Image processing and analysis system and method based on image

Abstract

The invention discloses an image processing and analysis system and method based on images, and relates to the field of image radiology technology. The system comprises a data aggregation module, an inward image detection module, an outward image detection module, a comprehensive scoring module and a report feedback module. The technical key points are as follows: by screening out all trainees whose evaluation indicators are less than a set threshold, and then summarizing their abnormal valuations and evaluation indicators to form a scoring data set, and building a scoring calculation model to obtain a comprehensive score for each trainee, and sorting them from large to small according to the score value to generate a sorting table, and by comprehensively considering the abnormal valuations and evaluation indicators, the physical fitness of the trainees is more comprehensively reflected; at the same time, through scoring and sorting, the physical fitness differences between the trainees are intuitively displayed, which provides a scientific basis for selecting excellent trainees, ensures the fairness and accuracy of the selection process, and improves the efficiency and quality of personnel selection.

Description

Technical Field

[0001] The present invention relates to the field of imaging radiology technology, and in particular to an image processing and analysis system and method based on imaging. Background Art

[0002] Radiography, also known as medical radiography, is a type of imaging examination that mainly uses X-rays to examine the human body. This examination method can observe the morphological structure of human organs and judge their functional status without changing or destroying the integrity of the body. Common radiography examinations include X-rays, fluoroscopy, gastrointestinal barium meals, bone density measurements, etc. With the advancement of medical science and technology, radiography technology is also constantly developing, such as digital X-ray photography technology, computer tomography (CT), magnetic resonance imaging (MRI), etc. These technologies provide doctors with clearer and more accurate diagnostic basis. Radiography is of great significance in medical diagnosis, but care should be taken to take protective measures when obtaining images to ensure the safety of patients.

[0003] The existing application publication number is CN116402816A, and the solution pointed out in the document named A method and system for managing physical examination CT image data includes: obtaining each pixel point of the lung CT image of the object to be analyzed as the center point, and determining the optimal window size according to the gradient angle of the pixel points in each initial window corresponding to each pixel point; determining the mutation pixel point according to the grayscale distribution of the pixel points in the optimal window corresponding to each pixel point, and then obtaining each connected domain; obtaining the corresponding membership degree according to the grayscale distribution of the pixel points in the optimal window corresponding to each pixel point in each connected domain, determining the Gaussian weight corresponding to each connected domain, and then performing Gaussian filtering on each connected domain in the lung CT image to obtain the lung CT image after filtering; the present invention has a better filtering effect on the lung CT image of the object to be analyzed, and can effectively prevent the loss of important features in the lung CT image of the object to be analyzed. It only processes the image and ensures the accuracy of the analysis result, but cannot analyze the physical abnormalities of different types of users based on the image.

[0004] Combining the above documents and existing technologies, traditional technologies have many problems; first, in terms of abnormal area identification and quantitative analysis based on the user's image data, traditional methods often lack accuracy and reliability, which may lead to analysis errors; for example, a single image modality analysis may not be able to accurately capture complex abnormal areas, such as early tumors; secondly, scar assessment has traditionally relied mainly on the doctor's subjective judgment, lacking objective quantitative standards, and not fully considering key constraints such as physical examination data and vision, which may lead to inconsistency and unfairness in the assessment results; for students who are working in civil aviation, traditional methods often rely on only a single indicator to assess their physical fitness, while ignoring other important factors, such as the condition of abnormal areas in body images and scar conditions, which cannot fully reflect the physical fitness and potential of the students, affecting the fairness and accuracy of subsequent selection. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] In view of the deficiencies in the prior art, the present invention provides an image processing and analysis system and method based on images. The solution running in the system can effectively solve the problems raised in the background technology.

[0007] (II) Technical solution

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0009] An image processing and analysis system based on images, comprising:

[0010] The data aggregation module obtains the real-time image data and clinical information of the target trainees; the real-time image data at least includes CT scan images and X-rays; the clinical information at least includes name, age and physical examination data; and simultaneously uses face recognition technology to authenticate the identity of the target trainees;

[0011] When the verification result is correct, no response action is taken;

[0012] When the verification result is wrong, the information proofreading action is performed;

[0013] The inward image detection module uses a deep learning model to analyze CT scan images, identify and mark preliminarily confirmed abnormal areas; it combines transfer learning technology for pre-training and fine-tunes the calibrated real-time image data;

[0014] Use multimodal image fusion technology to combine CT scan images and X-rays for analysis;

[0015] Determine the final abnormal area based on the analysis results;

[0016] Use deep learning algorithms to extract features and conduct quantitative analysis on the final abnormal area, build a quantitative analysis model, and simultaneously trigger the student judgment mechanism to generate abnormal valuation based on the target student type;

[0017] Compare the abnormality estimate with the preset acceptance threshold and decide whether to perform treatment intervention based on the comparison result;

[0018] The outward image detection module obtains the external surface image of the target student's body, and performs a quantitative assessment of the scars when scars are identified. It combines the physical examination data results and takes vision as a constraint to generate an evaluation index after comprehensive assessment. The evaluation index is compared with the set threshold, and corresponding system guidance is given based on the comparison results.

[0019] The comprehensive scoring module screens all target students who meet the evaluation index < the set threshold condition, and summarizes the final abnormal valuation and evaluation index of all target students to form a scoring data set, builds a scoring calculation model, obtains the comprehensive score value corresponding to each target student, and generates a ranking table Q;

[0020] The report feedback module summarizes the real-time imaging data and clinical information of each target student that has been correctly authenticated, the comparison results of the abnormal valuation and the acceptance threshold, the comparison results of the evaluation index and the set threshold, and the sorting table Q, generates report data, and transmits it in encrypted form and stores it in the analysis database.

[0021] Furthermore, the physical examination data in the clinical information includes at least: blood pressure, heart rate, and vision;

[0022] The information verification actions performed are as follows:

[0023] Automatically detect inconsistencies: compare the identity information obtained by face recognition with the records in the clinical information, and detect inconsistent fields, including at least misspellings of names, age mismatches, or differences in physical examination data; automatically correct information: if the name is misspelled, correct the displayed information based on the face recognition results; if there is an age mismatch or physical examination data difference, mark and temporarily store the correction suggestions; ensure verification correctness: after correction, perform a second information consistency check. If there are still inconsistent fields, trigger the manual review process, and retain the correction records and review tracks.

[0024] Furthermore, the process of obtaining a preliminary confirmed abnormal area is as follows:

[0025] Model selection: training convolutional neural network models;

[0026] Feature extraction: Using the model to extract key features in CT scan images, including lung nodules and bone abnormalities;

[0027] Abnormality recognition: Based on the extracted key features, abnormal areas in CT scan images are identified and initially classified.

[0028] Furthermore, the process of combining CT scan images and X-rays for analysis is as follows:

[0029] Register and align X-rays and CT scans;

[0030] Feature extraction is performed using the overall structural information of X-rays and the detailed cross-sectional information of CT scans;

[0031] The feature splicing method is used to fuse the extracted features and observe and analyze the abnormal area from several angles;

[0032] By comparing the fusion results with the analysis results of single image data;

[0033] If the analysis results are consistent, the initially confirmed abnormal area will be used as the final abnormal area;

[0034] If the analysis results are inconsistent, the fusion results shall prevail to determine the final abnormal area.

[0035] Furthermore, the features of the final abnormal region are extracted, including size and density;

[0036] When constructing a quantitative analysis model, the formula used is as follows:

[0037] ;

[0038] Where Av represents the final anomaly estimate, Both represent weight coefficients, and their value ranges are [0, 1]. X represents the size of the abnormal area, and D represents the density of the abnormal area, that is, the CT value.

[0039] Furthermore, the content that triggers the student judgment mechanism is as follows:

[0040] If the target student is a theoretical student, then output the current abnormal valuation;

[0041] If the target student is a non-theoretical student, the judgment adjustment strategy is executed, and under the condition that the abnormal area is determined to be a bone abnormality, the flight experience and age are used as correction data to generate a corrected abnormality estimate;

[0042] The formula used to generate the corrected anomaly estimate is as follows:

[0043] ;

[0044] ;

[0045] In the formula, represents the initial current abnormal estimation, α and β are correction coefficients, and their value range is [0, +∞), E represents flight experience, E_ref represents reference flight experience value, and A represents age.

[0046] Furthermore, the quantitative assessment of scars is based on the following methods:

[0047] ;

[0048] Where Sm represents the scar score obtained by quantitative evaluation of scars, ms represents the area of ​​scars, Lr_max represents the maximum length of the current scar, Bs represents the number of scars on the target student, w1, w2, a1, a2 are all weight coefficients, and the value range is [0, 1], i represents the number of the corresponding scar, i=1, 2, ..., n, n=Bs;

[0049] The method for generating evaluation indicators after comprehensive evaluation is as follows:

[0050] Taking vision as a constraint, when the vision of any eye of the target student reaches 0.7 or above, a comprehensive evaluation calculation is performed. When the vision of both eyes of the target student does not reach 0.7, the evaluation index is set to the set threshold;

[0051] The formula for calculating the evaluation index is as follows:

[0052] ;

[0053] Where, I represents the evaluation index, q1 and q2 are adjustment coefficients with a value range of [0, +∞), bp represents blood pressure, bp_normal represents the standard value of blood pressure, bp_range represents the range of blood pressure variation, hr represents heart rate, hr_normal represents the standard value of heart rate, hr_range represents the range of heart rate variation, Vl represents the visual acuity of the left eye, and Vr represents the visual acuity of the right eye;

[0054] The evaluation index is compared with the set threshold, and the comparison results are as follows:

[0055] When the evaluation index is ≥ the set threshold, the system guidance given is as follows:

[0056] It is recommended that target trainees consider careers other than aircrew;

[0057] When the evaluation index is less than the set threshold, the system guidance given is as follows:

[0058] It is recommended that the target trainees randomly select an aircrew or non-aircrew occupation.

[0059] Furthermore, the process of building the evaluation calculation model is as follows:

[0060] For the final outlier valuation:

[0061] ;

[0062] For evaluation metrics:

[0063] ;

[0064] Where S_Av and S_I represent the final anomaly estimation and the score of the evaluation index, respectively; Av_max and Av_min represent the maximum and minimum values ​​of the final anomaly estimation among all target students, respectively; I_max and I_min represent the maximum and minimum values ​​of the evaluation index among all target students, respectively;

[0065] The formula for calculating the comprehensive evaluation value is:

[0066] ;

[0067] In the formula, S represents the comprehensive score;

[0068] When generating the ranking table Q, the comprehensive score values ​​are sorted in order from large to small.

[0069] An image processing and analysis method based on image includes the following steps:

[0070] S1. Acquire the real-time image data and clinical information of the target student; wherein the real-time image data at least includes CT scan images and X-rays; the clinical information at least includes name, age and physical examination data; and simultaneously use face recognition technology to authenticate the identity of the target student;

[0071] When the verification result is correct, no response action is taken;

[0072] When the verification result is wrong, the information proofreading action is performed;

[0073] S2. Use deep learning models to analyze CT scan images, identify and mark preliminarily confirmed abnormal areas; combine transfer learning technology for pre-training, and fine-tune the calibrated real-time image data;

[0074] Use multimodal image fusion technology to combine CT scan images and X-rays for analysis;

[0075] Determine the final abnormal area based on the analysis results;

[0076] Use deep learning algorithms to extract features and conduct quantitative analysis on the final abnormal area, build a quantitative analysis model, and simultaneously trigger the student judgment mechanism to generate abnormal valuation based on the target student type;

[0077] Compare the abnormality estimate with the preset acceptance threshold and decide whether to perform treatment intervention based on the comparison result;

[0078] S3. Obtain the outer surface image of the target student's body, and if scars are identified, perform a quantitative assessment of the scars, combine the physical examination data results, and use vision as a constraint condition to generate an assessment index after comprehensive assessment; compare the assessment index with the set threshold, and give corresponding system guidance based on the comparison results;

[0079] S4. Screen all target students who meet the condition that the evaluation index is less than the set threshold, and summarize the final abnormal valuation and evaluation index of all target students to form a scoring data set, build a scoring calculation model, obtain the comprehensive score value corresponding to each target student, and generate a ranking table Q;

[0080] S5. Summarize the real-time image data and clinical information of each target student that has been correctly authenticated, the comparison results of the abnormal valuation and the acceptance threshold, the comparison results of the evaluation index and the set threshold, and the sorting table Q, generate report data, and encrypt and transmit them and store them in the analysis database.

[0081] (III) Beneficial effects

[0082] The present invention provides an image processing and analysis system and method based on images, which has the following beneficial effects:

[0083] (1) This solution achieves efficient and accurate identification and quantitative analysis of abnormal areas in the real-time image data of target trainees. It uses multimodal image fusion technology to combine CT scan images and X-ray films for analysis to ensure the accuracy of abnormal areas, extract features and quantitatively analyze abnormal areas, build a quantitative analysis model, and generate abnormal valuations that reflect the degree of impact. For non-theoretical trainees, the correction of abnormal valuations based on flight experience and age is also considered to make the results more in line with actual conditions. The overall solution improves the accuracy and reliability of abnormality detection, and also realizes automatic judgment of the degree of abnormality and intervention suggestions, providing a scientific basis for trainees' health management and treatment decisions.

[0084] (2) This scheme realizes the accurate identification and quantitative evaluation of scars on the outer surface of the target trainees’ bodies, and combines the physical examination data and vision constraints to comprehensively generate evaluation indicators, which can effectively identify scar characteristics, quantify scar scores, and consider vision as a key constraint. When vision meets the standard, the scar score is combined with other physical examination data to calculate the evaluation index. This scheme improves the objectivity and accuracy of scar evaluation. By comparing the evaluation index with the set threshold, the system can give targeted career choice suggestions, which not only guarantees the occupational safety requirements of air crew members, but also provides a scientific career planning reference for the target trainees.

[0085] (3) This scheme screens out all students whose evaluation indicators are less than the set threshold, then summarizes their abnormal valuations and evaluation indicators to form a scoring data set, and builds a scoring calculation model to obtain a comprehensive score for each student. The scores are sorted from large to small according to the score value to generate a ranking table. This solves the problem of incomplete evaluation of a single indicator and lack of comprehensive ranking in traditional evaluation. By comprehensively considering abnormal valuations and evaluation indicators, the physical fitness of students is more comprehensively reflected. At the same time, through scoring and ranking, the physical fitness differences between students are intuitively displayed, which provides a scientific basis for selecting outstanding students, ensures the fairness and accuracy of the selection process, and improves the efficiency and quality of personnel selection. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 It is a schematic diagram of the modular structure of the system in the present invention;

[0087] Figure 2 Schematic diagram of the overall process steps of the method of the present invention. DETAILED DESCRIPTION

[0088] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0089] Embodiment 1:

[0090] See also Figure 1 , this embodiment provides an image processing and analysis system based on images;

[0091] This solution proposes a comprehensive image processing and analysis system that integrates image radiography, physical examination information analysis and face recognition technology, aiming to provide efficient and accurate physical examination result analysis and identity verification services for students participating in flight-related work. The system integrates a variety of advanced algorithms and models to achieve in-depth mining and intelligent processing of image data, and displays or outputs image analysis results as guiding data, helping medical or staff to conduct final qualitative analysis, while enhancing the security and convenience of identity verification.

[0092] The system consists of several functional modules, namely:

[0093] Data aggregation module, inward image detection module, outward image detection module, comprehensive scoring module and report feedback module. Each of the above modules is operated in sequence according to needs, so as to form an overall operation plan, aiming to complete the analysis and evaluation of each college, and help the auditors to more clearly and quantitatively understand the comprehensive physical fitness and related conditions of the corresponding students;

[0094] System Function Module Description

[0095] Data aggregation module:

[0096] Acquire the real-time imaging data and clinical information of the target trainee; the clinical information includes at least name, age and physical examination data, and simultaneously use facial recognition technology to authenticate the identity of the target trainee;

[0097] When the verification result is correct, no response action is taken;

[0098] When the verification result is wrong, the information proofreading action is performed to ensure the accurate correspondence of the physical examination data;

[0099] Among them, real-time imaging data include X-rays and CT scan images;

[0100] The target trainee (such as a civil aviation pilot trainee) stands in front of an X-ray machine. After the X-rays penetrate the body, they are received by film or digital detectors to form an image, which is an X-ray film. The target trainee lies on a CT scanning bed. The X-rays rotate around the body and scan. After the detector receives the signal, the computer reconstructs it and generates a CT image, which is a CT scan image.

[0101] After acquiring real-time image data, it also includes using an adaptive filtering algorithm to automatically adjust denoising parameters according to the image type, reduce human intervention, and improve processing efficiency; automatically adjust the size of the filter window and the median replacement threshold according to the noise level and image details of the X-ray film; for example, in areas with high noise, the filter window can be appropriately increased; in areas with rich details, a smaller filter window is maintained to retain details; according to the noise distribution and image details of the CT scan image, the size of the filter window, the mean replacement threshold or the spatial distance and pixel grayscale difference weight of the bilateral filter are automatically adjusted; for example, in areas with uneven noise distribution, the size of the filter window can be dynamically adjusted; in areas with rich edge information, the weight of the pixel grayscale difference in the bilateral filter is increased;

[0102] In summary, by using the adaptive filtering algorithm, the denoising parameters can be automatically adjusted according to the image type (such as X-rays and CT scan images), reducing human intervention and improving processing efficiency; this method has important application value in real-time image data processing, especially in the field of medical image analysis, such as in the physical examination image processing of civil aviation pilot trainees, it can more accurately identify abnormal areas and provide a reliable basis for selection;

[0103] The physical examination data in the clinical information includes at least: blood pressure, heart rate and vision;

[0104] The target student's facial image is captured by a camera, and the identity verification is performed using deep learning facial recognition technology. If the name, age, and physical examination data displayed after identity verification are consistent with the obtained clinical information, the verification result is correct. If there is any inconsistency, the verification result is wrong.

[0105] The information verification actions performed are as follows:

[0106] Automatically detect inconsistencies: compare the identity information obtained by facial recognition (i.e. name, age and physical examination data) with the records in the clinical information, and automatically detect inconsistent fields, such as misspelled names, age mismatches or differences in physical examination data; automatically correct information: for inconsistencies that can be corrected automatically, such as misspelled names, the system will automatically correct the displayed information based on the facial recognition results; for inconsistencies that require manual confirmation, such as differences in age or physical examination data, the system will mark and temporarily store correction suggestions; ensure verification correctness: after correction, the system will perform an information consistency check again to ensure that all information matches; if there are still inconsistencies, the system will trigger a manual review process, while retaining the correction records and review tracks.

[0107] By adopting the above technical solutions, efficient and accurate processing and verification of real-time image data and clinical information of target trainees are achieved; the system can automatically obtain real-time image data such as X-rays and CT scan images, and use adaptive filtering algorithms to automatically adjust denoising parameters according to image types, reducing human intervention and significantly improving processing efficiency;

[0108] At the same time, combined with deep learning facial recognition technology, identity verification of target students is carried out to ensure the accuracy and consistency of clinical information such as name, age and physical examination data; during the verification process, the system can automatically detect and correct information inconsistencies, and mark and temporarily store correction suggestions for inconsistencies that require manual confirmation, ensuring the correctness of verification;

[0109] This solution effectively solves technical problems in traditional methods such as low efficiency in image data processing, frequent human intervention, and inaccurate identity information verification. Especially in the physical examination image processing of civil aviation pilot trainees, it can more accurately identify abnormal areas and correct information errors in a timely manner, providing a reliable and efficient basis for trainee selection.

[0110] Intrinsic image detection module:

[0111] First, the deep learning model is used to analyze the CT scan images in the real-time imaging data to identify and mark the initially confirmed abnormal areas, such as fractures and tumors. Combined with the transfer learning technology, a large amount of public medical imaging data is used for pre-training, and the calibrated real-time imaging data is fine-tuned to improve the recognition accuracy of the model.

[0112] The purpose of the above steps is: to obtain a preliminary confirmation of the abnormal area;

[0113] Among them, the process of obtaining the preliminary confirmed abnormal area is as follows:

[0114] Model selection: training deep learning models, such as convolutional neural networks (CNNs), to automatically detect abnormalities in images; feature extraction: using the model to automatically extract key features in images, including lung nodules and bone abnormalities; abnormality identification: based on the extracted features, the model can identify abnormal areas in the image and perform preliminary classification;

[0115] For example, in CT scan images, the model can automatically identify lung nodules that may indicate lung disease; in bone CT scans, the model can detect bone problems such as fractures and osteoporosis;

[0116] Fine-tuning process:

[0117] In deep learning, fine-tuning refers to the process of making small adjustments to a pre-trained model to adapt it to a new task. Specifically, fine-tuning freezes most of the layers of the pre-trained model and only trains the last layer or the added new layer to adapt to the characteristics of a specific dataset. For example, in the task of identifying abnormalities in pilot lung images, fine-tuning means using a pre-trained CNN model but only training its last classification layer, or adding and training a new classification layer. In this way, the model can retain its understanding of the common features of lung images while learning abnormal features specific to the pilot group, thereby improving recognition accuracy.

[0118] Secondly, multimodal image fusion technology is used to combine CT scan images and X-rays for analysis;

[0119] If the analysis results are consistent, it means that the abnormal area is accurately determined, and the initially confirmed abnormal area is used as the final abnormal area; if the analysis results are inconsistent, the fusion result shall prevail to determine the final abnormal area;

[0120] The purpose of the above steps is: to obtain the abnormal area after further confirmation;

[0121] The process of combining CT scan images and X-rays for analysis is as follows:

[0122] First, the X-ray and CT scan images are preprocessed, such as registration and alignment, to ensure that they are in the same spatial coordinate system; then, the characteristics of each image data, such as the overall structural information of the X-ray and the detailed tomographic information of the CT scan, are used to extract features; then, these features are fused, and methods such as weighted averaging, feature splicing, or deep learning fusion networks can be used. Specifically, feature splicing is used, and the fused features will contain richer information, allowing the abnormal area to be observed and analyzed from multiple angles; finally, by comparing the fusion results with the analysis results of the single image data, if they are consistent, the abnormal area is confirmed to be accurate; if they are inconsistent, the fusion results are used as the basis to determine the final abnormal area; for example, the fusion of X-rays and CT scan results can more accurately identify the location and size of lung nodules.

[0123] Finally, the deep learning algorithm is used to extract features of the abnormal area, including size and density. Then, these features are quantitatively analyzed, and a quantitative analysis model is constructed to generate an abnormality estimate that reflects the degree of impact. The student judgment mechanism is also triggered simultaneously. If the target student is a theoretical student, the current abnormality estimate is output.

[0124] If the target student is a non-theoretical student, the judgment adjustment strategy is executed, and under the condition that the abnormal area is determined to be a bone abnormality, the flight experience and age are used as correction data to generate a corrected abnormality estimate;

[0125] Compare the anomaly estimate to a preset acceptance threshold:

[0126] When the abnormality estimate exceeds the acceptance threshold, it means that the abnormality level exceeds the standard. The absolute value of the difference between the abnormality estimate and the acceptance threshold is positively correlated with the intervention priority. The target students with abnormality levels exceeding the standard will be treated with intervention.

[0127] When the abnormality estimate does not exceed the acceptance threshold, it means that the abnormality is not excessive and no immediate treatment intervention is required;

[0128] The purpose of the above steps is to: determine the degree of impact of the final abnormal area on the target student and decide whether to conduct treatment intervention according to the situation;

[0129] Among them, the formula used to build the quantitative analysis model is as follows:

[0130] ;

[0131] Where Av represents the final anomaly estimate, Both represent weight coefficients, and their value ranges are [0, 1]. X represents the size of the abnormal region (such as area or volume), and D represents the density of the abnormal region, that is, the CT value;

[0132] The weight coefficient is determined by the coefficient of variation method, which is a method of assigning weights to each indicator according to the degree of variation between the current value of each evaluation indicator and the target value. If the numerical difference of a certain indicator is large, it can clearly distinguish the evaluated objects, indicating that the indicator has rich information for distinguishing, and thus the indicator should be given a larger weight. On the contrary, if the numerical difference of each evaluated object on a certain indicator is small, then the ability of this indicator to distinguish the evaluated objects is weak, and thus the indicator should be given a smaller weight. This method directly uses the information contained in each indicator to obtain the weight of the indicator through calculation, so it is objective.

[0133] The design logic is as follows: the formula comprehensively considers the size and density of the abnormal area, and through logarithmic transformation and weight coefficient adjustment, the abnormal features of different ranges can be reasonably integrated to generate an abnormal estimation that reflects the degree of impact; log(X+1) is used to perform logarithmic transformation on the size to reduce the excessive impact of size differences on the abnormal estimation;

[0134] For example: If the size of an abnormal area is X=50, the density is D=70, and the setting The values ​​are 0.6 and 0.4 respectively, then the abnormal estimate Av=0.6×log(50+1)+0.4×70≈3.6+28=31.6;

[0135] It should be noted that:

[0136] Theoretical students have no experience in flying during internship or normal work, and only have theoretical experience from books or data analysis, without being affected by actual flying. Non-theoretical students are students with actual flying experience. They are easily affected by flying during actual flying, such as: a certain pressure on bones. Since such impact is a normal experience, it needs to be corrected and adjusted later.

[0137] Bone abnormality is the result of the above preliminary classification, so it can be directly obtained; if there is no bone abnormality, the abnormality estimation is still output according to the current abnormality estimation;

[0138] The corrected anomaly estimate is generated based on the following formula:

[0139] ;

[0140] ;

[0141] In the formula, represents the initial current abnormal estimation, α and β are correction coefficients, which respectively represent the degree of correction of the abnormal estimation by flight experience and age, and the value range is [0, +∞), E represents flight experience (i.e., flight hours), E_ref represents the reference flight experience value, which is used to adjust the influence of flight experience on the abnormal estimation, and the value is usually 1000 hours for entry level, and A represents age;

[0142] Design logic: The formula introduces flight experience in the form of E / (E+E_ref). This design ensures that the impact of flight experience on abnormal valuation is gradual, that is, the more flight experience, the greater the impact, but the growth trend gradually slows down, avoiding the infinite increase of abnormal valuation caused by infinite increase of flight experience; the logarithmic transformation of age is used to reduce the excessive impact of age differences on the correction value, making the impact of age on abnormal valuation smoother;

[0143] For example: If the original initial current abnormal estimate = 30, flight experience = 1500 hours, reference flight experience = 1000 hours, age = 40 years old, and α = 0.1, β = 0.05, then the final abnormal estimate after correction is Av = 30 × (1 + 0.1 × 1500 / (1500 + 1000) − 0.05 × log (40 + 1)) ≈ 30 × (1 + 0.06 − 0.077) ≈ 28.9;

[0144] When there is no generated corrected abnormal estimate, the final abnormal estimate Av = the initial current abnormal estimate .

[0145] By adopting the above technical solutions, efficient and accurate identification and quantitative analysis of abnormal areas in the real-time image data of target trainees are achieved; the solution uses a deep learning model to perform preliminary abnormality identification on CT scan images, and combines transfer learning technology to improve the recognition accuracy; at the same time, multimodal image fusion technology is used to combine CT scan images and X-ray films for analysis to ensure the accuracy of abnormal areas, and by extracting features and quantitatively analyzing abnormal areas, a quantitative analysis model is constructed to generate abnormal valuations that reflect the degree of impact; for non-theoretical trainees, the correction of abnormal valuations due to flight experience and age is also considered to make the results more in line with the actual situation;

[0146] This solution effectively solves the problems of low accuracy in abnormality recognition, insufficient quantitative analysis, and failure to consider individual differences in traditional image analysis. It improves the accuracy and reliability of abnormality detection. By setting an acceptance threshold, it achieves automatic judgment of the degree of abnormality and intervention recommendations, providing a scientific basis for students' health management and treatment decisions, especially in the selection and health management of civil aviation pilot students.

[0147] Outward image detection module:

[0148] Obtain the target student's body surface image, and if the scar is identified, conduct a quantitative assessment of the scar, combine the physical examination data results, and use vision as a constraint condition to generate an evaluation index after a comprehensive assessment;

[0149] Compare the evaluation indicators with the set thresholds, and provide corresponding system guidance based on the comparison results;

[0150] Among them, scar recognition is usually achieved through image processing technology, such as using convolutional neural network (CNN) to analyze the target student's body surface image. CNN can learn and recognize scar features in the image, such as shape, color, texture, etc., so as to accurately identify the location and range of the scar;

[0151] The quantitative assessment of scars is based on the following:

[0152] ;

[0153] Where Sm represents the scar score obtained by quantitative evaluation of scars, ms represents the area of ​​scars, Lr_max represents the maximum length of the current scar, Bs represents the number of scars on the target student, w1, w2, a1, a2 are all weight coefficients, and the value range is [0, 1], i represents the number of the corresponding scar, i=1, 2, ..., n, n=Bs, represents the area of ​​the i-th scar, represents the maximum length of the i-th scar;

[0154] The method for generating evaluation indicators after comprehensive evaluation is as follows:

[0155] Taking vision as a constraint, when the vision of any eye of the target student reaches 0.7 or above, a comprehensive evaluation calculation is performed. When the vision of both eyes of the target student does not reach 0.7, the evaluation index is set to the set threshold;

[0156] The formula for calculating the evaluation index is as follows:

[0157] ;

[0158] Where, I represents the evaluation index, q1 and q2 are adjustment coefficients used to control the influence of other physical examination data on the evaluation index, and the value range is [0, +∞), bp represents blood pressure, bp_normal represents the standard value of blood pressure (i.e. normal value), bp_range represents the range of blood pressure variation, hr represents heart rate, hr_normal represents the standard value of heart rate (i.e. normal value), hr_range represents the range of heart rate variation, Vl represents the left eye visual acuity, and Vr represents the right eye visual acuity;

[0159] Logical explanation: If the visual acuity of either eye (left or right eye) reaches or exceeds 0.7, the comprehensive evaluation index is calculated; if the visual acuity of both eyes is lower than 0.7, the comprehensive evaluation index is the set threshold; when the visual acuity meets the conditions, the evaluation index I is the product of the scar quantification score Sm and the influence of other physical examination data (such as blood pressure, heart rate);

[0160] For example, suppose a target student has a scar score Sm=4.2, blood pressure 140 / 90 (normal 120 / 80, range ±20), heart rate 85 (normal 70, range ±10), left eye vision 0.8, right eye vision 0.6; because the left eye vision 0.8 ≥ 0.7, the vision constraint condition is met;

[0161] Set q1=0.1, q2=0.1 (other influencing coefficients are omitted);

[0162] Then the comprehensive evaluation index I = 4.2 × (1 + 0.1 × 1) × (1 + 0.1 × 0.5) = 4.2 × 1.1 × 1.15 ≈ 5.313;

[0163] The evaluation index is compared with the set threshold, and the comparison results are as follows:

[0164] When the evaluation index is ≥ the set threshold, the system guidance given is as follows:

[0165] Recommending that the target trainee consider a career other than aircrew (meaning that the target trainee's scarring or other medical data may pose a greater risk to his or her ability to work as an aircrew member);

[0166] When the evaluation index is less than the set threshold, the system guidance given is as follows:

[0167] It is recommended that the target trainees randomly select an aircrew or non-aircrew occupation.

[0168] By adopting the above technical solution, the target trainees' body surface scars can be accurately identified and quantitatively evaluated. The evaluation index is generated by combining the physical examination data (including blood pressure, heart rate, etc.) and vision constraints. The solution uses convolutional neural networks (CNN) to effectively identify scar features, quantifies scars through a set formula, and considers vision as a key constraint. When vision meets the standard, the scar score is combined with other physical examination data to calculate the evaluation index.

[0169] This technical solution solves the problems in traditional methods of scar assessment, such as strong subjectivity, lack of quantitative standards, and failure to fully consider the comprehensive impact of physical examination data. By comparing the evaluation indicators with the set thresholds, the system can give targeted career choice suggestions, which not only guarantees the occupational safety requirements of air crew members, but also provides scientific career planning references for target trainees. This solution improves the objectivity and accuracy of scar assessment.

[0170] Comprehensive scoring module:

[0171] Screen all target students who meet the condition that the evaluation index is less than the set threshold, and summarize the final abnormal valuation and evaluation index of all target students to form a scoring data set, build a scoring calculation model, obtain the comprehensive score value corresponding to each target student, and sort the comprehensive score values ​​in order from large to small to generate a sorting table Q;

[0172] The process of building the evaluation calculation model is as follows:

[0173] For the final outlier valuation:

[0174] ;

[0175] For evaluation metrics:

[0176] ;

[0177] Where S_Av and S_I represent the final anomaly estimation and the score of the evaluation index, respectively; Av_max and Av_min represent the maximum and minimum values ​​of the final anomaly estimation among all target students, respectively; I_max and I_min represent the maximum and minimum values ​​of the evaluation index among all target students, respectively;

[0178] The formula for calculating the comprehensive evaluation value is:

[0179] ;

[0180] In the formula, S represents the comprehensive score;

[0181] Logical explanation: By setting the maximum and minimum values, we convert the original "smaller is better" indicator into a "bigger is better" score; the closer the original value (i.e. the final abnormal valuation and evaluation indicator) is to the minimum value, the larger the score; the closer the original value is to the maximum value, the smaller the score; use the linear mapping method to map the original value to a score range of 0 to 100; add up the score values ​​of each indicator to get the total score; the higher the total score, the better the physical fitness of the target student;

[0182] An example of a table format is as follows:

[0183]

[0184] The data in the above table is only for example, but it can be seen from the above table that student C has the highest total score, indicating that his physical fitness is the best; student B has the lowest total score, indicating that his physical fitness is relatively poor, so the ranking in the ranking table Q is {student C, student A, student B}.

[0185] By adopting the above technical solutions, a comprehensive assessment and ranking of the physical fitness of target students is achieved;

[0186] The scheme first screens out all students whose evaluation indicators are less than the set threshold, and then summarizes their abnormal valuations and evaluation indicators to form a scoring data set; then, a scoring calculation model is built to convert the original "smaller is better" indicator into a "bigger is better" scoring value through linear mapping, and the scoring range is set; finally, the abnormal valuation and evaluation indicator scores are added together to obtain the comprehensive score of each student, and the scores are sorted from large to small according to the score value to generate a sorting table;

[0187] This technical solution solves the problem of incomplete evaluation of a single indicator and lack of comprehensive ranking in traditional evaluation. By comprehensively considering abnormal valuations and evaluation indicators, it reflects the physical fitness of students more comprehensively. At the same time, through scoring and ranking, it intuitively shows the differences in physical fitness between students, provides a scientific basis for selecting outstanding students, ensures the fairness and accuracy of the selection process, and improves the efficiency and quality of personnel selection.

[0188] Report Feedback Module:

[0189] Summarize the real-time image data and clinical information of each target trainee that has been properly authenticated, the comparison results of the abnormal valuation and the acceptance threshold, the comparison results of the evaluation index and the set threshold, and the sorting table Q, generate report data, and encrypt and transmit them, and store them in the analysis database for the trainees and medical staff to view;

[0190] Among them, encrypted transmission technology:

[0191] Encrypted transmission refers to the encryption of transmitted data through encryption algorithms to ensure the security and confidentiality of data during transmission. Commonly used encryption transmission technologies include but are not limited to:

[0192] SSL / TLS Protocol:

[0193] SSL (Secure Sockets Layer) and its successor TLS (Transport Layer Security) are encryption protocols widely used for Internet communications; they provide data encryption, identity authentication, and message integrity verification to ensure that data is securely transmitted between the client and the server;

[0194] HTTPS:

[0195] HTTPS (HyperText Transfer Protocol Secure) is an encrypted version of HTTP that uses the SSL / TLS protocol to encrypt data transmission; it ensures the security of communication between the browser and the server and prevents data from being eavesdropped or tampered with;

[0196] VPN (Virtual Private Network):

[0197] VPN technology enables secure communication between remote users and internal networks by establishing encrypted tunnels over public networks; it is suitable for scenarios that require remote access to sensitive data, such as medical staff remotely viewing trainees' clinical information.

[0198] In this scenario, you can choose to use technologies such as HTTPS or VPN to encrypt and transmit sensitive data such as the target trainee’s real-time imaging data, clinical information, the comparison results of abnormal valuation and acceptance threshold, the comparison results of evaluation indicators and set thresholds, and the sorting table Q;

[0199] The process of storing in the analysis database is as follows:

[0200] Data preparation: Before transmission, package the summarized report data into one or more data files, such as JSON, XML or CSV format, and specifically use JSON format; ensure that the data file contains all necessary information and the data format meets the requirements of the analytical database;

[0201] Encrypted transmission: Data files are securely transmitted to the analytical database server using selected encrypted transmission technology (such as HTTPS); during the transmission process, data files are encrypted to prevent them from being intercepted or tampered with by unauthorized personnel;

[0202] Data reception and decryption: After the analytical database server receives the encrypted data file, it will use the corresponding decryption algorithm to decrypt the data; the decrypted data will be verified for its integrity and authenticity to ensure that the data has not been tampered with during transmission.

[0203] Data storage: Verified data will be stored in the analysis database for students and medical staff to view; the database will adopt appropriate storage structure and index mechanism to improve data query efficiency and performance.

[0204] Access control: The analytical database implements strict access control policies to ensure that only authorized users can view or manipulate stored data; this can be achieved through mechanisms such as user authentication, role management, and permission assignment;

[0205] In summary, encrypted transmission technologies such as HTTPS or VPN are used to ensure the security of data during transmission, while the process of storing it in the analysis database involves multiple links such as data preparation, encrypted transmission, data reception and decryption, data storage, and access control. Through these measures, it can be ensured that the sensitive data of the target students is properly protected and managed.

[0206] Embodiment 2:

[0207] See also Figure 2 Based on Example 1, this embodiment further provides an image processing and analysis method based on images, comprising the following specific steps:

[0208] S1. Acquire the real-time image data and clinical information of the target student; wherein the real-time image data at least includes CT scan images and X-rays; the clinical information at least includes name, age and physical examination data; and simultaneously use face recognition technology to authenticate the identity of the target student;

[0209] When the verification result is correct, no response action is taken;

[0210] When the verification result is wrong, the information proofreading action is performed;

[0211] S2. Use deep learning models to analyze CT scan images, identify and mark preliminarily confirmed abnormal areas; combine transfer learning technology for pre-training, and fine-tune the calibrated real-time image data;

[0212] Use multimodal image fusion technology to combine CT scan images and X-rays for analysis;

[0213] Determine the final abnormal area based on the analysis results;

[0214] Use deep learning algorithms to extract features and conduct quantitative analysis on the final abnormal area, build a quantitative analysis model, and simultaneously trigger the student judgment mechanism to generate abnormal valuation based on the target student type;

[0215] Compare the abnormality estimate with the preset acceptance threshold and decide whether to perform treatment intervention based on the comparison result;

[0216] S3. Obtain the outer surface image of the target student's body, and if scars are identified, perform a quantitative assessment of the scars, combine the physical examination data results, and use vision as a constraint condition to generate an assessment index after comprehensive assessment; compare the assessment index with the set threshold, and give corresponding system guidance based on the comparison results;

[0217] S4. Screen all target students who meet the condition that the evaluation index is less than the set threshold, and summarize the final abnormal valuation and evaluation index of all target students to form a scoring data set, build a scoring calculation model, obtain the comprehensive score value corresponding to each target student, and generate a ranking table Q;

[0218] S5. Summarize the real-time image data and clinical information of each target student that has been correctly authenticated, the comparison results of the abnormal valuation and the acceptance threshold, the comparison results of the evaluation index and the set threshold, and the sorting table Q, generate report data, and encrypt and transmit them and store them in the analysis database.

[0219] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product. A person of ordinary skill in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented in electronic hardware, or in a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution.

[0220] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0221] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. An image processing and analysis system based on images, characterized in that: The system includes: The data aggregation module obtains the real-time image data and clinical information of the target trainees; the real-time image data at least includes CT scan images and X-rays; the clinical information at least includes name, age and physical examination data; and simultaneously uses face recognition technology to authenticate the identity of the target trainees; When the verification result is correct, no response action is taken; When the verification result is wrong, the information proofreading action is performed; The inward image detection module uses a deep learning model to analyze CT scan images, identify and mark preliminarily confirmed abnormal areas; it combines transfer learning technology for pre-training and fine-tunes the calibrated real-time image data; Use multimodal image fusion technology to combine CT scan images and X-rays for analysis; Determine the final abnormal area based on the analysis results; Use deep learning algorithms to extract features and conduct quantitative analysis on the final abnormal area, build a quantitative analysis model, and simultaneously trigger the student judgment mechanism to generate abnormal valuation based on the target student type; Compare the abnormality estimate with the preset acceptance threshold and decide whether to perform treatment intervention based on the comparison result; The outward image detection module obtains the external surface image of the target student's body, and performs a quantitative assessment of the scars when scars are identified. It combines the physical examination data results and takes vision as a constraint to generate an evaluation index after comprehensive assessment. The evaluation index is compared with the set threshold, and corresponding system guidance is given based on the comparison results. The comprehensive scoring module screens all target students who meet the evaluation index < set threshold condition, and summarizes the final abnormal valuation and evaluation index of all target students to form a scoring data set, builds a scoring calculation model, obtains the comprehensive score value corresponding to each target student, and generates a ranking table Q.

2. The image processing and analysis system based on image according to claim 1, characterized in that: The physical examination data in the clinical information includes at least: blood pressure, heart rate and vision; The information verification actions performed are as follows: Automatically detect inconsistencies: compare the identity information obtained by face recognition with the records in the clinical information, and detect inconsistent fields, including at least misspellings of names, age mismatches, or differences in physical examination data; automatically correct information: if the name is misspelled, correct the displayed information based on the face recognition results; if there is an age mismatch or physical examination data difference, mark and temporarily store the correction suggestions; ensure verification correctness: after correction, perform a second information consistency check. If there are still inconsistent fields, trigger the manual review process, and retain the correction records and review tracks.

3. The image processing and analysis system based on image according to claim 1, characterized in that: The process of obtaining a preliminary confirmed abnormal area is as follows: Model selection: training convolutional neural network models; Feature extraction: Using the model to extract key features in CT scan images, including lung nodules and bone abnormalities; Abnormality recognition: Based on the extracted key features, abnormal areas in CT scan images are identified and initially classified.

4. The image processing and analysis system based on image according to claim 1, characterized in that: The process of combining CT scan images and X-rays for analysis is as follows: Register and align X-rays and CT scans; Feature extraction is performed using the overall structural information of X-rays and the detailed cross-sectional information of CT scans; The feature splicing method is used to fuse the extracted features and observe and analyze the abnormal area from several angles; By comparing the fusion results with the analysis results of single image data; If the analysis results are consistent, the initially confirmed abnormal area will be used as the final abnormal area; If the analysis results are inconsistent, the fusion results shall prevail to determine the final abnormal area.

5. The image processing and analysis system based on video according to claim 1, characterized in that: Extract features of the final abnormal area, including size and density; When constructing a quantitative analysis model, the formula used is as follows: ； Where Av represents the final anomaly estimate, Both represent weight coefficients, and their value ranges are [0, 1]. X represents the size of the abnormal area, and D represents the density of the abnormal area, that is, the CT value.

6. The image processing and analysis system based on video according to claim 5, characterized in that: The content that triggers the student judgment mechanism is as follows: If the target student is a theoretical student, then output the current abnormal valuation; If the target student is a non-theoretical student, the judgment adjustment strategy is executed, and under the condition that the abnormal area is determined to be a bone abnormality, the flight experience and age are used as correction data to generate a corrected abnormality estimate; The formula used to generate the corrected anomaly estimate is as follows: ； ； In the formula, represents the initial current abnormal estimation, α and β are correction coefficients, and their value range is [0, +∞), E represents flight experience, E_ref represents reference flight experience value, and A represents age.

7. The image processing and analysis system based on image according to claim 6, characterized in that: The quantitative assessment of scars is based on the following: ； Where Sm represents the scar score obtained by quantitative evaluation of scars, ms represents the area of ​​scars, Lr_max represents the maximum length of the current scar, Bs represents the number of scars on the target student, w1, w2, a1, a2 are all weight coefficients, and the value range is [0, 1], i represents the number of the corresponding scar, i=1, 2, ..., n, n=Bs; The method for generating evaluation indicators after comprehensive evaluation is as follows: Taking vision as a constraint, when the vision of any eye of the target student reaches 0.7 or above, a comprehensive evaluation calculation is performed. When the vision of both eyes of the target student does not reach 0.7, the evaluation index is set to the set threshold; The formula for calculating the evaluation index is as follows: ； Where, I represents the evaluation index, q1 and q2 are adjustment coefficients with a value range of [0, +∞), bp represents blood pressure, bp_normal represents the standard value of blood pressure, bp_range represents the range of blood pressure variation, hr represents heart rate, hr_normal represents the standard value of heart rate, hr_range represents the range of heart rate variation, Vl represents the visual acuity of the left eye, and Vr represents the visual acuity of the right eye; The evaluation index is compared with the set threshold, and the comparison results are as follows: When the evaluation index is ≥ the set threshold, the system guidance given is as follows: It is recommended that target trainees consider careers other than aircrew; When the evaluation index is less than the set threshold, the system guidance given is as follows: It is recommended that the target trainees randomly select an aircrew or non-aircrew occupation.

8. The image processing and analysis system based on video according to claim 7, characterized in that: The process of building an evaluation calculation model is as follows: For the final outlier valuation: ； For evaluation metrics: ； Where S_Av and S_I represent the final anomaly estimation and the score of the evaluation index, respectively; Av_max and Av_min represent the maximum and minimum values ​​of the final anomaly estimation among all target students, respectively; I_max and I_min represent the maximum and minimum values ​​of the evaluation index among all target students, respectively; The formula for calculating the comprehensive evaluation value is: ； In the formula, S represents the comprehensive score; When generating the ranking table Q, the comprehensive score values ​​are sorted in order from large to small.

9. The image processing and analysis system based on video according to claim 1, characterized in that: The system also includes: a report feedback module, which summarizes the real-time imaging data and clinical information of each target student that has been correctly authenticated, the comparison results of the abnormal valuation and the acceptance threshold, the comparison results of the evaluation index and the set threshold, and the sorting table Q, generates report data, and transmits it in encrypted form and stores it in the analysis database.

10. An image processing and analysis method based on video, using any system described in claim 1 to claim 9, characterized in that: The steps include: S1. Acquire the real-time image data and clinical information of the target student; wherein the real-time image data at least includes CT scan images and X-rays; the clinical information at least includes name, age and physical examination data; and simultaneously use face recognition technology to authenticate the identity of the target student; When the verification result is correct, no response action is taken; When the verification result is wrong, the information proofreading action is performed; S2. Use deep learning models to analyze CT scan images, identify and mark preliminarily confirmed abnormal areas; combine transfer learning technology for pre-training, and fine-tune the calibrated real-time image data; Use multimodal image fusion technology to combine CT scan images and X-rays for analysis; Determine the final abnormal area based on the analysis results; Use deep learning algorithms to extract features and conduct quantitative analysis on the final abnormal area, build a quantitative analysis model, and simultaneously trigger the student judgment mechanism to generate abnormal valuation based on the target student type; Compare the abnormality estimate with the preset acceptance threshold and decide whether to perform treatment intervention based on the comparison result; S3. Obtain the outer surface image of the target student's body, and if scars are identified, perform a quantitative assessment of the scars, combine the physical examination data results, and use vision as a constraint condition to generate an assessment index after comprehensive assessment; compare the assessment index with the set threshold, and give corresponding system guidance based on the comparison results; S4. Screen all target students who meet the condition that the evaluation index is less than the set threshold, and summarize the final abnormal valuation and evaluation index of all target students to form a scoring data set, build a scoring calculation model, obtain the comprehensive score value corresponding to each target student, and generate a ranking table Q; S5. Summarize the real-time image data and clinical information of each target student that has been correctly authenticated, the comparison results of the abnormal valuation and the acceptance threshold, the comparison results of the evaluation index and the set threshold, and the sorting table Q, generate report data, and encrypt and transmit them and store them in the analysis database.

Images