduce image artifacts in CT images. An adaptive Gaussian filter is a filter that dynamically adjusts the parameters of a Gaussian function based on the statistical characteristics of local image regions. It can automatically adjust parameters according to the size and distribution of artifacts in the CT image to remove artifacts. By removing artifacts and sharpening the CT images, image quality is improved.

[0046] In this embodiment, the process of analyzing the ROI region conformity values ​​for different age stages of the human body is as follows:

[0047]

[0048] In the formula, G i This represents the ROI region conformity value for the i-th age group of the human body, where i = 1, 2, ..., n, i represents the age group number, n represents the total number of age groups, and M... i N represents the volume of bone covered by the ROI region in a CT image of the i-th age group of the human body. i R represents the peak value of the HU value in the skeletal ROI region of a CT image at the i-th age stage of the human body. i O represents the trough value of the HU value in the skeletal ROI region of a CT image at the i-th age stage of the human body. i ΔM represents the edge length of the skeletal ROI region in a CT image of the i-th age stage of the human body. i ΔT represents the standard value of the bone volume covered by the ROI region in the CT image of the i-th age group. i ΔO represents the standard range of HU values ​​covering the skeletal ROI region in CT images of the i-th age group. i τ1 represents the standard value of the edge length of the ROI region in the CT image of the human body at the i-th age stage, τ2 represents the correction factor of the bone volume covered by the ROI region, τ3 represents the correction factor of the HU value range covered by the ROI region, and τ4 represents the correction factor of the edge length of the ROI region.

[0049] In this embodiment, there is a correlation between the bone volume covered by the ROI region, the peak and trough values ​​of the HU (Human Hull Number) value, and the edge length. A larger ROI region typically contains more bone tissue, resulting in an increased range of HU values. Simultaneously, as the volume increases, the edge length of the bone may also increase, especially in irregularly shaped bone regions. Increased volume can lead to more complex edges, thus affecting the edge length. Conversely, if the volume decreases, the range of HU values ​​may shrink, and the edge length may also decrease, ultimately affecting the accuracy of the assessment of bone morphology and density.

[0050] In this embodiment, the correction factor for the bone volume covered by the ROI region is a preset correction factor for the bone volume covered by the ROI region in the database. This factor represents the degree of influence of the bone volume covered by the ROI region on the ROI region's conformity value at different age stages. When using this method, the preset correction factor for the bone volume covered by the ROI region can be directly obtained from the database. The correspondence can be a pre-defined mapping relationship. For example, a mapping set can be formed between the bone volume covered by the ROI region and the correction factor for the bone volume covered by the ROI region in the database. The real-time bone volume covered by the ROI region is input into the mapping set to obtain the corresponding weight. The mapping relationship can be one-to-one or many-to-one. In this example, its value range is [0, 1].

[0051] In this embodiment, the correction factor for the HU value range covered by the ROI region is a preset correction factor for the HU value range covered by the ROI region in the database. This represents the degree of influence of the HU value range covered by the ROI region on the ROI region's conformity value at different age stages. When using this method, the preset correction factor for the HU value range covered by the ROI region can be directly obtained from the database. The correspondence can be a pre-defined mapping relationship. For example, a mapping set can be formed between the HU value range covered by the ROI region and the correction factor for the HU value range covered by the ROI region in the database. The real-time HU value range covered by the ROI region is input into the mapping set to obtain the corresponding weight. The mapping relationship can be one-to-one or many-to-one. In this example, its value range is [0, 1].

[0052] In this embodiment, the correction factor for the ROI region edge length is a preset correction factor for the ROI region edge length in the database. This factor represents the degree of influence of the ROI region edge length on the ROI region conformity value at different age stages of the human body. When using this method, the preset correction factor for the ROI region edge length can be directly obtained from the database. The correspondence can be a pre-defined mapping relationship. For example, a mapping set can be formed between the ROI region edge length and the correction factor for the ROI region edge length in the database. The real-time ROI region edge length can be input into the mapping set to obtain the corresponding weight. The mapping relationship can be one-to-one or many-to-one. In this example, its value range is [0, 1].

[0053] Optionally, the HU values ​​in the skeletal ROI region of the human body at different age stages are divided into intervals to obtain the HU value intervals of the skeletal ROI region of the human body at different age stages. The number of voxels of HU values ​​in each HU value interval of the skeletal ROI region of the human body at different age stages is counted, thereby constructing the skeletal HU-Histogram curve of the human body at different age stages.

[0054] It should be noted that by segmenting and statistically analyzing the HU values ​​within the ROI of human bones at different age stages, a HU-Histogram curve is constructed. The number of voxels is then grouped and counted according to a certain step size, thus obtaining a histogram curve reflecting the distribution of HU values. This is equivalent to dividing the entire skeletal HU range (e.g., 1000 to 2000 HU) into several intervals, counting the number of bone voxels in each interval, and obtaining a histogram curve reflecting the distribution of bone density. The horizontal axis represents the HU value interval, and the vertical axis represents the number of voxels within each interval. For example, using 0–2000 HU as the statistical range, with each interval being 5 HU, 200 cases were extracted from each of the five age groups: 10–20 years old, 20–30 years old, and 30–40 years old. This resulted in 10,000 ROI curves for each group. After averaging these curves, a representative standard HU-Histogram curve (standard curve) was obtained. The experimental results show that the peak value of the curve in the young group (20-30 years old) is concentrated in the range of 400-450 HU, while the peak value in the elderly group (60-70 years old) shifts to the right to the range of 450-500 HU. The curve width also increases from about 300 HU in the young group to about 350 HU in the elderly group. This indicates that as age increases, the distribution of bone mineral density tends to spread towards higher HU values, which can directly reflect the age characteristics of bone calcification and changes in bone mineral density.

[0055] The width and mean HU value of the HU-Histogram curves of human bones at different ages are extracted from the HU-Histogram curves of human bones at different ages. The standard HU-Histogram curves of human bones at different ages stored in the database are extracted and compared with the standard HU-Histogram curves of human bones at different ages to obtain the peak position offset of the standard HU-Histogram curves of human bones at different ages. The Histogram feature conformance value of human bones at different ages is obtained after processing. The Histogram feature conformance value of human bones at different ages is used to evaluate the degree of difference in bone density of human bones in CT images at different ages.

[0056] The analysis process for the histogram characteristics of the human body at different age stages is as follows:

[0057]

[0058] In the formula, ρ i X represents the Histogram feature conformance value of the i-th age stage of the human body, i = 1, 2, ..., n, where i represents the age stage number and n represents the total number of age stages. i Y represents the width of the Histogram curve for the i-th age stage of the human body. iZ represents the mean HU value of the Histogram curve for the i-th age stage of the human body. i ΔX represents the peak position shift of the HU-Histogram curve of the human skeleton at the i-th age stage. i The standard value of the width of the Histogram curve for the i-th age stage of the human body, ΔY i ΔZ represents the standard mean value of the HU value of the Histogram curve for the i-th age stage of the human body. i α1 represents the standard value of the peak position offset of the HU-Histogram curve for the i-th age stage of the human body, α2 represents the correction factor corresponding to the width of the preset Histogram curve, α3 represents the correction factor corresponding to the mean of the HU value of the preset Histogram curve, and α4 represents the correction factor corresponding to the peak position offset of the preset HU-Histogram curve.

[0059] It should be noted that histogram-based moments (such as mean and variance) can improve the sensitivity to bone tissue status. Preininger et al., in a micro-CT study of bone healing in mice, pointed out that using histogram moment features to describe bone tissue (extracting histogram moments after removing background peaks) can clearly distinguish mineralized tissue, cartilage, and other tissues. On human skeletal CT, the mean HU (humeral density) is highly correlated with bone mineral density, and an increase in histogram width is often accompanied by a decrease in bone mineralization. Therefore, by extracting and comparing these histogram feature values, the differences in bone mineral density at different age stages can be quantified. For example, the aforementioned study found that the average HU of the lumbar spine in normal (non-osteoporosis) individuals was approximately 120.8 HU, while the mean HU in the osteopenia and osteoporosis groups were 78.8 HU and 54.7 HU, respectively, corresponding to a significantly lower and wider histogram mean. Combining this with histogram peak shifts can lead to the construction of more comprehensive assessment indicators.

[0060] In this embodiment, the three parameters—the width of the HU-Histogram curve, the mean HU value, and the peak position shift of the HU-Histogram curve—are not independent. Changes in the width of the HU-Histogram curve not only affect the mean HU value but are also closely related to the peak position shift. For example, a wider HU-Histogram curve typically indicates a more dispersed HU value distribution, which may lead to a larger range of mean variation and potential peak position shifts. As the curve width increases, the mean HU value may also shift, especially when the HU distribution is uneven. An increase in curve width may cause the peak position to shift towards higher or lower HU value ranges. If the curve width decreases, the mean HU value will become more concentrated, and the peak position shift may also decrease, thus affecting the overall characteristic analysis of the HU value distribution.

[0061] In this embodiment, the correction factor for the width of the Histogram curve is a preset correction factor for the width of the Histogram curve in the database. This factor represents the degree of influence of the width of the Histogram curve on the Histogram feature conformity values ​​at different age stages. When using this method, the preset correction factor for the width of the Histogram curve can be directly obtained from the database. The correspondence can be a pre-defined mapping relationship. For example, a mapping set can be formed between the width of the Histogram curve and the correction factor for the width of the Histogram curve in the database. The width of the real-time Histogram curve can be input into the mapping set to obtain the corresponding weight. The mapping relationship can be one-to-one or many-to-one. In this example, its value range is [0, 1].

[0062] In this embodiment, the correction factor for the mean HU value of the Histogram curve is a preset correction factor for the mean HU value of the Histogram curve in the database. This factor represents the degree of influence of the mean HU value of the Histogram curve on the Histogram feature conformity value at different age stages. When using this method, the preset correction factor for the mean HU value of the Histogram curve can be directly obtained from the database. The correspondence can be a pre-defined mapping relationship. For example, a mapping set can be formed between the mean HU value of the Histogram curve and the correction factor for the mean HU value of the Histogram curve in the database. The mean HU value of the real-time Histogram curve is input into the mapping set to obtain the corresponding weight. The mapping relationship can be one-to-one or many-to-one. In this example, its value range is [0, 1].

[0063] In this embodiment, the correction factor for the peak position offset of the HU-Histogram curve is a preset correction factor for the peak position offset of the HU-Histogram curve in the database. This factor represents the degree of influence of the peak position offset of the HU-Histogram curve on the Histogram feature conformity value at different age stages of the human body. When using this method, the preset correction factor for the peak position offset of the HU-Histogram curve can be directly obtained from the database. The correspondence can be a pre-defined mapping relationship. For example, a mapping set can be formed between the peak position offset of the HU-Histogram curve and the correction factor for the peak position offset of the HU-Histogram curve in the database. The peak position offset of the real-time HU-Histogram curve can be input into the mapping set to obtain the corresponding weight. The mapping relationship can be one-to-one or many-to-one. In this example, its value range is [0, 1].

[0064] In this embodiment, it should be noted that the horizontal axis of the HU-Histogram curve represents the density of each voxel, and the horizontal axis represents the HU value. The range of the horizontal axis can be set according to the characteristics of bone tissue. Generally, the HU value range of bone is concentrated between +200 and +2000, but in order to cover the complete CT value, the horizontal axis usually starts from -1000 (the HU value of air) and extends to +3000 or higher. For example, it can be divided into several small intervals (bins), each bin may cover 10 to 50 HU values, that is, the interval is divided into [200, 250], [250, 300], etc. The vertical axis of the HU-Histogram curve represents the number of voxels belonging to that range within each HU value interval.

[0065] In this embodiment, the width of the Histogram curve is obtained by analyzing the range of HU values ​​covered by the Histogram curve. Specifically, the width represents the distribution span of the curve from the lowest to the highest HU value, that is, the difference between the minimum and maximum HU values; the mean HU value is obtained by calculating the weighted average of the HU values ​​in each interval of the HU-Histogram curve for each age group, specifically by using the number of voxels in each HU value interval as the weight to calculate the overall mean HU value; the peak position offset of the HU-Histogram curve is obtained by finding the peak positions of the curve and the standard curve, calculating the difference between the peak position of the actual curve (actual HU value) and the peak position of the standard curve (standard HU value), and taking the absolute value to obtain the peak position offset.

[0066] In this embodiment, we assume the width X of the Histogram curve is... i The value is 200-2000, and the mean Y of the HU value of the Histogram curve is... i The value is 100-150, and the peak position offset Z of the HU-Histogram curve is... i The value is 10-30, and the standard value of the width ΔX of the Histogram curve is... i The value is 500, and the mean standard value of the HU value of the Histogram curve is ΔY. i The value is 120, and the standard value of the peak position offset ΔZ of the HU-Histogram curve is 120. i The value is 20, the correction factor α1 corresponding to the width of the Histogram curve is 0.3, the correction factor α2 corresponding to the mean of the HU values ​​of the Histogram curve is 0.5, and the correction factor α3 corresponding to the peak position offset of the HU-Histogram curve is 0.2. The statistical table of the changes in Histogram feature values ​​of human beings at different age stages is shown in Table 1:

[0067] Table 1. Histogram characteristics of the human body at different age stages.

[0068]

[0069]

[0070] As shown in Table 1, the smaller the difference between the width of the Histogram curve and the standard value of the width of the Histogram curve, the greater the Histogram feature conformity value of different age groups of the human body, which means that the degree of difference in bone density of CT images of different age groups of the human body is smaller.

[0071] Specifically, the Histogram feature values ​​of each age group of the human body are extracted and compared with the Histogram feature matching thresholds of each age group of the human body stored in the database. If the Histogram feature matching values ​​of each age group of the human body are higher than or equal to the Histogram feature matching thresholds of each age group of the human body, the contrast and brightness of the skeletal ROI regions of each age group of the human body are enhanced. If the Histogram feature matching values ​​of each age group of the human body are lower than the Histogram feature matching thresholds of each age group of the human body, the CT images of each age group of the human body are enhanced and segmented.

[0072] It should be noted that the segmented ROI can be further used for quantitative bone mass analysis or virtual implant fitting. In osteoporosis prevention or preoperative planning, such optimization helps to accurately assess bone strength and shape. For example, in osteoporosis screening applications, enhanced images can help magnify small structures in the sacroiliac bones or vertebrae to detect bone micro-damage; in pre-hip replacement surgery, segmentation provides a clear three-dimensional bone model for implant selection.

[0073] Specifically, grayscale images of CT images of the human body at different ages after secondary optimization are extracted and segmented to obtain segmented grayscale images of CT images of the human body at different ages after secondary optimization. Based on the segmented grayscale images of CT images of the human body at different ages after secondary optimization, the geometric and structural features of the skeleton are obtained. The geometric and structural features of the skeleton are combined as the data for constructing the skeleton image data model of CT images of the human body at different ages.

[0074] In this embodiment, image processing software is used to enhance the contrast and brightness of the ROI regions of the human skeleton at different ages, making the images clearer. Professional image processing software is used to preprocess the acquired CT images, including noise reduction and grayscale correction. Then, image segmentation techniques such as threshold segmentation and edge detection are used to extract the ROI regions, and further contrast enhancement and detail optimization are performed on the ROI regions to finally obtain the enhanced segmented image.

[0075] In this embodiment, grayscale thresholds for CT images of the human body at different ages are obtained from a database. These thresholds can distinguish between skeletal and non-skeletal structures. The grayscale value of each pixel in the grayscale image of the CT image of the human body at different ages is compared with the grayscale threshold of the grayscale image of the CT image of the human body at different ages. If the grayscale value of a pixel falls within the set threshold range, the pixel is regarded as part of the skeletal structure and is retained in the final skeletal grayscale image. If the grayscale value is not within the threshold range, it is regarded as a non-skeletal structure and is set to black in the final skeletal grayscale image, thereby segmenting the skeletal grayscale images of CT images of the human body at different ages.

[0076] Among them, a skeletal image data model of CT images of the human body at different ages is constructed based on the skeletal image data model of CT images of the human body at different ages.

[0077] In this embodiment, grayscale images of bones from CT images of the human body at different age stages are extracted to obtain the geometric and structural features of the bones. A bone image data model of CT images of the human body at different age stages is constructed using decision tree regression. Decision tree regression is a method specifically designed to predict the values ​​of continuous variables. It models and predicts data based on a tree structure. It achieves the regression task by dividing the dataset into different regions and predicting a constant value within each region. Each internal node of the tree represents a feature, and each leaf node represents an output value. It selects the best feature for data partitioning, ensuring that the output values ​​of the partitioned subsets are as close as possible to the true values. This embodiment collects CT image data from different age groups, covering a diverse population. Information such as gender, race, and health status is collected in the images; these images are typically stored in DICOM (the standard format for medical imaging) files. Geometric features (bone volume, shape, surface area) and structural features (bone tissue, trabecular structure, periosteum, medullary cavity, and bone connections) are extracted from the bones. These features are used as input to the model, and model construction parameters are set through decision tree regression. This allows for the construction of skeletal image data models of CT images at various age stages. These CT image skeletal image data models can output bone features and skeletal attributes for each age stage, including bone density, bone weight, bone toughness, bone age, and bone shape and contour.

[0078] Specifically, preset bone density values, preset bone weight values, and preset bone toughness values ​​are obtained from the database, and combined with the bone image data model of CT images of human bodies at different ages, the bone density values, bone weight values, and bone toughness values ​​of human bodies at different ages are output, and processed to obtain the bone conformity characterization values ​​of human bodies at different ages. The bone conformity characterization values ​​of human bodies at different ages are used to judge the accuracy of the bone image data model construction of CT images of human bodies at different ages.

[0079] In this embodiment, the bone density, bone weight, and bone toughness values ​​of the human body at each age stage are extracted from the bone image data model output by CT images of the human body at each age stage. The bone density, bone weight, and bone toughness values ​​of the human body at each age stage are processed to obtain the bone conformity characterization values ​​of the human body at each age stage. The preset bone density, bone weight, and bone toughness values ​​are obtained from the database to obtain the deviation data of bone volume of the human body at each age stage. The deviation data of bone toughness of the human body at each age stage is obtained by comparing the bone toughness threshold with the bone toughness of the human body at each age stage.

[0080] In this embodiment, the skeletal image data model of CT images of the human body at different ages is evaluated. This helps to intuitively determine the accuracy of the skeletal image data model of CT images of the human body at different ages. When the accuracy is low, corresponding measures are taken to improve the accuracy of the model.

[0081] The skeletal conformity values ​​for different age stages are calculated using the following formula:

[0082]

[0083] In the formula, σ i This represents the skeletal characteristics of the human body at the i-th age stage, where i = 1, 2, ..., n, i represents the age stage number, n represents the total number of age stages, P0 represents the preset bone mineral density value, Q0 represents the preset bone weight value, S0 represents the preset bone toughness value, and P... i Q represents the bone mineral density value at the i-th age stage in the human body. i S represents the bone mass value of the human body at the i-th age stage. i Let represent the bone toughness value at the i-th age stage of the human body, and e represent the natural constant.

[0084] In this embodiment, preset bone density values, preset bone weight values, and preset bone toughness values ​​are obtained from the database. These values ​​are then combined with bone density values, bone weight values, and bone toughness values ​​for different age stages to obtain the skeletal conformity characterization values ​​for different age stages. The preset bone density values ​​are represented by the average of historical bone density values ​​obtained directly from the database. The preset bone weight values ​​are represented by the average of historical bone weight values ​​obtained directly from the database. The preset bone toughness values ​​are represented by the average of historical bone toughness values ​​obtained directly from the database.

[0085] This embodiment combines bone mineral density, bone mass, and bone toughness values ​​at different age stages to obtain the skeletal conformity characterization values ​​for each age stage. In this formula, there is a common regulatory relationship between bone mineral density, bone mass, and bone toughness values ​​at different age stages. For example, high bone mineral density is usually associated with strong bone toughness, and increased bone mass may lead to increased bone toughness. The comprehensive assessment of the three helps to gain a more comprehensive understanding of bone health status.

[0086] Specifically, let's assume that the bone mineral density value P of the human body at the i-th age stage is... i The value is 1.0-1.5 grams per cubic centimeter (g / cm³). 3 The bone mass value Q of the human body at the i-th age stage. i The value is 18-28 kg, and the bone toughness value S of the human body in the i-th age stage. i The value is 300-2100 kg, such as Figure 2 As shown, this is a statistical chart of the changes in the skeletal conformation characterization values ​​provided in an embodiment of the present invention. Figure 2 As shown, when the preset bone mineral density is 1.0 g / cm³, the preset bone weight is 22.5 kg, and the preset bone toughness is 1800 kg, the bone weight value Q for the i-th age stage of the human body is... i The bone density value is 20 kg. As the bone density value of the human body increases in the i-th age stage, the bone conformity characteristic value of the human body in the i-th age stage gradually increases. When the bone toughness value of the human body in the i-th age stage is 700, the bone conformity characteristic value curve of the human body in the i-th age stage is shown in Figure a. When the bone toughness value of the human body in the i-th age stage is 1500, the bone conformity characteristic value curve of the human body in the i-th age stage is shown in Figure b. When the bone toughness value of the human body in the i-th age stage is 2200, the bone conformity characteristic value curve of the human body in the i-th age stage is shown in Figure c. By analyzing the bone density value, bone weight value and bone toughness value of the human body in the i-th age stage, the performance of the bone model is judged.

[0087] The process involves extracting skeletal conformance values ​​for different age groups of the human body and comparing them with preset skeletal conformance value thresholds. If the skeletal conformance values ​​for different age groups of the human body are higher than or equal to the preset skeletal conformance value thresholds, it indicates that the model performance is good. If the skeletal conformance values ​​for different age groups of the human body are lower than the preset skeletal conformance value thresholds, more CT images of different age groups of the human body are obtained to increase the training data of the model.

[0088] In this embodiment, the preset skeletal conformance value threshold is represented by the average value of skeletal conformance values ​​in the database. When the skeletal conformance value of each age group is higher than or equal to the preset threshold, it indicates that the output data of the skeletal image data model of CT images of each age group matches the true value, and the performance of the skeletal image data model of CT images of each age group is good. If the skeletal conformance value of each age group is lower than the preset threshold, it indicates that the output data of the skeletal image data model of CT images of each age group does not match the true value, and the performance of the skeletal image data model of CT images of each age group is poor. It is necessary to collect more CT image data, including CT images of different genders and different skeletal parts in each age group, to increase the sample size. For example, this can help improve the prediction accuracy of the model and enable it to more accurately reflect the true situation of the data.

[0089] Compared with the prior art, the present invention has the following advantages:

[0090] In this embodiment of the invention, a dataset of human skeletal CT images of different age groups is acquired. By labeling and analyzing the ROI regions of the skeleton in the CT images, ROI region conformance values ​​for each age group are obtained. After preliminary optimization, the CT images are further analyzed, HU-Histogram curves are fitted, Histogram feature conformance values ​​are extracted, and secondary optimization is performed based on these features. Finally, skeletal image data models for different age groups are constructed, and their performance is analyzed to obtain skeletal conformance characterization values, which can improve the accuracy of skeletal structure analysis.

[0091] In this embodiment of the invention, CT images are processed to generate grayscale images, noise is removed, and bone contours are extracted to identify and extract the Regions of Interest (ROIs) of the bones. Within the extracted bone ROIs, voxels are traversed to collect feature information, obtaining the Hub (HU) value for each voxel. By analyzing these HU values, features such as bone volume, peak and trough values ​​of the HU values, and edge length are extracted to calculate the conformity value of the bone ROI, thus evaluating its conformity. Finally, the conformity value is compared with a threshold to determine whether to preprocess or sharpen and remove artifacts from the CT images, effectively improving the quality of the CT images and making subsequent analysis more accurate.

[0092] In this embodiment of the invention, the width and average HU value of the skeletal HU-Histogram curves for each age group are extracted and compared with standard curves in the database to calculate the peak position offset, thereby obtaining the Histogram feature conformance value. By comparing these conformance values ​​with stored thresholds, it is determined whether to enhance the contrast and brightness of the skeletal ROI region, or to enhance and segment the image, providing a data foundation for subsequent model construction.

[0093] In this embodiment of the invention, grayscale image segmentation is performed on CT images of the human body at different ages, and the grayscale images of bones and their geometric and structural features are extracted as input to construct a bone image data model. By comparing the bone conformity representation value with a preset threshold, the quality of model construction is evaluated, and a decision is made on whether to add training data to optimize the model. The model is flexibly adjusted and optimized to ensure its quality.

[0094] The following points need to be explained:

[0095] (1) The accompanying drawings of the embodiments of the present invention only involve the structures involved in the embodiments of the present invention. Other structures can refer to the general design.

[0096] (2) For clarity, the thickness of layers or regions is enlarged or reduced in the drawings used to describe embodiments of the present invention; that is, these drawings are not drawn to actual scale. It is understood that when an element such as a layer, film, region, or substrate is referred to as being “above” or “below” another element, the element may be “directly” located “above” or “below” the other element, or there may be intermediate elements.

[0097] (3) Where there is no conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other to obtain new embodiments.

[0098] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for constructing a skeletal image data model based on CT images, characterized in that, This includes obtaining a human skeleton CT image dataset, which includes CT images of the human body at various age stages. The skeletal ROI regions are marked in CT images of the human body at different ages. The skeletal ROI regions of the human body at different ages are analyzed to obtain the ROI region compliance value of the human body at different ages. The CT images of the human body at different ages are then preliminarily optimized. The CT images of the human body at different ages were analyzed after preliminary optimization. The HU-Histogram curves of the human body at different ages were fitted and processed to obtain the Histogram feature conformance values ​​of the human body at different ages. Based on the Histogram feature conformance values ​​of the human body at different ages, the CT images of the human body at different ages were optimized again. The specific process for obtaining the Histogram feature conformity values ​​of the human body at different age stages is as follows: The width and mean HU value of the HU-Histogram curves of human bones at different ages are extracted from the HU-Histogram curves of human bones at different ages. The standard HU-Histogram curves of human bones at different ages stored in the database are extracted and compared with the standard HU-Histogram curves of human bones at different ages to obtain the peak position offset of the standard HU-Histogram curves of human bones at different ages. The Histogram feature conformance values ​​of human bones at different ages are processed to obtain the Histogram feature conformance values ​​of human bones at different ages. The Histogram feature conformance values ​​of human bones at different ages are used to evaluate the degree of difference in bone density of human bones in CT images at different ages. The analysis process for the histogram characteristics of the human body at different age stages is as follows: ； In the formula, This represents the Histogram feature conformity value for the i-th age stage of the human body. , where i represents the age group number and n represents the total number of age groups. This represents the width of the Histogram curve for the i-th age stage of the human body. This represents the mean HU value of the Histogram curve for the i-th age stage of the human body. This represents the shift in the peak position of the HU-Histogram curve of the human skeleton at the i-th age stage. This represents the standard value of the width of the Histogram curve for the i-th age stage of the human body. The standard value of the mean HU value represents the Histogram curve for the i-th age stage of the human body. This represents the standard value of the peak position offset of the HU-Histogram curve for the i-th age stage of the human body. This represents the correction factor corresponding to the width of the preset Histogram curve. This represents the correction factor corresponding to the mean HU value of the preset Histogram curve. This represents the correction factor corresponding to the peak position offset of the preset HU-Histogram curve. Extract CT images of the human body at different ages after secondary optimization, obtain skeletal image data model construction data for CT images of the human body at different ages, construct skeletal image data models for CT images of the human body at different ages, obtain skeletal conformity characterization values ​​for the human body at different ages, and provide performance analysis and suggestions for the skeletal image data models for CT images of the human body at different ages.

2. The method for constructing a skeletal image data model based on CT images according to claim 1, characterized in that, The specific process for obtaining the skeletal ROI regions of the human body at different age stages is as follows: Grayscale processing is performed on CT images of the human body at different ages to obtain grayscale images of CT images of the human body at different ages. Noise is removed from the grayscale images of CT images of the human body at different ages, and the skeletal contours on the grayscale images of CT images of the human body at different ages are extracted. The skeletal ROI regions for each age group are obtained from the skeletal contours on grayscale images of CT images of the human body at different ages.

3. The method for constructing a skeletal image data model based on CT images according to claim 2, characterized in that, The specific process for obtaining the ROI region conformity values ​​for different age stages of the human body is as follows: The voxels in the skeletal ROI region of the human body at different ages are traversed, and the feature information of each voxel is extracted and collected to obtain the HU value of each voxel in the skeletal ROI region of the human body at different ages in CT images. The volume of bone covered by the ROI region, the peak value, the valley value, and the edge length of the HU value are extracted from the HU values ​​of each voxel in the skeletal ROI region of CT images of human body at different ages. The ROI region conformity value of human body at different ages is obtained by processing the ROI region conformity value of human body at different ages. The ROI region conformity value of human body at different ages is used to evaluate the degree of conformity of ROI regions of human body at different ages. The ROI region conformity values ​​of each age group of the human body are compared with the ROI region conformity value thresholds of each age group of the human body. If the ROI region conformity values ​​of each age group of the human body are higher than or equal to the ROI region conformity value thresholds of each age group of the human body, the CT images of each age group of the human body are directly preprocessed. If the ROI region conformity values ​​of each age group of the human body are lower than the ROI region conformity value thresholds of each age group of the human body, the CT images are sharpened and artifact removal operations are performed.

4. The method for constructing a skeletal image data model based on CT images according to claim 2, characterized in that, The specific process of analyzing the preliminarily optimized CT images of the human body at different age stages and fitting the HU-Histogram curves for each age stage is as follows: The HU values ​​in the skeletal ROI regions of different age groups of the human body are divided into intervals to obtain the HU value intervals of the skeletal ROI regions of different age groups. The number of voxels of HU values ​​in each HU value interval of the skeletal ROI regions of different age groups of the human body is counted, thereby constructing the skeletal HU-Histogram curves of different age groups of the human body.

5. The method for constructing a skeletal image data model based on CT images according to claim 4, characterized in that, The specific process of performing secondary optimization on CT images of the human body at different age stages based on the Histogram feature conformity values ​​of the human body at different age stages is as follows: The histogram feature values ​​of each age group of the human body are extracted and compared with the histogram feature matching thresholds of each age group stored in the database. If the histogram feature matching values ​​of each age group of the human body are higher than or equal to the histogram feature matching thresholds of each age group of the human body, the contrast and brightness of the skeletal ROI regions of each age group of the human body are enhanced. If the histogram feature matching values ​​of each age group of the human body are lower than the histogram feature matching thresholds of each age group of the human body, the CT images of each age group of the human body are enhanced and segmented.

6. The method for constructing a skeletal image data model based on CT images according to claim 1, characterized in that, The specific process of extracting the optimized CT images of the human body at different age stages and obtaining the skeletal image data model construction data of the CT images of the human body at different age stages is as follows: The grayscale images of CT images of the human body at different ages after secondary optimization are extracted and segmented to obtain the segmented grayscale images of CT images of the human body at different ages after secondary optimization. The geometric and structural features of the skeleton are obtained based on the segmented grayscale images of CT images of the human body at different ages after secondary optimization. The geometric and structural features of bones are combined to form the bone image data model for CT images of the human body at different ages.

7. The method for constructing a skeletal image data model based on CT images according to claim 6, characterized in that, The specific process for constructing skeletal image data models of CT images of the human body at different age stages is as follows: Based on the skeletal image data model of CT images of the human body at different ages, a skeletal image data model of CT images of the human body at different ages is constructed.

8. The method for constructing a skeletal image data model based on CT images according to claim 7, characterized in that, The specific process for obtaining the skeletal conformity characteristics of the human body at different age stages is as follows: Preset bone density, bone weight, and bone toughness values ​​are obtained from the database. Combined with the bone image data model of CT images of human body at different ages, the bone density, bone weight, and bone toughness values ​​of human body at different ages are output. The bone conformity characterization values ​​of human body at different ages are processed to obtain the bone conformity characterization values ​​of human body at different ages. The bone conformity characterization values ​​of human body at different ages are used to judge the accuracy of the bone image data model of CT images of human body at different ages.

9. The method for constructing a skeletal image data model based on CT images according to claim 8, characterized in that, The specific process for performance analysis and feedback on the skeletal image data model of CT images of the human body at different age stages is as follows: The skeletal conformance values ​​of human bodies at different ages are extracted and compared with preset skeletal conformance value thresholds. If the skeletal conformance values ​​of human bodies at different ages are higher than or equal to the preset skeletal conformance value thresholds, it indicates that the model performance is good. If the skeletal conformance values ​​of human bodies at different ages are lower than the preset skeletal conformance value thresholds, more CT images of human bodies at different ages are obtained to increase the training data of the model.

Images