Osteoporosis screening system and method based on AI algorithm using X-rays
The osteoporosis screening system using AI algorithms automatically identifies osteoporosis symptoms and adjusts X-ray machine parameters in real time, solving the problem of image quality and parameter incompatibility with individual differences in traditional technologies, and achieving high-precision and efficient osteoporosis screening.
Patent Information
- Application Number
- CN202411179277.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-27
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-08-27
AI Technical Summary
In the early screening of osteoporosis, existing technologies are limited by image quality and resolution, making it difficult to accurately identify subtle changes. Traditional technologies rely on fixed imaging parameters, which are not suitable for patient groups with large individual differences, increasing the risk of misdiagnosis and the complexity of judgment.
An AI-based osteoporosis screening system is employed, comprising an image quality control and classification module, a bone feature analysis module, an imaging parameter optimization module, and a judgment and verification module. It utilizes deep belief networks and convolutional neural networks to automatically identify osteoporosis symptoms, adjust X-ray machine parameters in real time, and optimize image quality and feature extraction.
It improves the image analysis accuracy and automation level of osteoporosis screening, reduces human error, ensures the best image quality for each scan, and enhances the reliability and accuracy of screening.
Smart Images

Figure CN118691615B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of medical image processing technology, and in particular to an osteoporosis screening system and method using X-rays based on an AI algorithm. Background Art
[0002] Medical image processing is a field that uses computer technology to analyze and interpret medical image data, including images obtained from various imaging technologies such as X-rays, MRIs, and CT scans. The purpose of this field is to help doctors diagnose diseases more accurately by enhancing image quality and extracting key features in images. Medical image processing technology includes the application of machine learning and deep learning, which greatly improves the automation and accuracy of diagnosis, especially in parsing the complexity of images and extracting imperceptible details.
[0003] Among them, the osteoporosis screening system using X-rays uses X-ray imaging technology combined with image processing algorithms to identify and assess the risk of osteoporosis. Its main purpose is to detect osteoporosis in the early stages so that timely intervention and treatment can be carried out. By analyzing bone density and structural changes in X-ray images, the system can help doctors determine the patient's bone health status for prevention.
[0004] Existing technologies are immature in automation and identification of pathological features, especially in the early screening of osteoporosis. It is difficult to accurately identify subtle changes due to image quality and resolution limitations. Traditional technologies rely on relatively fixed imaging parameter settings, which leads to the loss of key screening information in patient groups with large individual differences. For example, fixed exposure and focal length are not suitable for the bone density and body shape of all patients, resulting in the loss of key information. Existing technologies require more manual intervention to adjust and interpret the results when processing images, which not only increases the complexity of the judgment process, but also increases the risk of misjudgment, affecting the efficiency and accuracy of screening. Summary of the invention
[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an osteoporosis screening system and method based on AI algorithm and X-ray film.
[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solution: an osteoporosis screening system using X-rays based on AI algorithm, the system comprising:
[0007] The image quality control classification module performs image quality assessment based on X-ray image data, screens images whose resolution meets the judgment criteria, performs graded processing on the images, and reclassifies them according to the image clarity index to obtain an image screening set;
[0008] The bone feature analysis module automatically identifies the features of osteoporosis symptoms based on the image screening set and uses a deep belief network, records the feature values of each image, and compares the features with known disease standards to obtain morphological feature data;
[0009] The imaging parameter optimization module adjusts the X-ray machine parameters, including exposure and focal length, based on the preliminary abnormality detection results, and uses a convolutional neural network to re-collect images to optimize the visibility of key details. It also compares the differences in image details before and after the adjustment to obtain imaging quality optimization information.
[0010] The judgment and verification module judges and analyzes the image based on the morphological feature data and imaging quality optimization information, calculates the medical indicator score associated with osteoporosis, and compares it with the medical history data to obtain the screened bone image data.
[0011] The present invention is improved in that the steps of obtaining the image screening set are specifically as follows:
[0012] Based on X-ray image data, the intensity and weight of the pixel points are used, using the formula:
[0013]
[0014] Calculate the clarity index of each X-ray image ,in, and represents the pixel position of the image, and Indicates the width and height of the image, It's location The pixel intensity, is the weight that is dynamically adjusted according to the pixel position, and is the weight adjustment coefficient;
[0015] Based on the clarity index, the sensitivity to image quality differences is optimized by comparison with a resolution standard, using the formula:
[0016]
[0017] Get the clarity ratio result ,in, is the clarity index, is the standard value of clarity, is the scaling factor adjusted according to the image characteristics;
[0018] Based on the clarity ratio result, determine whether the image meets the quality standard, using the judgment formula:
[0019]
[0020] Get image quality assessment results ,in, is the result of the clarity ratio, is the threshold value that determines whether the image quality is acceptable.
[0021] The present invention is improved in that the steps of automatically identifying the features are specifically as follows:
[0022] Based on the image filter set, a single image is loaded and optimized for feature recognizability using the formula:
[0023]
[0024] Get the image preprocessing results ,in, and represents the pixel position of the image, and Indicates the width and height of the image, It's location The pixel intensity, and is a processing parameter used to optimize the recognizability of image features;
[0025] Based on the image preprocessing results, a deep belief network is used for feature extraction using the formula:
[0026]
[0027] Obtain key features associated with osteoporosis ,in, is the image preprocessing result, It is a parameter that adjusts the sensitivity of feature extraction.
[0028] The present invention is improved in that the step of acquiring the morphological feature data is specifically as follows:
[0029] Based on the features output by the deep belief network, the osteoporosis symptoms are initially identified and the correlation with disease screening is optimized using the formula:
[0030]
[0031] Get the weighted feature set ,in, Representative Image No. eigenvalues, is the weight of each feature dimension, is the nonlinear adjustment coefficient of the optimization feature, is the total number of dimensions;
[0032] Based on the weighted feature set, the comparison with the disease morphology standard is carried out using the formula:
[0033]
[0034] Get morphological feature data ,in, It is Image No. weighted eigenvalues, It is The disease morphology standard threshold of each feature, It is The standard deviation of the features, is the total number of features.
[0035] The present invention is improved in that the steps of adjusting the X-ray machine parameters are specifically as follows:
[0036] Based on the preliminary anomaly detection results, the image exposure deviation is calculated using the formula:
[0037]
[0038] Get exposure bias assessment data ,in, and represents the pixel position of the image, and Indicates the width and height of the image, It's location The pixel intensity, is the total number of pixels, is the target brightness threshold;
[0039] Based on the exposure deviation evaluation data, the exposure of the X-ray machine is adjusted and the image quality is optimized, using the formula:
[0040]
[0041] Get the adjusted exposure settings ,in, is the original exposure, is the adjustment factor, is the exposure bias assessment data;
[0042] Based on the adjusted exposure settings, the focus of the X-ray machine is evaluated and adjusted, and if detail is lost, a focus correction is performed using the formula:
[0043]
[0044] Get a new focal length ,in, is the original focal length, and are the target and current level of detail, It is the responsiveness parameter for adjusting the focal length.
[0045] The present invention is improved in that the step of obtaining the imaging quality optimization information is specifically as follows:
[0046] Based on the adjusted exposure and focal length, the image is recaptured and analyzed using a convolutional neural network to optimize key details in the image using the formula:
[0047]
[0048] Get the image quality score adjusted by CNN ,in, is the adjusted image input data;
[0049] Based on the image quality score, the image details before and after adjustment are compared using the formula:
[0050]
[0051] Get the difference in image quality before and after adjustment ,in, is the adjusted image quality score, is the image quality score before adjustment.
[0052] The present invention is improved in that the calculation steps of the medical indicator score associated with osteoporosis are specifically as follows:
[0053] Based on the morphological feature data and imaging quality optimization information, bone density features are extracted, and the edge detection algorithm is used to calculate the edge strength of each area using the formula:
[0054]
[0055] Get the total edge strength index ,in, It is The edge strength value of the image region, is the weight factor of edge strength, Represents the number of image regions involved in calculating the total edge strength;
[0056] Based on the total edge strength index, the image is analyzed to evaluate the consistency of the texture using the formula:
[0057]
[0058] Get texture consistency index ,in, Representative The texture consistency of the region is the weight factor for texture consistency, Represents the number of image regions involved in calculating the overall texture consistency;
[0059] Based on the total edge strength index and texture consistency index, the formula is adopted:
[0060]
[0061] Calculating osteoporosis medical index scores ,in, and is the weight coefficient, is the overall edge strength index, It is an indicator of texture consistency.
[0062] The present invention is improved in that the steps of obtaining the bone image data after screening are specifically as follows:
[0063] Compare the calculated scores with historical medical data and classify the image data using the formula:
[0064]
[0065] Get the screened bone image dataset ,in, is the risk threshold, It is the medical indicator score.
[0066] The osteoporosis screening method based on AI algorithm using X-ray film is performed based on the above-mentioned osteoporosis screening system based on AI algorithm using X-ray film, and comprises the following steps:
[0067] S1: Based on X-ray image data, the resolution and clarity index of each image are evaluated, images that meet the conditions are screened, the images are reclassified and clarity grades are set, and an image screening set is obtained;
[0068] S2: Based on the image screening set, using a deep belief network, analyzing the osteoporosis symptom characteristics in each image, recording the feature value of each image, and comparing the feature value with a known disease standard to generate morphological feature data;
[0069] S3: Based on the preliminary abnormality detection results, adjust the exposure and focus of the X-ray machine, use the convolutional neural network to re-capture the image, optimize the display of key details in the image, compare the images before and after the adjustment, and generate imaging quality optimization information;
[0070] S4: calculating a medical indicator score associated with osteoporosis based on the imaging quality optimization information and the morphological feature data, comparing the score with the medical history data, and generating a preliminary medical evaluation result;
[0071] S5: Using the preliminary medical assessment results, perform risk assessment on the screened images, perform risk grading on the patients according to the assessment results, and generate a risk grading log.
[0072] Compared with the prior art, the advantages and positive effects of the present invention are:
[0073] In the present invention, through the application of deep belief network and convolutional neural network, the image analysis accuracy and automation level of osteoporosis screening are improved, and the key features of osteoporosis, such as bone density and microstructural changes, can be accurately and automatically identified. Automated feature recognition reduces dependence on professional knowledge, reduces human errors in the screening process, and improves the sensitivity of lesion detection. AI-driven real-time adjustment of imaging parameters ensures that the best image quality can be obtained in each scan, thereby improving the reliability of screening. BRIEF DESCRIPTION OF THE DRAWINGS
[0074] Figure 1 The module diagram of the osteoporosis screening system using X-rays based on AI algorithm proposed by the present invention;
[0075] Figure 2 This is a flowchart for obtaining an image screening set in the present invention;
[0076] Figure 3 A flow chart for automatically identifying features in the present invention;
[0077] Figure 4 This is a flow chart for obtaining morphological feature data in the present invention;
[0078] Figure 5 is a flow chart for adjusting the parameters of the X-ray machine in the present invention;
[0079] Figure 6 A flowchart for obtaining imaging quality optimization information in the present invention;
[0080] Figure 7 is a flow chart for calculating the medical indicator score associated with osteoporosis in the present invention;
[0081] Figure 8 This is a flow chart for obtaining bone image data after screening in the present invention. DETAILED DESCRIPTION
[0082] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.
[0083] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.
[0084] Example
[0085] See also Figure 1 The present invention provides a technical solution: an osteoporosis screening system using X-rays based on an AI algorithm includes:
[0086] The image quality control classification module performs image quality assessment based on X-ray image data, screens images whose resolution meets the judgment criteria, performs graded processing on the images, and reclassifies them according to the image clarity index to obtain an image screening set;
[0087] The bone feature analysis module uses a deep belief network based on the image screening set to automatically identify the characteristics of osteoporosis symptoms, record the feature values of each image, and compare the features with known disease standards to obtain morphological feature data;
[0088] The imaging parameter optimization module adjusts the X-ray machine parameters, including exposure and focal length, based on the preliminary abnormality detection results, and uses a convolutional neural network to re-collect images to optimize the visibility of key details. It also compares the differences in image details before and after the adjustment to obtain imaging quality optimization information.
[0089] The judgment and verification module judges and analyzes the image based on the morphological feature data and imaging quality optimization information, calculates the medical indicator scores associated with osteoporosis, and compares them with the medical history data to obtain the bone image data after screening.
[0090] The image screening set includes images that meet the conditions, image grading information, and reclassified image sets. The morphological feature data includes feature recognition patterns, pathological feature comparisons, and automatic recognition feature sets. The imaging quality optimization information includes adjusted detail resolution, image clarity enhancement items, and imaging parameter adjustment details. The bone image data after screening includes medical scores and data comparison and analysis results.
[0091] See also Figure 2 , the steps to obtain the image screening set are as follows:
[0092] Based on X-ray image data, the intensity and weight of the pixel points are used, using the formula:
[0093]
[0094] Calculate the clarity index of each X-ray image ,in, and represents the pixel position of the image, and Indicates the width and height of the image, It's location The pixel intensity, is the weight that is dynamically adjusted according to the pixel position, and is the weight adjustment coefficient;
[0095] Based on the clarity index, the sensitivity to image quality differences is optimized by comparison with a resolution standard, using the formula:
[0096]
[0097] Get the clarity ratio result ,in, is the clarity index, is the standard value of clarity, is the scaling factor adjusted according to the image characteristics;
[0098] Based on the clarity ratio results, determine whether the image meets the quality standards using the judgment formula:
[0099]
[0100] Get image quality assessment results ,in, is the result of the clarity ratio, is the threshold value that determines whether the image quality is acceptable.
[0101] Assume a small image, whose size is 2x2 pixels;
[0102] The intensity of each pixel for:
[0103]
[0104]
[0105]
[0106]
[0107] Weight Assume that:
[0108]
[0109]
[0110]
[0111]
[0112] Calculation process:
[0113]
[0114]
[0115] The calculated sharpness index is 72.5, which reflects the overall sharpness of the image at a given weight and pixel intensity.
[0116] use , assuming is 100, is 1.1;
[0117] Calculation process:
[0118]
[0119] The calculated ratio It is 0.659, which means that the image clarity is about 66% lower than the standard value, which means that the image does not meet the quality requirements.
[0120] Assumptions is 0.7, use is compared with 0.659. Since 0.659 is less than 0.7, 0 means the image is unqualified
[0121] See also Figure 3 , the steps for automatic feature recognition are as follows:
[0122] Based on the image filter set, load a single image and optimize the feature identifiability using the formula:
[0123]
[0124] Get the image preprocessing results ,in, and represents the pixel position of the image, and Indicates the width and height of the image, It's location The pixel intensity, and is a processing parameter used to optimize the recognizability of image features;
[0125] Based on the image preprocessing results, the deep belief network is used for feature extraction, using the formula:
[0126]
[0127] Obtain key features associated with osteoporosis ,in, is the image preprocessing result, It is a parameter that adjusts the sensitivity of feature extraction.
[0128] Suppose there is a 2x2 pixel image with the following pixel values:
[0129]
[0130]
[0131]
[0132]
[0133] Select and Perform the calculation:
[0134]
[0135]
[0136]
[0137] The value is 6380, which reflects the overall feature strength of the image after preprocessing enhancement, indicating that the image has a strong feature response under the selected parameters and is suitable for further feature extraction.
[0138] Use the above , assuming :
[0139]
[0140]
[0141]
[0142] The value is close to 1, indicating that under the selected parameters, the image features are very significant and have been highly responded to by the model, indicating that the image contains important features related to osteoporosis and is suitable for further analysis and screening.
[0143] See also Figure 4 , the specific steps for obtaining morphological feature data are:
[0144] Based on the features output by the deep belief network, the symptoms of osteoporosis are initially identified and the correlation with disease screening is optimized using the formula:
[0145]
[0146] Get the weighted feature set ,in, Representative Image No. eigenvalues, is the weight of each feature dimension, is the nonlinear adjustment coefficient of the optimization feature, is the total number of dimensions;
[0147] Based on the weighted feature set, the comparison is made with the disease morphology standard using the formula:
[0148]
[0149] Get morphological feature data ,in, It is Image No. weighted eigenvalues, It is The disease morphology standard threshold of each feature, It is The standard deviation of the features, is the total number of features.
[0150] Assume there are three feature dimensions ( ):
[0151] Eigenvalues:
[0152] , , ;
[0153] Weight:
[0154] ;
[0155] Nonlinear coefficient
[0156] ;
[0157] Calculation process:
[0158]
[0159]
[0160]
[0161] The value is 1.37, which is the result of weighting and nonlinear adjustment for comparison with the disease standard threshold, with higher values indicating a high degree of disease association.
[0162] Assume the following data:
[0163] ;
[0164] ;
[0165] ;
[0166] ;
[0167] ;
[0168] .
[0169] Calculation process:
[0170] Assumptions :
[0171]
[0172]
[0173]
[0174] What you get When it is -1, it means that the feature set of the image does not match the disease standard threshold after comparison, indicating that the image does not have typical disease features.
[0175] See also Figure 5 , the specific steps for adjusting the X-ray machine parameters are:
[0176] Based on the preliminary anomaly detection results, the image exposure deviation is calculated using the formula:
[0177]
[0178] Get exposure bias assessment data ,in, and represents the pixel position of the image, and Indicates the width and height of the image, It's location The pixel intensity, is the total number of pixels, is the target brightness threshold;
[0179] Based on the exposure deviation evaluation data, the exposure of the X-ray machine is adjusted and the image quality is optimized using the formula:
[0180]
[0181] Get the adjusted exposure settings ,in, is the original exposure, is the adjustment factor, is the exposure bias assessment data;
[0182] Based on the adjusted exposure setting, the focus of the X-ray machine is evaluated and adjusted. If detail is lost, focus correction is performed using the formula:
[0183]
[0184] Get a new focal length ,in, is the original focal length, and are the target and current level of detail, It is the responsiveness parameter for adjusting the focal length.
[0185] Assume the image size is 100x100 pixels, that is ,The average brightness of the image was measured to be 180, which is lower than the ideal brightness;
[0186] Assume it is 200.
[0187] Calculation process:
[0188]
[0189]
[0190]
[0191] It means that the current image brightness is 20 units lower than the ideal value, and the exposure needs to be increased to compensate for the difference; 1800000 is the sum of the brightness values of all pixels.
[0192] Assume it is 500;
[0193] Set to 0.1, indicating that each unit of exposure deviation adjusts the exposure by 10%;
[0194] .
[0195] Calculation process:
[0196]
[0197]
[0198]
[0199] This means that in order to achieve the ideal brightness, the original exposure setting needs to be fine-tuned by 2 units.
[0200] Assume it is 50mm;
[0201] Set to 0.05, which means that the focal length is adjusted by 5% for each unit detail difference;
[0202] Assume 100%;
[0203] Assume 95%.
[0204] Calculation process:
[0205]
[0206]
[0207]
[0208]
[0209] mm means that in order to increase detail clarity, the original focal length needs to be fine-tuned to increase by 0.25mm.
[0210] See also Figure 6 , the specific steps for obtaining the imaging quality optimization information are:
[0211] Based on the adjusted exposure and focus, the image is recaptured and analyzed using a convolutional neural network to optimize key details in the image using the formula:
[0212]
[0213] Get the image quality score adjusted by CNN ,in, is the adjusted image input data;
[0214] Based on the image quality score, the image details before and after adjustment are compared using the formula:
[0215]
[0216] Get the difference in image quality before and after adjustment ,in, is the adjusted image quality score, is the image quality score before adjustment.
[0217] Assuming the adjusted graph There is improved contrast and details. The CNN model is trained to identify and enhance these details, outputting an image quality score, for example, from 70 to 85, reflecting the enhancement of key details;
[0218] Assume that the resized image is input to CNN, CNN evaluates the image and outputs a quality score of 70 increased to 85;
[0219] The adjusted images show higher quality scores in the CNN evaluation, indicating that the visibility of key details has been significantly improved.
[0220] If the image quality score before adjustment It is 70, the adjusted score It's 85.
[0221] Calculate the difference between two ratings:
[0222]
[0223] The 15 shows an improvement of 15 points in the image quality score, indicating a significant improvement in the visibility of key details.
[0224] See also Figure 7 , the calculation steps of the medical indicator score associated with osteoporosis are as follows:
[0225] Based on the morphological feature data and imaging quality optimization information, bone density features are extracted, and the edge detection algorithm is used to calculate the edge strength of each area using the formula:
[0226]
[0227] Get the total edge strength index ,in, It is The edge strength value of the image region, is the weight factor of edge strength, Represents the number of image regions involved in calculating the total edge strength;
[0228] Based on the total edge strength index, the image is analyzed to evaluate the consistency of the texture using the formula:
[0229]
[0230] Get texture consistency index ,in, Representative The texture consistency of the region is the weight factor for texture consistency, Represents the number of image regions involved in calculating the overall texture consistency;
[0231] Based on the total edge strength index and texture consistency index, the formula is adopted:
[0232]
[0233] Calculating osteoporosis medical index scores ,in, and is the weight coefficient, is the overall edge strength index, It is an indicator of texture consistency.
[0234] Suppose there are three regions with edge strengths of 10, 15, and 20;
[0235] Assume that the weight factor of each edge is is 0.5.
[0236] calculate:
[0237]
[0238] This indicates that the total weighted edge strength is 22.5, which will be used for further medical score calculation.
[0239] Suppose there are three regions with texture consistency of 5, 10, and 15 respectively;
[0240] Assume that the weight factor for each texture is is 0.3.
[0241] calculate:
[0242]
[0243] This indicates that the total weighted texture consistency is 9.
[0244] Assumptions and ;
[0245] calculate:
[0246]
[0247] A score of 40.5 reflects the patient's risk of osteoporosis.
[0248] See also Figure 8 , the steps for obtaining the bone image data after screening are as follows:
[0249] Compare the calculated scores with historical medical data and classify the image data using the formula:
[0250]
[0251] Get the screened bone image dataset ,in, is the risk threshold, It is the medical indicator score.
[0252] Assumptions The score is 0.924, is 0.8, use the formula to judge, 0.924 is greater than 0.8, then The corresponding X-ray film is classified as high risk.
[0253] The osteoporosis screening method using X-rays based on AI algorithms includes the following steps:
[0254] S1: Based on X-ray image data, the resolution and clarity index of each image are evaluated, images that meet the conditions are screened, the images are reclassified and clarity grades are set, and an image screening set is obtained;
[0255] S2: Based on the image screening set, a deep belief network is used to analyze the osteoporosis symptom characteristics in each image, record the feature value of each image, and compare the feature value with the known disease standard to generate morphological feature data;
[0256] S3: Based on the preliminary abnormality detection results, adjust the exposure and focus of the X-ray machine, use the convolutional neural network to re-capture the image, optimize the display of key details in the image, compare the images before and after the adjustment, and generate imaging quality optimization information;
[0257] S4: Based on the imaging quality optimization information and morphological feature data, calculate the medical index score associated with osteoporosis, compare the score with the medical history data, and generate a preliminary medical assessment result;
[0258] S5: Using the preliminary medical assessment results, conduct risk assessment on the screened images, risk grade the patients according to the assessment results, and generate a risk grade log.
[0259] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.
Claims
1. Osteoporosis screening system based on AI algorithm using X-rays, It is characterized in that The system comprises: The image quality control classification module performs image quality assessment based on X-ray image data, screens images whose resolution meets the judgment criteria, performs graded processing on the images, and reclassifies them according to the image clarity index to obtain an image screening set; The bone feature analysis module automatically identifies the features of osteoporosis symptoms based on the image screening set and uses a deep belief network, records the feature values of each image, and compares the features with known disease standards to obtain morphological feature data; The imaging parameter optimization module adjusts the X-ray machine parameters, including exposure and focal length, based on the preliminary abnormality detection results, and uses a convolutional neural network to re-collect images to optimize the visibility of key details. It also compares the differences in image details before and after the adjustment to obtain imaging quality optimization information. The judgment and verification module judges and analyzes the image based on the morphological feature data and the imaging quality optimization information, calculates the medical index score associated with osteoporosis, and compares it with the medical history data to obtain the bone image data after screening; The steps for obtaining the morphological feature data are specifically as follows: Based on the features output by the deep belief network, the osteoporosis symptoms are initially identified and the correlation with disease screening is optimized using the formula: Get the weighted feature set ,in, Representative Image No. eigenvalues, is the weight of each feature dimension, is the nonlinear adjustment coefficient of the optimization feature, is the total number of dimensions; Based on the weighted feature set, the comparison with the disease morphology standard is carried out using the formula: Get morphological feature data ,in, It is Image No. weighted eigenvalues, It is The disease morphology standard threshold of each feature, It is The standard deviation of the features, is the total number of features; The adjustment steps of the X-ray machine parameters are specifically as follows: Based on the preliminary anomaly detection results, the image exposure deviation is calculated using the formula: Get exposure bias assessment data ,in, and represents the pixel position of the image, and Indicates the width and height of the image, It's location The pixel intensity, is the total number of pixels, is the target brightness threshold; Based on the exposure deviation evaluation data, the exposure of the X-ray machine is adjusted and the image quality is optimized, using the formula: Get the adjusted exposure settings ,in, is the original exposure, is the adjustment factor, is the exposure bias assessment data; Based on the adjusted exposure settings, the focus of the X-ray machine is evaluated and adjusted, and if detail is lost, a focus correction is performed using the formula: Get a new focal length ,in, is the original focal length, and are the target and current level of detail, is the responsiveness parameter for adjusting the focal length; The steps for obtaining the imaging quality optimization information are specifically as follows: Based on the adjusted exposure and focal length, the image is recaptured and analyzed using a convolutional neural network to optimize key details in the image using the formula: Get the image quality score adjusted by CNN ,in, is the adjusted image input data; Based on the image quality score, the image details before and after adjustment are compared using the formula: Get the difference in image quality before and after adjustment ,in, is the adjusted image quality score, is the image quality score before adjustment; The calculation steps of the medical indicator score associated with osteoporosis are specifically as follows: Based on the morphological feature data and imaging quality optimization information, bone density features are extracted, and the edge detection algorithm is used to calculate the edge strength of each area using the formula: Get the total edge strength index ,in, It is The edge strength value of the image region, is the weight factor of edge strength, Represents the number of image regions involved in calculating the total edge strength; Based on the total edge strength index, the image is analyzed to evaluate the consistency of the texture using the formula: Get texture consistency index ,in, Representative The texture consistency of the region is the weight factor for texture consistency, Represents the number of image regions involved in calculating the overall texture consistency; Based on the total edge strength index and texture consistency index, the formula is adopted: Calculating the osteoporosis medical index score ,in, and is the weight coefficient, is the overall edge strength index, It is an indicator of texture consistency.
2. The osteoporosis screening system using X-rays based on AI algorithm according to claim 1, It is characterized in that The steps of obtaining the image screening set are specifically as follows: Based on X-ray image data, the intensity and weight of the pixel points are used, using the formula: Calculate the clarity index of each X-ray image ,in, and represents the pixel position of the image, and Indicates the width and height of the image, It's location The pixel intensity, is the weight dynamically adjusted according to the pixel position, and is the weight adjustment coefficient; Based on the clarity index, the sensitivity to image quality differences is optimized by comparison with a resolution standard, using the formula: Get the clarity ratio result ,in, is the clarity index, is the standard value of clarity, is the scaling factor adjusted according to the image characteristics; Based on the clarity ratio result, determine whether the image meets the quality standard, using the judgment formula: Get image quality assessment results ,in, is the clarity ratio result, is the threshold value that determines whether the image quality is acceptable.
3. The osteoporosis screening system based on AI algorithm and X-ray according to claim 1, It is characterized in that The steps of automatically identifying the features are specifically as follows: Based on the image filter set, a single image is loaded and optimized for feature recognizability using the formula: Get the image preprocessing results ,in, and represents the pixel position of the image, and Indicates the width and height of the image, It's location The pixel intensity, and is a processing parameter used to optimize the recognizability of image features; Based on the image preprocessing results, a deep belief network is used for feature extraction using the formula: Obtain key features associated with osteoporosis ,in, is the image preprocessing result, It is a parameter that adjusts the sensitivity of feature extraction.
4. The osteoporosis screening system based on AI algorithm and using X-rays according to claim 1, It is characterized in that The steps for obtaining the bone image data after screening are specifically as follows: Compare the calculated scores with historical medical data and classify the image data using the formula: Get the screened bone image dataset ,in, is the risk threshold, It is the medical indicator score.
5. Osteoporosis screening method based on AI algorithm using X-rays, It is characterized in that The osteoporosis screening system based on the AI algorithm using X-rays according to any one of claims 1 to 4 is implemented, comprising the following steps: Based on the X-ray image data, the resolution and clarity index of each image are evaluated, the images meeting the conditions are screened, the images are reclassified and the clarity grades are set, and the image screening set is obtained; Based on the image screening set, using a deep belief network, analyzing the osteoporosis symptom characteristics in each image, recording the characteristic value of each image, and comparing the characteristic value with a known disease standard to generate morphological characteristic data; According to the preliminary abnormality detection results, the exposure and focus of the X-ray machine are adjusted, and the convolutional neural network is used to re-capture the image to optimize the display of key details in the image. The images before and after the adjustment are compared to generate imaging quality optimization information; Calculating a medical indicator score associated with osteoporosis based on the imaging quality optimization information and the morphological feature data, comparing the score with the medical history data, and generating a preliminary medical assessment result; The preliminary medical assessment results are used to perform risk assessment on the screened images, and the patients are risk graded according to the assessment results to generate a risk graded log.
