Image processing method and system for scoliosis detection

Abstract

The invention discloses an image processing method and system for scoliosis detection, and relates to the field of image processing technology. The image processing method for scoliosis detection obtains spinal image data of a patient to be detected and performs preprocessing; performs regional extraction on the spinal image data of the patient to be detected after preprocessing to obtain spinal regional data; performs regional division on the spinal regional data to obtain the cervical spine region, the thoracic spine region, and the lumbar spine region. The invention can more accurately evaluate the structural state of the cervical spine, the thoracic spine, and the lumbar spine through detailed regional division and regional analysis. Compared with the traditional method of relying only on the overall spinal image for single measurement, this partition processing method can better capture the abnormality of the local area, improve the accuracy of scoliosis detection, and effectively reduce the measurement error and improve the reliability of scoliosis detection by comprehensively analyzing the scoliosis index of each area.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to an image processing method and system for scoliosis detection. Background Art

[0002] Scoliosis is a common spinal deformity disease, which is usually manifested as the spine deviating from the normal vertical straight line in the front view, forming an abnormal lateral curvature. This disease often occurs in adolescence. If it is not discovered and treated in time, it may cause serious health problems, including abnormal posture, breathing difficulties and spinal dysfunction. Accurately detecting the degree and location of scoliosis is crucial to formulating an effective treatment plan.

[0003] Prior art, such as the invention patent application with announcement number: CN117765062B, discloses an image processing method and system for detecting adolescent scoliosis, which includes obtaining a training set of a classification model to train the classification model, obtaining a trained classification model, inputting the scoliosis image to be classified into the trained classification model, and classifying the scoliosis image; inputting the scoliosis image to be predicted into the trained deep learning model, and predicting the center point of the spine; the scoliosis image includes a single-bend image of the spine, a double-bend image of the spine, and a triple-bend image of the spine; different models are selected according to different image structures to predict the center point of the spine, so that the output center point of the spine is more accurate, overcoming the difficulty of predicting the center point due to the different structures of the curved segments of the spine, thereby improving the calculation accuracy of the Cobb angle.

[0004] Based on the above scheme, it is found that the limitations of the existing technology include at least the following problems. First, the existing scoliosis detection method mainly relies on the deep learning model to predict the center point of the spine, and then determines the bending end point of the spine based on the center point, and calculates the Cobb angle. Although this method improves the measurement accuracy of the Cobb angle to a certain extent, due to the complexity of the spine, especially in the presence of multiple bends (such as S-shaped or C-shaped spine), the model's prediction of the center point of the spine still has errors, resulting in inaccurate Cobb angle measurement results. In addition, the detection method of the existing technology relies more on a single image (such as an X-ray image) for analysis, and does not fully utilize the advantages of multimodal imaging data, which can easily miss some detailed information and reduce the reliability of the overall detection. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides an image processing method and system for scoliosis detection, which solves the problem that the existing scoliosis detection method mainly relies on a deep learning model to predict the center point of the spine, and then determines the bending end point of the spine based on the center point, and calculates the Cobb angle. Although this method improves the measurement accuracy of the Cobb angle to a certain extent, due to the complexity of the spine, especially in the presence of multiple bends (such as S-shaped or C-shaped spine), the model's prediction of the center point of the spine still has errors, resulting in inaccurate Cobb angle measurement results. In addition, the detection method of the prior art relies more on a single image (such as an X-ray image) for analysis, and does not fully utilize the advantages of multimodal imaging data, which easily misses some detailed information, reducing the reliability of the overall detection.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: an image processing method for scoliosis detection, comprising the following steps: obtaining spinal image data of a patient to be detected and performing preprocessing; performing regional extraction on the spinal image data of the patient to be detected after preprocessing to obtain spinal regional data; performing regional division on the spinal regional data to obtain cervical spine region, thoracic spine region, and lumbar spine region; performing data analysis on the cervical spine region, thoracic spine region, and lumbar spine region respectively to obtain scoliosis indexes of the cervical spine region, thoracic spine region, and lumbar spine region; judging and analyzing the scoliosis indexes of the cervical spine region, thoracic spine region, and lumbar spine region respectively with preset corresponding regional thresholds, marking the regions that do not meet the corresponding regional thresholds, and analyzing the degree of regional scoliosis.

[0007] Furthermore, the spinal image data specifically includes the pixel value of each pixel point of the spinal column, and the spinal region data specifically includes the grayscale spinal pixel value of each spinal grayscale pixel point in the spinal image data.

[0008] Furthermore, the specific process of preprocessing the spinal image data of the patient to be detected is: performing denoising processing on the spinal image data of the patient to be detected, and performing contrast enhancement processing on the spinal image data of the patient to be detected after the denoising processing.

[0009] Furthermore, the specific process of obtaining the spinal area data is as follows: reading the spinal image data of the patient to be tested after contrast enhancement processing and performing grayscale processing to obtain the grayscale pixel value of each pixel point containing the spine; performing grayscale analysis on the grayscale pixel value of each pixel point containing the spine to obtain a grayscale histogram; using the grayscale pixel value corresponding to the peak value in the grayscale histogram as the pixel judgment threshold; based on the pixel judgment threshold, screening the grayscale pixel value of each pixel point containing the spine to obtain the grayscale spine pixel value of each spinal grayscale pixel point in the spinal image data.

[0010] Furthermore, the specific steps for obtaining the cervical spine region, thoracic spine region, and lumbar spine region are as follows: connect each spine grayscale pixel point located at the boundary in the spine region data and regard them as the spine contour; perform mean analysis on the horizontal coordinate values ​​of several spine grayscale pixel points located at the bottom of the spine contour to obtain the horizontal coordinate value of the center of the bottom of the spine; draw a vertical line perpendicular to the horizontal coordinate value of the center of the bottom of the spine toward the top of the spine contour in the spine region data, and use the vertical coordinate value of the spine grayscale pixel point at the top of the spine contour intersecting with the vertical line as the vertical coordinate value of the center of the top of the spine; read the vertical coordinate value of the center of the bottom of the spine, and perform length analysis combined with the vertical coordinate value of the center of the top of the spine to obtain the total length of the spine; input the vertical coordinate value of the center of the top of the spine and the total length of the spine into the regional division model for regional division to obtain the cervical spine region, thoracic spine region, and lumbar spine region.

[0011] Furthermore, the area division model is specifically as follows: ;in, is the ordinate of the end of the cervical region of the spine, is the vertical coordinate value of the center of the top of the spine, is the cervical spine area ratio coefficient stored in the database, is the total length of the spine, is the ordinate of the end of the thoracic region of the spine, is the thoracic vertebra area ratio stored in the database, is the ordinate of the end of the lumbar region of the spine, is the lumbar area ratio coefficient stored in the database, .

[0012] Furthermore, the specific steps for obtaining the scoliosis index of the cervical spine, thoracic spine, and lumbar spine are as follows: obtain the reference curvature values ​​of the cervical spine, thoracic spine, and lumbar spine; analyze the average curvature values, regional length values, and regional spine vertical displacement values ​​of the cervical spine, thoracic spine, and lumbar spine based on the spine region data; conduct a comprehensive analysis of the reference curvature values ​​and the average curvature values, regional length values, and regional spine vertical displacement values ​​of the cervical spine, thoracic spine, and lumbar spine regions to obtain the scoliosis index of the cervical spine, thoracic spine, and lumbar spine regions.

[0013] Furthermore, the specific formulas for calculating the scoliosis index of the cervical spine region, the thoracic spine region, and the lumbar spine region are as follows: ;in, is the scoliosis index of the cervical spine area, is the average curvature of the cervical spine, is the reference curvature value of the cervical spine area, is the regional spinal vertical displacement value of the cervical spine area, is the length of the cervical spine area, is the scoliosis index of the thoracic spine region, is the average curvature of the thoracic region of the spine, is the reference curvature value of the thoracic spine region, is the regional spinal vertical displacement value of the thoracic spine area, is the length of the thoracic spine region, is the scoliosis index of the lumbar region of the spine, is the average curvature of the lumbar region of the spine, is the reference curvature value of the lumbar region of the spine, is the regional spinal vertical displacement value of the lumbar region of the spine, It is the length value of the lumbar region of the spine.

[0014] An image processing system for scoliosis detection includes: a data acquisition module, a region extraction module, a region division module, a scoliosis analysis module, and an abnormal marking module; the data acquisition module is used to acquire spinal image data of a patient to be detected and perform preprocessing; the region extraction module is used to perform region extraction on the spinal image data of the patient to be detected after preprocessing to obtain spinal region data; the region division module is used to perform region division on the spinal region data to obtain cervical spine region, thoracic spine region, and lumbar spine region; the scoliosis analysis module is used to perform data analysis on the cervical spine region, thoracic spine region, and lumbar spine region, respectively, to obtain scoliosis indexes of the cervical spine region, thoracic spine region, and lumbar spine region; the abnormal marking module is used to judge and analyze the scoliosis indexes of the cervical spine region, thoracic spine region, and lumbar spine region with preset corresponding region thresholds, mark the regions that do not meet the corresponding region thresholds, and analyze the degree of regional scoliosis.

[0015] The present invention has the following beneficial effects:

[0016] (1) This image processing method for scoliosis detection can more accurately evaluate the structural status of the cervical, thoracic and lumbar vertebrae through detailed regional division and regional analysis. Compared with the traditional method of relying solely on the overall spinal image for single measurement, this partitioning processing method can better capture abnormalities in local areas and improve the accuracy of scoliosis detection. In addition, by comprehensively analyzing the scoliosis index of each area, the measurement error is effectively reduced and the reliability of scoliosis detection is improved.

[0017] (2) This image processing method for scoliosis detection effectively improves the visibility of the spinal region by applying advanced denoising and contrast enhancement techniques in the image preprocessing stage, ensuring the accuracy of subsequent data analysis. By extracting and analyzing the grayscale value of each pixel, this method can accurately identify and process the fine structure of the spine. This sophisticated image processing method makes the segmentation and analysis of the spinal region more accurate, thereby providing more detailed and reliable data support in the entire detection process.

[0018] (3) The image processing method for scoliosis detection effectively identifies the specific location and degree of scoliosis by marking and further analyzing the abnormal areas that do not meet the preset threshold. This marking process not only provides an intuitive reference for clinical diagnosis, but also provides doctors with more targeted treatment suggestions through in-depth analysis of the abnormal areas. This comprehensive analysis and marking mechanism greatly improves the detection sensitivity and diagnostic accuracy of abnormal spinal areas, and provides a scientific basis for the patient's subsequent treatment.

[0019] (4) The image processing system for scoliosis detection, through modular design, decomposes the key steps in the scoliosis detection process into independent functional modules, including data acquisition module, region extraction module, region division module, scoliosis analysis module and abnormality marking module. This design not only simplifies the development and maintenance of the system, but also facilitates the subsequent upgrade and expansion of the system. The independence of each module enables the system to more flexibly adapt to different application scenarios, such as different types of spinal image data processing or analysis of specific clinical needs. Through modular design, the system can easily introduce new algorithms or technologies, thereby improving the adaptability of the system and the possibility of future expansion, thereby providing more comprehensive and flexible technical support for scoliosis detection.

[0020] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a flow chart of an image processing method for scoliosis detection according to the present invention.

[0022] Figure 2 The present invention is a flowchart of the specific steps for obtaining spinal region data in an image processing method for scoliosis detection.

[0023] Figure 3 This is a block diagram of an image processing system for scoliosis detection according to the present invention. DETAILED DESCRIPTION

[0024] The embodiment of the present application solves the problem that the existing scoliosis detection method mainly relies on a deep learning model to predict the center point of the spine, then determines the bending end point of the spine based on the center point, and calculates the Cobb angle. Although this method improves the measurement accuracy of the Cobb angle to a certain extent, due to the complexity of the spine, especially in the presence of multiple bends (such as S-shaped or C-shaped spines), the model's prediction of the center point of the spine still has errors, resulting in inaccurate Cobb angle measurement results. In addition, the detection method of the prior art relies more on a single image (such as an X-ray image) for analysis, and does not fully utilize the advantages of multimodal imaging data, which makes it easy to miss some detailed information, reducing the reliability of the overall detection.

[0025] The overall idea of ​​the problem in the embodiment of this application is as follows:

[0026] First, the spinal image data of the patient to be tested is obtained, and the image is preprocessed, including denoising and contrast enhancement processing to ensure the clarity and accuracy of the image. The preprocessed spinal image data is subjected to regional extraction to obtain the pixel value of each pixel point containing the spine. Then, the spinal regional data is further divided into cervical region, thoracic region and lumbar region. Data analysis is performed on each region to obtain its average curvature value, regional length value and vertical displacement value. Subsequently, these analysis results are compared with the preset reference curvature value, and the scoliosis index of each region is comprehensively calculated. The scoliosis index of each region is judged and analyzed with the preset corresponding region threshold. The regions that do not meet the threshold are marked, and the scoliosis degree of these regions is further analyzed.

[0027] See also Figure 1 The embodiment of the present invention provides a technical solution: an image processing method for scoliosis detection, comprising the following steps: obtaining spinal X-ray image data of a patient to be detected and performing preprocessing; performing regional extraction on the spinal X-ray image data of the patient to be detected after preprocessing to obtain spinal regional data; performing regional division on the spinal regional data to obtain cervical spine region, thoracic spine region, and lumbar spine region; performing data analysis on the cervical spine region, thoracic spine region, and lumbar spine region respectively to obtain scoliosis indexes of the cervical spine region, thoracic spine region, and lumbar spine region; judging and analyzing the scoliosis indexes of the cervical spine region, thoracic spine region, and lumbar spine region respectively with preset corresponding regional thresholds, marking the regions that do not meet the corresponding regional thresholds, and analyzing the degree of regional scoliosis.

[0028] The specific process of analyzing the degree of regional scoliosis is as follows: Once the areas with scoliosis index outside the normal range are identified, the next step is to mark these areas on the image:

[0029] Visual marking: Use image processing software (such as OpenCV, MATLAB, etc.) to highlight abnormal areas on the original spinal X-ray image. This usually involves drawing markers or color coding on the image to make the abnormal area clearly visible.

[0030] Detailed analysis of marked abnormal areas to determine possible health problems and the need for further testing or treatment:

[0031] Detailed evaluation: Based on the specific value of the scoliosis index and the location of the abnormal area, possible pathological changes, such as the degree of scoliosis, local structural changes, etc., are evaluated.

[0032] Clinical control: The image analysis results are compared with clinical symptoms and patient history to evaluate the correlation between the abnormalities shown in the image and the patient's actual symptoms.

[0033] Recommended further testing: Depending on the results of the analysis, more detailed imaging studies (such as MRI) or functional testing may be recommended to obtain more information about the structural and functional status of the spine.

[0034] Finally, all analysis results and recommendations are summarized into a detailed diagnostic report:

[0035] Diagnostic report: includes the detection method, abnormal areas found, specific scoliosis index, relevant clinical significance explanation and follow-up examination or treatment recommendations.

[0036] Patient Communication: Ensure reports are clear and easy to understand to facilitate communication between doctors and patients, and ensure patients fully understand the test results and next steps.

[0037] The spinal X-ray image data specifically includes the pixel value of each pixel point of the spine, and the spinal region data specifically includes the grayscale spine pixel value of each spinal grayscale pixel point in the spinal X-ray image data. These data are obtained through the hospital's recording system.

[0038] Specifically, the specific process of preprocessing the spinal X-ray image data of the patient to be detected is: performing denoising processing on the spinal X-ray image data of the patient to be detected, and performing contrast enhancement processing on the spinal X-ray image data of the patient to be detected after the denoising processing.

[0039] Among them, denoising adopts bilateral filtering. Bilateral filter is an effective denoising tool. It takes into account both spatial neighbors and pixel value similarity. It can retain edges while removing noise. Its principle is: bilateral filter combines the characteristics of domain filtering and value filtering. Domain filtering is based on the spatial proximity between pixels, while value filtering is based on the similarity of pixel values.

[0040] Contrast enhancement processing adopts CLAHE processing, which is a technology for improving local image contrast. It is particularly suitable for application scenarios in medical imaging that have high requirements for details. Its principle is: CLAHE limits the contrast amplification in the histogram equalization process, by dividing the image into small areas called "blocks" and then applying histogram equalization independently to avoid excessive enhancement of noise in uniform areas.

[0041] In this implementation, bilateral filtering, as an advanced denoising technology, can not only effectively remove noise from the image, but also retain important details such as the edge of the spine. This is because bilateral filtering takes into account both the spatial position of pixels and the similarity of pixel values, thereby avoiding image blurring while removing noise. For spinal X-ray images, this means that the structure of the spine can be displayed more clearly, especially for subtle bone edges and complex spinal curves, which helps to improve the accuracy of subsequent image analysis. CLAHE technology effectively improves the local contrast of the image by performing histogram equalization on the local area of ​​the image. Since the details and bone structure in the spinal X-ray image may not be obvious due to the overall brightness distribution, CLAHE can maintain the global brightness distribution while retaining the image. The contrast of these local areas is significantly enhanced, making the structure of the spine more clear and prominent in the image, which is of great significance to improving the accuracy and visibility of scoliosis detection, especially when the spinal curvature is small or the abnormality is relatively hidden. Through the denoising and contrast enhancement processing in the preprocessing stage, the unnecessary information and noise in the original X-ray image are effectively suppressed, and the structural information of the spine is significantly enhanced. This not only provides higher quality data input for subsequent regional extraction, division and analysis, but also helps to reduce errors in subsequent processing steps, ensuring that the analysis results are more reliable and accurate during the scoliosis detection process. Ultimately, these improvements can provide a more accurate reference basis for clinical diagnosis and support doctors to make more scientific treatment decisions.

[0042] Specifically, Figure 2As shown in FIG. 1 , the specific process of obtaining the spinal region data is as follows: the spinal X-ray image data of the patient to be detected after contrast enhancement processing is read and grayscale processing is performed, specifically, the RGB value is converted into a grayscale value through the formula Y=0.299R+0.587G+0.114B, wherein R, G, and B are the values ​​of the three channels of red, green, and blue, respectively, and Y is the obtained grayscale value, and the grayscale pixel value of each pixel point containing the spine is obtained; the grayscale pixel value of each pixel point containing the spine is grayscale analyzed to obtain a grayscale histogram, specifically, for each pixel in the image, the count of the corresponding grayscale level in the histogram is increased according to its grayscale value, and the grayscale pixel value is grayscaled. The grayscale level of the grayscale image is usually from 0 to 255, so the histogram will have 256 possible grayscale levels; the grayscale pixel value corresponding to the peak in the grayscale histogram is used as the pixel judgment threshold. The peak refers to the position of the highest point in the histogram, which usually corresponds to the most common grayscale level in the image. The peak can be found by searching for the maximum value in the histogram to complete the grayscale pixel value with the most pixels; based on the pixel judgment threshold, the grayscale pixel value of each pixel point containing the spine is screened (that is, the pixel points with grayscale pixel values ​​lower than the pixel judgment threshold are removed) to obtain the grayscale spine pixel value of each spine grayscale pixel point in the spine X-ray image data.

[0043] In this embodiment, the image processing process is simplified by converting the color image into a grayscale image. The color image contains three channels (red, green, and blue), and each channel needs to be processed separately, while the grayscale image only contains one channel, which significantly reduces the amount of data and improves the efficiency of image processing. In addition, the value of each pixel in the grayscale image directly reflects the brightness information of the position, which helps to capture the subtle structure of the spine more accurately, especially when identifying the spinal area, it can more effectively distinguish between bones and soft tissues. By constructing a grayscale histogram and analyzing its peak position, the most common grayscale level in the image can be determined, which usually corresponds to the main grayscale value of the spinal area. Using this information, a suitable threshold can be set to focus on extracting the spinal area. The relevant grayscale pixel values, this threshold screening method based on the grayscale histogram, helps to eliminate the interference information of the background and non-spinal areas, so as to more accurately extract the data of the spinal area and reduce the possibility of misidentification. By performing grayscale analysis and threshold screening on the image, it can provide cleaner and more accurate input data for subsequent image processing steps (such as region division and curvature analysis), and only retain grayscale pixels that meet the characteristics of the spinal area, which greatly reduces the impact of the non-spinal area on subsequent analysis and ensures the accuracy and effectiveness of subsequent processing steps. This preprocessing method not only improves the overall image processing effect, but also provides reliable basic data for the final scoliosis detection, ensuring the accuracy and credibility of the test results.

[0044] Specifically, the specific steps for obtaining the cervical spine region, thoracic spine region, and lumbar spine region are as follows: connect each spine grayscale pixel point located at the boundary in the spine region data and regard them as the spine contour; perform mean analysis on the horizontal coordinate values ​​of several spine grayscale pixel points located at the bottom of the spine contour to obtain the horizontal coordinate value of the center of the bottom of the spine; draw a vertical line perpendicular to the horizontal coordinate value of the center of the bottom of the spine in the spine region data toward the top of the spine contour, and use the vertical coordinate value of the spine grayscale pixel point at the top of the spine contour intersecting with the vertical line as the vertical coordinate value of the center of the top of the spine; read the vertical coordinate value of the center of the bottom of the spine, and perform length analysis in combination with the vertical coordinate value of the center of the top of the spine to obtain. The total length of the spine; the central ordinate value of the top of the spine and the total length of the spine are input into the regional division model for regional division to obtain the cervical spine region, thoracic spine region and lumbar spine region. The specific process is to input the central ordinate value of the top of the spine and the total length of the spine into the regional division model to calculate the ordinate of the terminal end of the cervical spine region, the thoracic spine region and the lumbar spine region. Taking the ordinate of the terminal end of the cervical spine region and the thoracic spine region as the reference, horizontal lines are drawn respectively, and the spinal region data is divided into three regions, namely, the cervical spine region, thoracic spine region and lumbar spine region are obtained.

[0045] The regional division model is as follows: ;in, is the ordinate of the end of the cervical region of the spine (i.e. the ordinate of the center point of the end), is the vertical coordinate value of the center of the top of the spine, is the cervical spine area ratio coefficient stored in the database, is the total length of the spine, is the ordinate of the end of the thoracic region of the spine (i.e. the ordinate of the center point of the end), is the thoracic vertebra area ratio stored in the database, is the ordinate of the end of the lumbar region of the spine (i.e. the ordinate of the center point of the end), is the lumbar area ratio coefficient stored in the database, .

[0046] It needs to be explained that , , The specific calculation process is as follows: obtain several groups of cervical vertebrae region lengths, thoracic vertebrae region lengths, and lumbar vertebrae region lengths, and perform mean analysis on them respectively to obtain the mean of the cervical vertebrae region length, the mean of the thoracic vertebrae region length, and the mean of the lumbar vertebrae region length. Then sum the three to obtain the regional sum value. Then perform ratio analysis on the mean of the cervical vertebrae region length, the mean of the thoracic vertebrae region length, and the mean of the lumbar vertebrae region length with the regional sum value. The ratio analysis result is the corresponding proportion coefficient.

[0047] In this implementation, the spine contour is accurately defined by connecting the grayscale pixel points on the border of the spine region, thereby ensuring that each part of the spine can be accurately identified and analyzed. The coordinate values ​​of the center points of the bottom and top of the spine are used to provide a stable and consistent reference point, which greatly improves the accuracy of the spine region division. Especially in clinical applications, ensuring the accurate division of each spinal region (cervical spine, thoracic spine, lumbar spine) is crucial for subsequent diagnosis and treatment plan design. The total length of the spine and the proportion of each region are calculated as key parameters, combined with the standard proportion coefficient stored in the database, to ensure the scientificity and repeatability of the division process. This division method based on statistical methods avoids It eliminates the influence of subjective factors, making the spinal region division of different patients consistent and comparable. In addition, the use of standard data in the database can be fine-tuned according to individual differences, so that each patient's regional division is more in line with their actual anatomical structure. The terminal vertical coordinates of the cervical, thoracic and lumbar regions of the spine are accurately calculated to clarify the boundaries of each region. This process not only ensures the independence of each spinal region, but also provides reliable basic data for the specific analysis of each region (such as curvature measurement, scoliosis index calculation). Accurate regional division makes subsequent scoliosis detection and analysis more accurate, provides high-quality imaging data support for clinicians, and helps to improve the accuracy of diagnosis and the effectiveness of treatment.

[0048] Specifically, the specific steps for obtaining the scoliosis index of the cervical spine, thoracic spine, and lumbar spine are as follows: obtain the reference curvature values ​​of the cervical spine, thoracic spine, and lumbar spine; analyze the average curvature values, regional length values, and regional spine vertical displacement values ​​(upper and lower offsets) of the cervical spine, thoracic spine, and lumbar spine based on the spine region data; conduct a comprehensive analysis of the reference curvature values ​​and the average curvature values, regional length values, and regional spine vertical displacement values ​​(upper and lower offsets) of the cervical spine, thoracic spine, and lumbar spine regions to obtain the scoliosis index of the cervical spine, thoracic spine, and lumbar spine regions.

[0049] The reference curvature value is obtained by consulting relevant literature or databases to find the normal curvature range of the cervical spine, thoracic spine and lumbar spine.

[0050] The specific process of analyzing the average curvature values ​​of the cervical spine region, the thoracic spine region, and the lumbar spine region is as follows: a vertical line perpendicular to the horizontal coordinate value of the center of the bottom of the spine is drawn toward the top of the spine contour in the spine regional data as the spine centerline, and the regional centerlines of the cervical spine region, the thoracic spine region, and the lumbar spine region are read respectively, and a mathematical fitting method such as polynomial fitting or spline curve fitting is used to describe the mathematical model of the regional centerlines in the cervical spine region, the thoracic spine region, and the lumbar spine region. This usually requires calculation from the centerline coordinate points obtained by skeletonization, and a polynomial of an appropriate order (such as quadratic or cubic) can be selected. times), use the least squares method to fit these points, or use a spline curve (such as a Bezier curve or a B-spline) for fitting, use a curvature formula, such as a curvature calculation formula for discrete points, calculate the curvature of each point on the regional center line in the cervical spine region, thoracic spine region, and lumbar spine region (calculated by combining the mathematical model of the regional center line in the cervical spine region, thoracic spine region, and lumbar spine region), average each curvature value in the cervical spine region, thoracic spine region, and lumbar spine region to obtain the average curvature value of the cervical spine region, thoracic spine region, and lumbar spine region.

[0051] The specific process of analyzing the regional length values ​​of the cervical spine region, the thoracic spine region, and the lumbar spine region is: the length values ​​of the regional center lines in the cervical spine region, the thoracic spine region, and the lumbar spine region are used as the corresponding regional length values.

[0052] The specific process of analyzing the regional spine vertical displacement values ​​(upper and lower offsets) of the cervical spine, thoracic spine, and lumbar spine regions is as follows: the regional spine vertical displacement values ​​(upper and lower offsets) are obtained by analyzing the differences in the vertical coordinates of the highest point and the lowest point in the cervical spine, thoracic spine, and lumbar spine regions.

[0053] The specific formula for calculating the scoliosis index of the cervical spine, thoracic spine, and lumbar spine is as follows: ;in, is the scoliosis index of the cervical spine area, is the average curvature of the cervical spine, is the reference curvature value of the cervical spine area, is the regional spinal vertical displacement value of the cervical spine area, is the length of the cervical spine area, is the scoliosis index of the thoracic spine region, is the average curvature of the thoracic region of the spine, is the reference curvature value of the thoracic spine region, is the regional spinal vertical displacement value of the thoracic spine area, is the length of the thoracic spine region, is the scoliosis index of the lumbar region of the spine, is the average curvature of the lumbar region of the spine, is the reference curvature value of the lumbar region of the spine, is the regional spinal vertical displacement value of the lumbar region of the spine, It is the length value of the lumbar region of the spine.

[0054] In this implementation, through the comprehensive analysis of reference curvature values, average curvature values, regional length values ​​and vertical displacement values, the structural changes of the spine can be comprehensively evaluated. This multi-dimensional analysis method helps to accurately capture abnormalities in various areas of the spine, especially for early or relatively hidden scoliosis problems, and can provide more accurate detection results. Compared with a single indicator, the combined use of these parameters can better reflect the severity of scoliosis and abnormalities in specific areas, ensuring the accuracy and reliability of the detection. This method ensures the scientific calculation of the curvature and displacement of different areas of the spine through systematic mathematical modeling and data analysis steps, uses mathematical fitting methods (such as polynomials or spline curves) to describe the centerline of the spine, and calculates the curvature value of each point through the curvature formula to ensure The standardization and consistency of the calculation process are ensured. Through these steps, errors caused by subjective judgment can be avoided and the standardization level of the overall analysis process can be improved. This systematic calculation method provides a solid foundation for subsequent clinical decision-making. By independently calculating and analyzing the scoliosis index of the cervical, thoracic and lumbar spine, it can provide more personalized diagnostic support for different spinal regions of patients. Since the scoliosis index of each region is calculated based on specific reference curvature values ​​and individual parameters, it can more accurately reflect individual differences, thereby providing doctors with personalized treatment recommendations. This targeted analysis method ensures the accuracy of the diagnostic results, and can tailor the most appropriate treatment plan for the patient's specific situation, thereby improving the effectiveness of diagnosis and treatment.

[0055] See also Figure 3The embodiment of the present invention provides a technical solution: an image processing system for scoliosis detection, comprising: a data acquisition module, a region extraction module, a region division module, a scoliosis analysis module, and an abnormal marking module; the data acquisition module is used to acquire the spinal X-ray image data of the patient to be detected and perform preprocessing; the region extraction module is used to perform region extraction on the spinal X-ray image data of the patient to be detected after preprocessing to obtain spinal region data; the region division module is used to perform region division on the spinal region data to obtain the cervical spine region, the thoracic spine region, and the lumbar spine region; the scoliosis analysis module is used to perform data analysis on the cervical spine region, the thoracic spine region, and the lumbar spine region respectively to obtain the scoliosis index of the cervical spine region, the thoracic spine region, and the lumbar spine region; the abnormal marking module is used to judge and analyze the scoliosis index of the cervical spine region, the thoracic spine region, and the lumbar spine region respectively with the preset corresponding region thresholds, mark the regions that do not meet the corresponding region thresholds, and analyze the regional scoliosis degree.

[0056] In summary, this application has at least the following effects:

[0057] Through detailed regional division and region-by-region analysis, the structural status of the cervical, thoracic and lumbar vertebrae of the spine can be evaluated more accurately. Compared with the traditional method of relying solely on the overall spinal image for a single measurement, this partitioning processing method can better capture abnormalities in local areas and improve the accuracy of scoliosis detection. In addition, by comprehensively analyzing the scoliosis index of each area, the measurement error is effectively reduced and the reliability of scoliosis detection is improved.

[0058] By applying advanced denoising and contrast enhancement techniques in the image preprocessing stage, the visibility of the spinal area is effectively improved, ensuring the accuracy of subsequent data analysis. By extracting and analyzing the grayscale value of each pixel, this method can accurately identify and process the fine structure of the spine. This sophisticated image processing method makes the segmentation and analysis of the spinal area more accurate, thereby providing more detailed and reliable data support throughout the detection process.

[0059] After detecting abnormal areas that do not meet the preset threshold, the specific location and degree of scoliosis can be effectively identified through marking and further analysis. This marking process not only provides an intuitive reference for clinical diagnosis, but also through in-depth analysis of abnormal areas, it can provide doctors with more targeted treatment suggestions. This comprehensive analysis and marking mechanism greatly improves the detection sensitivity and diagnostic accuracy of abnormal spinal areas, and provides a scientific basis for the patient's subsequent treatment.

[0060] Through modular design, the key steps in the scoliosis detection process are decomposed into independent functional modules, including data acquisition module, region extraction module, region division module, scoliosis analysis module and abnormality marking module. This design not only simplifies the development and maintenance of the system, but also facilitates the subsequent upgrade and expansion of the system. The independence of each module enables the system to adapt more flexibly to different application scenarios, such as different types of spinal image data processing or analysis of specific clinical needs. Through modular design, the system can easily introduce new algorithms or technologies, improve the adaptability of the system and the possibility of future expansion, thereby providing more comprehensive and flexible technical support for scoliosis detection.

[0061] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0062] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. An image processing method for scoliosis detection, characterized in that: The following steps are involved: Acquire spinal image data of the patient to be tested and perform preprocessing; Performing region extraction on the spinal image data of the patient to be tested after preprocessing to obtain spinal region data; The spinal region data is divided into regions to obtain the cervical spine region, the thoracic spine region, and the lumbar spine region; The data of the cervical spine region, thoracic spine region, and lumbar spine region are analyzed respectively to obtain the scoliosis index of the cervical spine region, thoracic spine region, and lumbar spine region; The scoliosis index of the cervical spine region, the thoracic spine region, and the lumbar spine region are judged and analyzed with the preset thresholds of the corresponding regions, and the regions that do not meet the thresholds of the corresponding regions are marked; The specific steps for obtaining the scoliosis index of the cervical spine region, the thoracic spine region, and the lumbar spine region are as follows: Obtain reference curvature values ​​of the cervical spine region, the thoracic spine region, and the lumbar spine region; Based on the spinal region data, the average curvature value, regional length value, and regional spinal vertical displacement value of the cervical spine region, thoracic spine region, and lumbar spine region were analyzed respectively; The reference curvature values, average curvature values, regional length values, and regional spinal vertical displacement values ​​of the cervical spine, thoracic spine, and lumbar spine regions were analyzed comprehensively to obtain the scoliosis index of the cervical spine, thoracic spine, and lumbar spine regions. The specific process of analyzing the average curvature values ​​of the cervical spine region, the thoracic spine region, and the lumbar spine region is as follows: calculating the curvature of each point on the regional center line of the cervical spine region, the thoracic spine region, and the lumbar spine region, averaging each curvature value in the cervical spine region, the thoracic spine region, and the lumbar spine region to obtain the average curvature values ​​of the cervical spine region, the thoracic spine region, and the lumbar spine region; The specific process of analyzing the regional length values ​​of the cervical spine region, the thoracic spine region, and the lumbar spine region is as follows: the length values ​​of the regional center lines in the cervical spine region, the thoracic spine region, and the lumbar spine region are used as the corresponding regional length values; The specific process of analyzing the regional spinal vertical displacement values ​​of the cervical spine region, the thoracic spine region, and the lumbar spine region is as follows: by analyzing the difference between the vertical coordinates of the highest point and the lowest point in the cervical spine region, the thoracic spine region, and the lumbar spine region, the regional spinal vertical displacement values ​​are obtained; The specific formula for calculating the scoliosis index of the cervical spine, thoracic spine, and lumbar spine is as follows: ； in, is the scoliosis index of the cervical spine area, is the average curvature of the cervical spine, is the reference curvature value of the cervical spine area, is the regional spinal vertical displacement value of the cervical spine area, is the length of the cervical spine area, is the scoliosis index of the thoracic spine region, is the average curvature of the thoracic region of the spine, is the reference curvature value of the thoracic spine region, is the regional spinal vertical displacement value of the thoracic spine area, is the length of the thoracic spine region, is the scoliosis index of the lumbar region of the spine, is the average curvature of the lumbar region of the spine, is the reference curvature value of the lumbar region of the spine, is the regional spinal vertical displacement value of the lumbar region of the spine, It is the length value of the lumbar region of the spine.

2. The image processing method for scoliosis detection according to claim 1, characterized in that: The spinal image data specifically includes the pixel value of each pixel point of the spinal column, and the spinal region data specifically includes the grayscale spinal pixel value of each spinal grayscale pixel point in the spinal image data.

3. The image processing method for scoliosis detection according to claim 1, characterized in that: The specific process of preprocessing the spinal image data of the patient to be detected is: performing denoising processing on the spinal image data of the patient to be detected, and performing contrast enhancement processing on the spinal image data of the patient to be detected after the denoising processing.

4. The image processing method for scoliosis detection according to claim 2, characterized in that: The specific process of obtaining the spinal region data is as follows: Read the spine image data of the patient to be tested after contrast enhancement processing and perform grayscale processing to obtain the grayscale pixel value of each pixel point containing the spine; Perform grayscale analysis on the grayscale pixel value of each pixel point containing the spine to obtain a grayscale histogram; The grayscale pixel value corresponding to the peak value in the grayscale histogram is used as the pixel judgment threshold; The grayscale pixel value of each pixel point including the spine is screened based on the pixel judgment threshold to obtain the grayscale spine pixel value of each spine grayscale pixel point in the spine image data.

5. The image processing method for scoliosis detection according to claim 4, characterized in that: The specific steps for obtaining the cervical spine region, the thoracic spine region, and the lumbar spine region are as follows: Connect each spine grayscale pixel point located at the boundary in the spine region data and regard it as the spine contour; Perform mean analysis on the abscissa values ​​of several grayscale pixel points of the spine located at the bottom of the spine contour to obtain the abscissa value of the center of the spine bottom; In the spine area data, a vertical line perpendicular to the horizontal coordinate value of the center of the spine bottom is drawn toward the top of the spine contour, and the vertical coordinate value of the spine grayscale pixel point where the top of the spine contour intersects with the vertical line is used as the vertical coordinate value of the center of the spine top; Read the ordinate value of the center of the spine bottom, and perform length analysis in combination with the ordinate value of the center of the spine top to obtain the total length of the spine; The vertical coordinate value of the center of the top of the spine and the total length of the spine are input into the regional division model for regional division to obtain the cervical spine region, thoracic spine region, and lumbar spine region.

6. The image processing method for scoliosis detection according to claim 5, characterized in that: The regional division model is as follows: ； in, is the ordinate of the end of the cervical region of the spine, is the vertical coordinate value of the center of the top of the spine, is the cervical spine area ratio coefficient stored in the database, is the total length of the spine, is the ordinate of the end of the thoracic region of the spine, is the thoracic vertebra area ratio stored in the database, is the ordinate of the end of the lumbar region of the spine, is the lumbar area ratio coefficient stored in the database, .

7. An image processing system for scoliosis detection, using the image processing method for scoliosis detection according to any one of claims 1 to 6, characterized in that: include: Data acquisition module, region extraction module, region division module, scoliosis analysis module, abnormality marking module; The data acquisition module is used to acquire the spinal image data of the patient to be detected and perform preprocessing; The region extraction module is used to perform region extraction on the spinal image data of the patient to be detected after preprocessing to obtain spinal region data; The region division module is used to divide the spinal column region data into regions to obtain the cervical spine region, the thoracic spine region, and the lumbar spine region; The scoliosis analysis module is used to analyze the data of the cervical spine region, the thoracic spine region, and the lumbar spine region respectively to obtain the scoliosis index of the cervical spine region, the thoracic spine region, and the lumbar spine region; The abnormal marking module is used to judge and analyze the scoliosis index of the cervical spine area, thoracic spine area, and lumbar spine area respectively with the preset corresponding area thresholds, mark the areas that do not meet the corresponding area thresholds, and analyze the degree of regional scoliosis.

Images