Intelligent digital urinary image processing and diagnosing system and method

Abstract

The invention discloses an intelligent digital urinary image processing and diagnosing system and method, and relates to the technical field of digital image information processing. The intelligent digital urinary image processing diagnosis method comprises the following steps: acquiring current urinary image data and current physiological data of a patient to be diagnosed, and sequentially performing reduction analysis and region division to obtain a plurality of current corrected urinary image region data of the patient to be diagnosed; carrying out feature extraction to obtain a urinary feature vector of each current corrected urinary image region; according to the method, the urinary feature vector of each current abnormal urinary image area is comprehensively analyzed to obtain the regional lesion index, the regional lesion index is compared and analyzed, and corresponding treatment suggestions are given based on the comparison and analysis result, so that the diagnosis accuracy is improved, and the diagnosis accuracy is improved. Therefore, the lesion of each corrected urinary region image can be independently and accurately diagnosed.

Description

Technical Field

[0001] The present invention relates to the technical field of digital image information processing, and in particular to an intelligent digital urology image processing and diagnosis system and method. Background Art

[0002] Intelligent digital technology has become the core driving force for innovation in all walks of life. Especially in the medical field, intelligent digital technology has shown great potential in diagnosis, treatment, monitoring and other aspects. Intelligent digital technology refers to the use of computer algorithms, artificial intelligence (AI), deep learning and other technologies to convert data into intelligent decisions to help improve the accuracy and efficiency of the system. In the field of medical image processing, the application of intelligent digital technology enables automatic analysis of medical images, improving the speed and accuracy of disease diagnosis. Urinary image processing and diagnosis refers to the scanning and imaging of the urinary system through imaging methods (such as CT, MRI, ultrasound, etc.), and the use of computers to process and analyze image data to help doctors discover lesions and assess the condition. Intelligent urinary image processing and diagnosis can automatically process and analyze image data of the urinary system through deep learning and computer vision algorithms, extract key lesion features, and make accurate predictions and diagnoses of the disease.

[0003] The prior art, such as the patent application with announcement number: CN110507287A, discloses an intelligent digital image processing system and method for urology, wherein the intelligent digital image processing system for urology includes: an image acquisition module, a main control module, an image correction module, an image enhancement module, an image optimization module, an image feature extraction module, a disease analysis module, an image storage module, and a display module. The present invention can enhance the edge of the demosaiced image according to the regional characteristics of the input image through the image enhancement module, and can suppress the brightness noise and color noise at the same time to provide a clear and clean image; at the same time, the image optimization module can correct and optimize the brightness compression range and the chroma compression range according to the brightness compression ratio and the chroma compression ratio, and according to the corrected and optimized brightness compression range and chroma compression range, the source image is compressed in color gamut, so that the display is used to display the source image after color gamut compression.

[0004] Based on the above scheme, it is found that the limitations of the existing technology include at least the following problems. First, it is difficult for the existing technology to accurately correct the image deviation caused by the complexity of the disease, which leads to insufficient diagnostic information of the image. Especially when dealing with more complex lesion areas, it is easy to cause distortion of the image data or fail to obtain sufficiently detailed feature information, thereby affecting the accuracy of subsequent lesion analysis. Secondly, the existing technology lacks effective fusion analysis of the patient's current physiological data, resulting in the disease analysis relying only on the image data itself, failing to fully consider the patient's overall physiological state, and thus lacking personalized diagnosis, which in turn makes the diagnostic scheme for different diseases limited. Finally, when dealing with complex urinary system lesions, the existing technology is difficult to achieve accurate regional division, so that the identification of atypical lesions is not sensitive enough, which in turn affects the final lesion diagnosis and treatment recommendations. Summary of the invention

[0005] In view of the deficiencies of the prior art, the present invention provides an intelligent digital urology image processing and diagnosis system and method, which solves the problem that the prior art is difficult to accurately correct image deviations caused by the complexity of the disease, resulting in insufficient diagnostic information of the image, especially when dealing with more complex lesion areas, it is easy to cause distortion of the image data or fail to obtain sufficiently detailed feature information, thereby affecting the accuracy of subsequent lesion analysis. Secondly, the prior art lacks effective fusion analysis of the patient's current physiological data, resulting in the disease analysis relying only on the image data itself, failing to fully consider the patient's overall physiological state, thereby lacking personalized diagnosis, and thus making the diagnostic schemes for different diseases limited. Finally, when dealing with complex urinary system lesions, the prior art is difficult to achieve accurate regional division, so that the identification of atypical lesions is not sensitive enough, thereby affecting the final lesion diagnosis and treatment recommendations.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: an intelligent digital urinary image processing and diagnosis method, comprising the following steps: obtaining current urinary image data and current physiological data of a patient to be diagnosed; inputting the current urinary image data and current physiological data of the patient to be diagnosed into a pre-established restoration model for restoration analysis to obtain current corrected urinary image data of the patient to be diagnosed, and performing regional division to obtain a number of current corrected urinary image regional data of the patient to be diagnosed; performing feature extraction on each current corrected urinary image regional data of the patient to be diagnosed to obtain a urinary feature vector of each current corrected urinary image region of the patient to be diagnosed; performing lesion analysis on the urinary feature vector of each current corrected urinary image region of the patient to be diagnosed to obtain the lesion type of each current abnormal urinary image region of the patient to be diagnosed; and performing comprehensive analysis on the urinary feature vector of each current abnormal urinary image region of the patient to be diagnosed to obtain a regional lesion index of each current abnormal urinary image region of the patient to be diagnosed; comparing and analyzing the regional lesion index of each current abnormal urinary image region of the patient to be diagnosed with a preset regional lesion index interval, and giving corresponding treatment suggestions based on the comparison and analysis results.

[0007] Furthermore, the current urinary image data specifically includes the pixel value and two-dimensional coordinates of each pixel point in the current urinary image, the current corrected urinary image data specifically includes the corrected pixel value and corrected two-dimensional coordinates of each corrected pixel point in the current corrected urinary image, the urinary feature vector specifically includes density feature, morphological feature, and texture feature, and the current physiological data specifically includes heart rate value, pulse pressure value, body fat percentage value, bladder filling value, pulse wave conduction velocity value, and plasma amino acid concentration value.

[0008] Furthermore, the restoration model is specifically a variational autoencoder, which includes an encoder and a decoder, and the pre-establishment process of the variational autoencoder is as follows: obtain several groups of image data and current physiological data of several patients to be diagnosed, and the image data include pixel values ​​and two-dimensional coordinates of each pixel point in the urinary image, and the current physiological data include heart rate value, pulse pressure value, body fat rate value, bladder filling value, pulse wave conduction velocity value, plasma amino acid concentration value; divide each group of image data and current physiological data of each patient to be diagnosed to generate a urinary image training set and a urinary image verification set; based on the urinary image training set, back propagate each training cycle in the set number of training cycles to generate implicit variables, and calculate the reconstruction loss function and KL divergence function respectively, and perform weighted analysis to generate a final loss function; and after each training cycle, perform evaluation analysis based on the urinary image verification set, and update the parameters of the encoder and decoder until the output result of the variational autoencoder meets the expected standard.

[0009] Furthermore, the specific steps for obtaining the current corrected urinary image data of the patient to be diagnosed are as follows: preprocessing the pixel value, two-dimensional coordinates, heart rate value, pulse pressure value, body fat rate value, bladder filling value, pulse wave conduction velocity value, and plasma amino acid concentration value of each pixel point in the current urinary image of the patient to be diagnosed; encoding the pixel value, two-dimensional coordinates, heart rate value, pulse pressure value, body fat rate value, bladder filling value, pulse wave conduction velocity value, and plasma amino acid concentration of each pixel point in the current urinary image of the patient to be diagnosed after the preprocessing in combination with the encoder in the variational autoencoder to generate the mean and standard deviation of the latent space of the patient to be diagnosed, and performing sampling analysis to obtain the latent variables of the patient to be diagnosed; and decoding the latent variables of the patient to be diagnosed in combination with the decoder in the variational autoencoder to obtain the current corrected urinary image data of the patient to be diagnosed, that is, the corrected pixel value and corrected two-dimensional coordinate of each corrected pixel point in the current corrected urinary image.

[0010] Furthermore, the specific steps for obtaining several current corrected urinary image area data of the patient to be diagnosed are as follows: grayscale processing is performed on the corrected pixel value of each corrected pixel point in the current corrected urinary image of the patient to be diagnosed to obtain the corrected grayscale pixel value of each corrected pixel point in the current corrected urinary image of the patient to be diagnosed; and the corrected grayscale pixel value of each corrected pixel point in the current corrected urinary image of the patient to be diagnosed is converted to obtain the density information value of each corrected pixel point in the current corrected urinary image of the patient to be diagnosed; the density information value of each corrected pixel point in the current corrected urinary image of the patient to be diagnosed is respectively judged and analyzed with several preset density information threshold intervals to obtain several types of pixel points in the current corrected urinary image of the patient to be diagnosed; regional connectivity processing is performed on each type of pixel point in the current corrected urinary image of the patient to be diagnosed to obtain several current corrected urinary image area data of the patient to be diagnosed, that is, the corrected pixel value and corrected two-dimensional coordinate of each corrected pixel point in each current corrected urinary image area of ​​the patient to be diagnosed.

[0011] Furthermore, the specific steps of obtaining the urinary feature vector of each current corrected urinary image area of ​​the patient to be diagnosed are as follows: reading the density information value of each corrected pixel point in each current corrected urinary image area of ​​the patient to be diagnosed, and performing a comprehensive analysis to obtain the density uniformity index, i.e., the density feature, in each current corrected urinary image area of ​​the patient to be diagnosed; performing a statistical analysis on each corrected pixel point in each current corrected urinary image area of ​​the patient to be diagnosed to obtain the area of ​​each current corrected urinary image area of ​​the patient to be diagnosed, reading the regional two-dimensional coordinates of several boundary corrected pixel points in each current corrected urinary image area of ​​the patient to be diagnosed, and performing a comprehensive analysis to obtain the area of ​​each current corrected urinary image area of ​​the patient to be diagnosed. The perimeter of each current corrected urinary image area of ​​the patient to be diagnosed is obtained; the area and perimeter of each current corrected urinary image area of ​​the patient to be diagnosed are comprehensively analyzed to obtain the morphological index of each current corrected urinary image area of ​​the patient to be diagnosed, that is, the morphological feature; the corrected grayscale pixel value of each corrected pixel point of each current corrected urinary image area of ​​the patient to be diagnosed and the corrected grayscale pixel values ​​of several neighborhood pixels in the corresponding preset neighborhood are read, and a comprehensive analysis is performed to obtain the LBP value of each current corrected urinary image area of ​​the patient to be diagnosed, that is, the texture feature; a urinary feature vector is established based on the density feature, morphological feature, and texture feature of each current corrected urinary image area of ​​the patient to be diagnosed.

[0012] Furthermore, the formula for calculating the density uniformity index and LBP value in each current urinary image region to be diagnosed of the patient to be diagnosed is as follows: Among them, JyZ h MdX is the density uniformity index of the hth current corrected urinary image area of ​​the patient to be diagnosed. hm is the density information value of the mth corrected pixel point in the hth current corrected urinary image area of ​​the patient to be diagnosed, WtL h is the LBP value of the hth current corrected urinary image area of ​​the patient to be diagnosed, DxS hmn is the corrected grayscale pixel value of the nth neighboring pixel point in the preset neighborhood of the mth corrected pixel point in the hth current corrected urinary image area of ​​the patient to be diagnosed, DxS hm is the corrected grayscale pixel value of the mth corrected pixel point in the hth current corrected urinary image area of ​​the patient to be diagnosed, S(DxS) is the sign function, h=1, 2, 3, …, h0, h0 is the number of current corrected urinary image areas, m=1, 2, 3, …, m0, m0 is the number of corrected pixels, n=1, 2, 3, …, n0, n0 is the number of pixels in the preset neighborhood corresponding to the corrected pixel point.

[0013] Furthermore, the specific steps for obtaining the lesion type of each current corrected urinary image area of ​​the patient to be diagnosed are as follows: reading the density features, morphological features, and texture features in the urinary feature vector of each current corrected urinary image area of ​​the patient to be diagnosed, and performing standardization processing; and comprehensively analyzing the density features, morphological features, and texture features of each current corrected urinary image area of ​​the patient to be diagnosed after the standardization processing, to obtain the diagnostic index of each current corrected urinary image area of ​​the patient to be diagnosed; and analyzing the diagnostic index of each current corrected urinary image area of ​​the patient to be diagnosed with the preset corresponding diagnostic rules, to obtain the lesion type of each current abnormal urinary image area of ​​the patient to be diagnosed; wherein, the formula for calculating the diagnostic index of each current corrected urinary image area of ​​the patient to be diagnosed is as follows: Among them, ZdS h is the diagnostic index of the hth current corrected urinary image region of the patient to be diagnosed, JyZ′ h is the density feature of the hth current corrected urinary image region of the patient to be diagnosed after standardization, α1 is the density index stored in the database, XtY′ h is the morphological feature of the hth current corrected urinary image region of the patient to be diagnosed after standardization, α2 is the morphological index stored in the database, WtL′ h is the texture feature of the hth current corrected urinary image area of ​​the patient to be diagnosed after standardized processing, α3 is the texture index stored in the database, α1+α2+α3=1, h=1, 2, 3, ..., h0, h0 is the number of current corrected urinary image areas, and e is a natural constant.

[0014] Furthermore, the specific steps for obtaining the regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed are as follows: obtaining image data of a standard urinary system anatomical atlas, wherein the image data is specifically the normal pixel value and normal two-dimensional coordinates of each normal pixel point in the standard urinary system anatomical atlas, and performing regional division and feature extraction in sequence to obtain a normal urinary feature vector of each normal urinary image area of ​​the patient to be diagnosed; performing a comprehensive analysis on the urinary feature vector of each current abnormal urinary image area of ​​the patient to be diagnosed and the normal urinary feature vector of the corresponding normal urinary image area to obtain a regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed; wherein the formula for calculating the regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed is as follows: Among them, ZdS i is the regional lesion index of the i-th abnormal urinary image of the patient to be diagnosed, XzT i YsT is the urinary feature vector of the i-th current abnormal urinary image region of the patient to be diagnosed, iis the normal urinary feature vector of the normal urinary image area corresponding to the i-th current abnormal urinary image area of ​​the patient to be diagnosed, i=1, 2, 3, ..., i0, i0 is the number of current abnormal urinary image areas, and e is a natural constant.

[0015] An intelligent digital urinary image processing and diagnosis system comprises: a data acquisition module, a restoration analysis module, a feature extraction module, a lesion analysis module, a comprehensive analysis module, and a comparison analysis module; the data acquisition module is used to acquire the current urinary image data and current physiological data of the patient to be diagnosed; the restoration analysis module is used to input the current urinary image data and current physiological data of the patient to be diagnosed into a pre-established restoration model for restoration analysis, to obtain the current corrected urinary image data of the patient to be diagnosed, and to perform regional division to obtain a plurality of current corrected urinary image regional data of the patient to be diagnosed; the feature extraction module is used to perform feature extraction on each current corrected urinary image regional data of the patient to be diagnosed, The urinary feature vector of each current corrected urinary image area of ​​the patient to be diagnosed is obtained; the lesion analysis module is used to perform lesion analysis on the urinary feature vector of each current corrected urinary image area of ​​the patient to be diagnosed, and obtain the lesion type of each current abnormal urinary image area of ​​the patient to be diagnosed; the comprehensive analysis module is used to perform comprehensive analysis on the urinary feature vector of each current abnormal urinary image area of ​​the patient to be diagnosed, and obtain the regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed; the comparison analysis module is used to compare and analyze the regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed with a preset regional lesion index range, and give corresponding treatment suggestions based on the comparison analysis results.

[0016] The present invention has the following beneficial effects:

[0017] (1) This intelligent digital urinary image processing and diagnostic method introduces a restoration model to correct the urinary image data, thereby removing the influence of the patient's physiological data on the urinary image data, thereby ensuring the authenticity and accuracy of the urinary image data during analysis, especially in the complex lesion areas of the urinary system. The corrected data can better reflect the patient's actual pathological condition, and the regional division further refines the analysis of each corrected urinary regional image, thereby ensuring that the lesion of each corrected urinary regional image can be independently and accurately diagnosed.

[0018] (2) The intelligent digital urological image processing and diagnosis method can identify the specific lesion type of each area by analyzing the urological feature vector of each corrected urological image area, thereby improving the accuracy of diagnosis. Then, by comprehensively analyzing the lesion feature vector, the lesion degree of the lesion type can be further identified, and more targeted treatment suggestions can be provided according to the different lesion degrees of the lesion type. In addition, the comprehensive analysis process also enhances the processing capability of complex lesions, especially for the identification of early lesions.

[0019] (3) This intelligent digital urological image processing and diagnostic method compares the regional lesion index corresponding to the lesion type of each corrected urological image area with the preset lesion index range, thereby assisting doctors in making better treatment decisions. Through detailed analysis of the regional lesion index, it can accurately determine the severity and development trend of the lesion, thereby formulating corresponding intervention measures, making the treatment plan more personalized and targeted, thereby improving the treatment effect and the patient's overall recovery process.

[0020] (4) The intelligent digital urology image processing and diagnosis system realizes a highly automated diagnostic process through modular design. Steps such as data acquisition, restoration analysis, feature extraction, and lesion analysis are completed through independent modules, reducing the need for manual intervention and improving diagnostic efficiency. In addition, through the cooperation of the comprehensive analysis module and the comparison analysis module, the regional lesion index can be quickly given, and treatment recommendations can be provided based on the comparison results, thereby speeding up the time for diagnostic decision-making, thereby ensuring diagnostic accuracy, and then improving the overall medical work efficiency and saving time costs.

[0021] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a flow chart of an intelligent digital urology image processing and diagnosis method of the present invention.

[0023] Figure 2 The present invention is a flowchart of the steps of obtaining a plurality of current corrected urinary image region data of a patient to be diagnosed in an intelligent digital urinary image processing and diagnosis method.

[0024] Figure 3 This is a block diagram of an intelligent digital urology image processing and diagnosis system of the present invention. DETAILED DESCRIPTION

[0025] The embodiment of the present application solves the problem that the existing technology is difficult to accurately correct image deviations caused by the complexity of the disease, resulting in insufficient diagnostic information of the image, especially when dealing with more complex lesion areas, which is easy to cause distortion of image data or fail to obtain sufficiently detailed feature information, thereby affecting the accuracy of subsequent lesion analysis. Secondly, the existing technology lacks effective fusion analysis of the patient's current physiological data, resulting in the disease analysis relying only on the image data itself, failing to fully consider the patient's overall physiological state, thereby lacking personalized diagnosis, and thus making the diagnosis scheme for different diseases limited. Finally, when dealing with complex urinary system lesions, the existing technology is difficult to achieve accurate regional division, so that the identification of atypical lesions is not sensitive enough, which in turn affects the final lesion diagnosis and treatment recommendations.

[0026] The overall idea of ​​the problem in the embodiment of this application is as follows:

[0027] First, the current urinary image data and current physiological data of the patient to be diagnosed are obtained and input into a pre-established restoration model for restoration analysis to obtain the current corrected urinary image data of the patient to be diagnosed, and regional division is performed to obtain several current corrected urinary image regional data of the patient to be diagnosed. Then, feature extraction is performed on the data of each current corrected urinary image region to obtain the urinary feature vector of each region, and lesion analysis is performed to obtain the lesion type of each current abnormal urinary image region of the diagnosed patient. Then, a comprehensive analysis is performed on the urinary feature vector of each current abnormal urinary image region of the patient to be diagnosed to obtain the regional lesion index of each current abnormal urinary image region of the patient to be diagnosed. Finally, a comparative analysis is performed, and corresponding treatment recommendations are provided based on the comparison results.

[0028] See also Figure 1The embodiment of the present invention provides a technical solution: an intelligent digital urinary image processing and diagnosis method, comprising the following steps: obtaining current urinary image data and current physiological data of a patient to be diagnosed; inputting the current urinary image data and current physiological data of the patient to be diagnosed into a pre-established restoration model for restoration analysis to obtain current corrected urinary image data of the patient to be diagnosed, and performing regional division to obtain a number of current corrected urinary image regional data of the patient to be diagnosed; performing feature extraction on each current corrected urinary image regional data of the patient to be diagnosed to obtain a urinary feature vector of each current corrected urinary image region of the patient to be diagnosed; performing lesion analysis on the urinary feature vector of each current corrected urinary image region of the patient to be diagnosed to obtain the lesion type of each current abnormal urinary image region of the patient to be diagnosed; and performing comprehensive analysis on the urinary feature vector of each current abnormal urinary image region of the patient to be diagnosed to obtain a regional lesion index of each current abnormal urinary image region of the patient to be diagnosed; comparing and analyzing the regional lesion index of each current abnormal urinary image region of the patient to be diagnosed with a preset regional lesion index interval, and giving corresponding treatment suggestions based on the comparison and analysis results.

[0029] The specific process of providing corresponding treatment suggestions based on the comparison and analysis results is as follows: if the regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed is lower than the preset lower limit of the regional lesion index interval (i.e., the minimum value of the regional lesion index interval), suggestions for adjusting lifestyle (i.e., improving eating habits, exercising, etc.) or drug treatment (i.e., providing corresponding drugs for treatment for lesion types with milder symptoms) are provided; if the regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed is within the preset regional lesion index interval, hospitalization (i.e., treatment under the close monitoring of the medical team), regular check-ups (i.e., Regular observation and follow-up, monitoring the development of the current revised urinary image area, and performing regular imaging examinations to evaluate whether the lesion area has a trend of further development); if the regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed is higher than the preset upper limit of the regional lesion index interval (i.e., the maximum value of the regional lesion index interval), emergency treatment (i.e., deciding whether to perform surgery, chemotherapy, or radiotherapy for the current revised urinary image area) and enhanced drug therapy (i.e., deciding to use an enhanced drug treatment plan for the current revised urinary image area, such as high-dose antibiotics, immunosuppressants, or anticancer drugs) are recommended.

[0030] The current urinary image data specifically includes the pixel value and two-dimensional coordinates of each pixel point in the current urinary image. The current corrected urinary image data specifically includes the corrected pixel value and corrected two-dimensional coordinates of each corrected pixel point in the current corrected urinary image. The urinary feature vector specifically includes density feature, morphological feature, and texture feature. The current physiological data specifically includes heart rate value, pulse pressure value, body fat percentage value, bladder filling value, pulse wave conduction velocity value, and plasma amino acid concentration value.

[0031] In this embodiment, the corrected two-dimensional coordinates are based on the coordinate system of the two-dimensional coordinates, that is, the upper left corner of the current urinary image is the origin, the horizontal direction is the X axis, and the vertical direction is the Y axis.

[0032] The heart rate value can be obtained through photoplethysmography, and a too fast heart rate will lead to uneven kidney enhancement effect. When performing renal vascular imaging, a too fast heart rate will cause the blood flow to accelerate, and the rapid blood flow will affect the stability and clarity of the imaging, and will also cause bladder wall instability, making it difficult to accurately assess bladder morphology or lesions (such as tumors, stones, etc.), and cause abnormal enhancement of prostate images, thereby affecting the identification of lesion areas. A too slow heart rate may lead to insufficient blood flow, thereby affecting the contrast effect of the organs, especially in the image enhancement process of organs such as the kidneys and bladder. Insufficient contrast may occur, affecting the diagnosis of lesions.

[0033] The pulse pressure value is the blood pressure difference between cardiac contraction and cardiac relaxation, that is, pulse pressure value = cardiac contraction pressure value - cardiac diastolic pressure value, and both cardiac contraction pressure value and cardiac diastolic pressure value can be obtained through the blood pressure sensor. A pulse pressure that is too high may cause increased tension in the blood vessel wall, thereby affecting the display or contrast of the urinary system blood vessels in the image. A pulse pressure that is too low may affect the excretion of urine, causing urine stagnation and affecting the imaging performance of the bladder and urethra.

[0034] The bladder filling value refers to the amount of urine stored in the bladder, which can be obtained through ultrasound (such as Doppler ultrasound). If the bladder is filled too much or too little, it will cause changes in the bladder wall, thereby affecting the image clarity. For example, overfilling may cause the bladder wall to expand and affect the display of surrounding organs, while insufficient filling will cause the bladder wall to become loose and irregular in shape.

[0035] The pulse wave velocity value is the speed at which the pulse wave propagates when blood passes through the arteries. It reflects the degree of vascular sclerosis and blood flow dynamics and can be obtained through a pressure sensor (such as a systemic arterial blood pressure monitor). A higher pulse wave velocity means arteriosclerosis or poor vascular elasticity, which leads to poor blood flow, especially in the perfusion of important organs such as the kidneys and bladder, affecting the blood supply to the urinary system and thus affecting the image quality. A lower pulse wave velocity means that the blood flow is too fast, which leads to motion artifacts in the image and affects the quality of urinary images.

[0036] The plasma amino acid concentration value is the sum of the content values ​​of all amino acids in the blood (such as arginine, glutamic acid, and phenylalanine), and the content value of each amino acid can be obtained by sampling. A blood sample of the patient to be diagnosed is obtained, and the measurement is performed in combination with liquid chromatography. The measurement results are entered into a database. Abnormal plasma amino acid concentration (such as too high or too low) will affect the physiological state of the urinary system, change the composition of urine, and thus affect the quality of urinary images.

[0037] Specifically, the restoration model is a variational autoencoder, which includes an encoder and a decoder, and the pre-establishment process of the variational autoencoder is as follows: obtain several groups of image data and current physiological data of several patients to be diagnosed, and the image data include pixel values ​​and two-dimensional coordinates of each pixel point in the urinary image, and the current physiological data include heart rate value, pulse pressure value, body fat rate value, bladder filling value, pulse wave conduction velocity value, plasma amino acid concentration value; divide each group of image data and current physiological data of each patient to be diagnosed to generate a urinary image training set and a urinary image verification set; based on the urinary image training set, back propagate each training cycle in the set number of training cycles (i.e., use the reparameterization technique) to generate implicit variables, and calculate the reconstruction loss function and KL divergence function respectively, and perform weighted analysis to generate a final loss function; and after each training cycle, perform evaluation analysis based on the urinary image verification set, and update the parameters of the encoder and decoder until the output result of the variational autoencoder meets the expected standard.

[0038] It needs to be explained that the encoder is specifically an encoder neural network, the decoder is specifically a decoder neural network, and the encoder neural network is specifically an encoding input layer, an encoding convolution layer, an encoding pooling layer, an encoding fully connected layer, a re-parameterization technique, an encoding activation function, and an encoding output layer, and the decoder neural network is specifically a decoding input layer, a decoding fully connected layer, a decoding deconvolution layer, a decoding upsampling layer, a decoding activation function, and a decoding output layer.

[0039] The encoding input layer is used to receive the raw data to be processed (i.e. the urinary imaging data of the patient to be diagnosed and the related current physiological data)

[0040] The encoding convolution layer is used to slide the input image using filters (convolution kernels), gradually extract features, and generate a convolution feature map. The purpose of the convolution operation is to map the input image to a feature space so that the high-level abstract features of the image are more prominent.

[0041] The encoding pooling layer is used to reduce the spatial size of the feature map (i.e., reduce the amount of computation and parameters) while retaining important features.

[0042] The encoding fully connected layer is used to flatten the features output from the convolution and pooling layers into a one-dimensional vector and process it through a series of neurons (i.e., fully connected layers). It is the mean (indicating the expected value of the latent variable) and standard deviation (indicating the distribution width of the latent variable) of the latent space.

[0043] The encoding output layer is used to generate the mean and standard deviation of the latent space.

[0044] The encoding activation function (such as ReLU) is used to increase the nonlinear expression ability of the model, effectively prevent the gradient vanishing problem, and enable the model to better fit complex data.

[0045] The reparameterization technique is used to sample latent variables in the latent space.

[0046] The decoding input layer is used to decode the latent variables output by the encoder from the latent space. These latent variables are low-dimensional vectors that contain the main features of the input data.

[0047] The decoding fully connected layer is used to gradually increase the dimension of the feature map and map the latent variables to the high-dimensional features required to recover the data space.

[0048] The decoding deconvolution layer is used to gradually restore the feature map to the spatial resolution of the original image, which is the inverse process of convolution. It gradually converts the low-dimensional feature map into high-dimensional image data (i.e., restores the details and structure of the image) by learning a set of convolution kernels.

[0049] The decoding upsampling layer is used to increase the spatial resolution of the image through interpolation methods.

[0050] The decoding activation function (such as Sigmoid) is used to introduce nonlinearity so that the decoder can capture more complex data distributions. In the decoder neural network, except for the output layer using the Sigmoid activation function, all other layers use LeakyReLU as the activation function.

[0051] The decoding output layer is used to output the restored urinary image data, that is, the corrected pixel value and corrected two-dimensional coordinates of each restored pixel point in the restored image.

[0052] In this implementation scheme, the variational autoencoder combines the urinary imaging data and physiological data of the patient to be diagnosed, uses the encoder to perform efficient feature extraction on the data, generates a potential representation in the latent space, thereby retaining the main information of the input data, and effectively removing the influence of physiological data on the urinary imaging, thereby generating high-quality and accurate corrected urinary imaging data, thereby improving the accuracy of subsequent analysis, providing a more reliable data basis for diagnosis, and improving diagnostic accuracy. Secondly, by using back propagation and reparameterization techniques in the VAE training process, the network parameters of the encoder and decoder can be effectively optimized. During the back propagation process, the model gradually adjusts the parameters to reduce the error according to the weighted analysis of the reconstruction loss and KL divergence of each training cycle, and finally outputs a high-quality corrected image that meets the expected standards, thereby ensuring that the model can not only process complex imaging data, but also effectively control the variability in the generation process, thereby generating more accurate image correction results, and then significantly improving the stability and generalization ability of the model, avoiding overfitting problems, and ensuring the accuracy and reliability of the system in practical applications.

[0053] Specifically, the specific steps for obtaining the current corrected urinary image data of the patient to be diagnosed are as follows: preprocessing the pixel value, two-dimensional coordinates, heart rate value, pulse pressure value, body fat rate value, bladder filling value, pulse wave conduction velocity value, and plasma amino acid concentration value of each pixel point in the current urinary image of the patient to be diagnosed (for example, scaling the pixel value to between [0,1] and other necessary processing (such as cropping, scaling or alignment, and converting the numerical value of the current physiological data into a suitable range); encoding the pixel value, two-dimensional coordinates, heart rate value, pulse pressure value, body fat rate value, bladder filling value, pulse wave conduction velocity value, and plasma amino acid concentration of each pixel point in the current urinary image of the patient to be diagnosed after the preprocessing by combining the encoder in the variational autoencoder (that is, passing the combination of the input image data and the current physiological data to the encoder network, and the encoder network extracts features through a series of convolutional layers and fully connected layers, and outputs the latent space (that is, the low-dimensional representation space obtained by the encoder network). The mean and standard deviation of the latent space of the patient to be diagnosed are generated, and sampling analysis is performed (i.e., using the reparameterization technique, i.e., sampling a noise vector from the standard normal distribution, and representing the latent variable through the reparameterization formula), and the latent variable of the patient to be diagnosed is obtained (i.e., a low-dimensional latent space vector, which contains the most important feature information of the input data and is compressed and extracted with the help of the encoder); and the latent variable of the patient to be diagnosed is decoded in combination with the decoder in the variational autoencoder (i.e., the latent variable of the latent space is converted into a higher-dimensional feature representation in combination with the decoding fully connected layer, and finally restored to the size and shape of the input data, and the low-dimensional latent representation is gradually converted back to the image pixel information in combination with the decoding deconvolution layer, and the spatial characteristics of the image are gradually restored), and the current corrected urinary image data of the patient to be diagnosed is obtained, i.e., the corrected pixel value and corrected two-dimensional coordinate of each corrected pixel point in the current corrected urinary image.

[0054] In this implementation scheme, by preprocessing the urinary imaging data and physiological data of the patient to be diagnosed, the original data is converted into a standardized format, thereby avoiding data deviation and noise interference, and by combining the imaging data with the physiological data and inputting them into the encoder network in the variational autoencoder, the key features in the image and the potential relationship of the physiological information can be effectively extracted, thereby removing the influence of the physiological information. The reparameterization technique enables the latent variables to not only maintain the key information of the data, but also reduce unnecessary variations in the data, thereby helping to remove noise in the imaging data. The generated latent space vector provides a higher quality and more reliable feature representation, making the subsequent image correction process more accurate. Secondly, by combining physiological data (such as heart rate, Pulse pressure, bladder filling volume, etc.) are combined with urinary imaging data. The variational autoencoder can capture the complex relationship between image features and the patient's physical condition, thereby removing the impact of physiological data on urinary imaging data, thereby helping the system to better understand the pathological information in the image. In addition, the role of the decoder is to convert the latent variables back into high-dimensional features of the original image, gradually restore the spatial structure and details of the image, and at the same time, restore the resolution of the image by decoding the deconvolution layer, and gradually reconstruct accurate pixel information, thereby improving the accuracy of image restoration, and then effectively reducing diagnostic errors caused by various interferences during image acquisition, and generating real and corrected images that conform to the actual pathological conditions, thereby helping relevant personnel make more accurate diagnostic decisions.

[0055] Specifically, Figure 2 As shown, the specific steps of obtaining several current corrected urinary image area data of the patient to be diagnosed are as follows: grayscale processing is performed on the corrected pixel value of each corrected pixel point in the current corrected urinary image of the patient to be diagnosed to obtain the corrected grayscale pixel value of each corrected pixel point in the current corrected urinary image of the patient to be diagnosed; and the corrected grayscale pixel value of each corrected pixel point in the current corrected urinary image of the patient to be diagnosed is converted to obtain the density information value of each corrected pixel point in the current corrected urinary image of the patient to be diagnosed (that is, the density characteristics represented by the pixel point); the density information value of each corrected pixel point in the current corrected urinary image of the patient to be diagnosed is respectively judged and analyzed with several preset density information threshold intervals to obtain several kinds of pixel points in the current corrected urinary image of the patient to be diagnosed (including but not limited to bladder pixel points, kidney pixel points, and ureter pixel points); regional connectivity processing is performed on each pixel point in the current corrected urinary image of the patient to be diagnosed to obtain several current corrected urinary image area data of the patient to be diagnosed, that is, the corrected pixel value and corrected two-dimensional coordinates of each corrected pixel point in each current corrected urinary image area of ​​the patient to be diagnosed.

[0056] The formula for calculating the density information value of each corrected pixel in the current corrected urinary image of the patient to be diagnosed is as follows: MdXm =ω*HdX m -ξ; where MdX m is the density information value of the mth corrected pixel point in the current corrected urinary image of the patient to be diagnosed, ω is the grayscale gain coefficient in the current corrected urinary image of the patient to be diagnosed, which is 2 in this implementation example, HdX m is the corrected grayscale pixel value of the mth corrected pixel point in the current corrected urinary image of the patient to be diagnosed, ξ is the density information reduction coefficient in the current corrected urinary image of the patient to be diagnosed, and in this embodiment, the value is 1000, m=1, 2, 3, ..., m0, m0 is the number of corrected pixels.

[0057] And the specific process of regional connectivity processing is:

[0058] Create a label matrix of the same size as the image to store the region label of each pixel (i.e. the type of pixel). Initially, all pixel labels are 0, indicating that they are not labeled.

[0059] Starting from the upper left corner of the image, scan each pixel row by row and column by column. If the pixel belongs to the target area (that is, the pixel value is 1, or it is 1 after threshold processing) and the pixel has not been marked, the region is marked as connected.

[0060] For each unlabeled target pixel, a depth-first search algorithm is used to find all pixels connected to the pixel and mark them as the same area.

[0061] After the scan is completed, all connected areas will be marked with different labels, and each label corresponds to an independent area. Areas such as the bladder, kidneys, ureters, etc. will have unique labels.

[0062] In this embodiment, by calculating the density information of each corrected pixel point, the image characteristics of different tissues and organs can be effectively distinguished. For example, by grayscale processing and density information value conversion, the characteristics of the bladder, kidney, ureter and other regions can be extracted, thereby providing accurate regional division for subsequent diagnosis and avoiding redundant information. Secondly, the regional connectivity processing can accurately identify the various anatomical regions in the urinary system through algorithms such as depth-first search, and assign a unique label to each region, thereby ensuring that each urinary image region (such as bladder, kidney, etc.) is individually marked, which is convenient for subsequent analysis and comparison. At the same time, each pixel of the image is automatically processed by the algorithm, thereby reducing the workload of manual analysis, thereby reducing the occurrence of human errors, and then improving diagnostic efficiency and accuracy. In addition, the automated regional connectivity algorithm can quickly identify and mark multiple regions to meet the processing requirements of different image data. Finally, based on threshold judgment and regional connectivity algorithm, it can not only process typical organs such as bladder, kidney, ureter, but also adapt to changes under different imaging conditions, thus having good flexibility and adaptability.

[0063] Specifically, the specific steps of obtaining the urinary feature vector of each current corrected urinary image area of ​​the patient to be diagnosed are as follows: read the density information value of each corrected pixel point in each current corrected urinary image area of ​​the patient to be diagnosed, and perform comprehensive analysis to obtain the density uniformity index, that is, the density feature, in each current corrected urinary image area of ​​the patient to be diagnosed; perform statistical analysis on each corrected pixel point in each current corrected urinary image area of ​​the patient to be diagnosed to obtain the area of ​​each current corrected urinary image area of ​​the patient to be diagnosed (that is, the total number of corrected pixel points in each current corrected urinary image area of ​​the patient to be diagnosed); read the regional two-dimensional coordinates of several boundary corrected pixel points in each current corrected urinary image area of ​​the patient to be diagnosed, A comprehensive analysis is performed to obtain the perimeter of each current corrected urinary image area of ​​the patient to be diagnosed; a comprehensive analysis is performed on the area and perimeter of each current corrected urinary image area of ​​the patient to be diagnosed to obtain the morphological index, i.e., the morphological feature, of each current corrected urinary image area of ​​the patient to be diagnosed; the corrected grayscale pixel value of each corrected pixel point of each current corrected urinary image area of ​​the patient to be diagnosed and the corrected grayscale pixel values ​​of several neighborhood pixel points in the corresponding preset neighborhood are read, and a comprehensive analysis is performed to obtain the LBP value, i.e., the texture feature, of each current corrected urinary image area of ​​the patient to be diagnosed; a urinary feature vector is established based on the density feature, morphological feature, and texture feature of each current corrected urinary image area of ​​the patient to be diagnosed.

[0064] The formula for calculating the perimeter and morphological factor of each current corrected urinary image region of the patient to be diagnosed is as follows:

[0065]

[0066] Among them, QzC h is the perimeter of the hth current corrected urinary image region of the patient to be diagnosed, (X hj , Y hj ) is the two-dimensional coordinate of the jth boundary correction pixel point of the hth current correction urinary image region of the patient to be diagnosed, (X h(j+1) , Y h(j+1) ) is the regional two-dimensional coordinate of the j+1th boundary correction pixel point of the hth current correction urinary image region of the patient to be diagnosed, is the two-dimensional coordinate of the j0th boundary correction pixel point of the hth current correction urinary image region of the patient to be diagnosed, (X h1 , Y h1 ) is the two-dimensional coordinate of the first boundary correction pixel point of the hth current correction urinary image area of ​​the patient to be diagnosed, XtY h is the morphological index of the hth current corrected urinary image region of the patient to be diagnosed, ψ is the morphological proportional coefficient, which is 4 in this implementation example, π is the circumference of a circle, which is 3.14 in this implementation example, MjZ h is the area of ​​the hth current corrected urinary image region of the patient to be diagnosed, h=1, 2, 3, ..., h0, h0 is the number of current corrected urinary image regions, j=1, 2, 3, ..., j0, j0 is the number of boundary correction pixels.

[0067] The formula for calculating the density uniformity index and LBP value in each current urinary image area to be diagnosed of the patient to be diagnosed is as follows: Among them, JyZ h MdX is the density uniformity index of the hth current corrected urinary image area of ​​the patient to be diagnosed. hm is the density information value of the mth corrected pixel point in the hth current corrected urinary image area of ​​the patient to be diagnosed, WtL h is the LBP value of the hth current corrected urinary image area of ​​the patient to be diagnosed, DxS hmn is the corrected grayscale pixel value of the nth neighboring pixel point in the preset neighborhood of the mth corrected pixel point in the hth current corrected urinary image area of ​​the patient to be diagnosed, DxS hm is the corrected grayscale pixel value of the mth corrected pixel point in the hth current corrected urinary image area of ​​the patient to be diagnosed, S(DxS) is the sign function, h=1, 2, 3, …, h0, h0 is the number of current corrected urinary image areas, m=1, 2, 3, …, m0, m0 is the number of corrected pixels, n=1, 2, 3, …, n0, n0 is the number of pixels in the preset neighborhood corresponding to the corrected pixel point.

[0068] It needs to be explained that S(DxS) is the discriminant of the sign function: Among them, S(DxS) is the sign function, DxS hmn is the corrected grayscale pixel value of the nth neighboring pixel point in the preset neighborhood of the mth corrected pixel point in the hth current corrected urinary image area of ​​the patient to be diagnosed, DxS hm is the corrected grayscale pixel value of the mth corrected pixel point in the hth current corrected urinary image area of ​​the patient to be diagnosed, h = 1, 2, 3, ..., h0, h0 is the number of current corrected urinary image areas, m = 1, 2, 3, ..., m0, m0 is the number of corrected pixels, n = 1, 2, 3, ..., n0, n0 is the number of pixels in the preset neighborhood corresponding to the corrected pixel point.

[0069] In this embodiment, by combining multiple analysis methods such as density features, morphological features, and texture features, the urinary imaging area of ​​the patient to be diagnosed can be comprehensively evaluated from multiple dimensions, so that the characteristics of the target area can be more comprehensively characterized, and more abundant information can be provided for subsequent analysis and judgment. Secondly, by calculating the area and perimeter of the image area, the size and boundary morphology of the target area can be effectively understood, which helps to evaluate the health status of the organ. The calculation of the morphological index helps to understand the morphological changes of the target area, such as the circularity and uniformity of the area, so it is very important to identify lesions such as tumors, because lesions usually cause irregular changes in morphology. In addition, by calculating the density uniformity of each pixel point, the uniformity and structural integrity of the tissue can be judged, because normal tissue usually has a more uniform density distribution, while abnormal lesion tissue (such as tumors) It may show local uneven density, thus providing an objective quantitative standard for detecting lesion areas, which is particularly helpful in early diagnosis. Then, by analyzing the grayscale relationship between each corrected pixel and its neighboring pixels, the texture features of the image can be extracted to identify subtle local structural changes. Because its texture analysis is highly sensitive to distinguishing different types of lesions, evaluating the boundaries and extensions of tumors, etc., it provides strong support for auxiliary diagnosis. Especially when there are subtle changes in tissue structure that are difficult to detect, the LBP value can reveal hidden details. Finally, by combining density features, morphological features, and texture features into a urinary feature vector, a comprehensive diagnostic feature value can be assigned to each image area, making the diagnosis more quantitative, convenient for computer-aided diagnosis system processing and classification, and with high efficiency and accuracy.

[0070] Specifically, the specific steps for obtaining the lesion type of each current corrected urinary image area of ​​the patient to be diagnosed are as follows: reading the density features, morphological features, and texture features in the urinary feature vector of each current corrected urinary image area of ​​the patient to be diagnosed, and performing standardization processing (i.e., removing the unit); and performing a comprehensive analysis on the density features, morphological features, and texture features of each current corrected urinary image area of ​​the patient to be diagnosed after the standardization processing, to obtain the diagnostic index of each current corrected urinary image area of ​​the patient to be diagnosed; and analyzing the diagnostic index of each current corrected urinary image area of ​​the patient to be diagnosed with the preset corresponding diagnostic rules, to obtain the lesion type of each current abnormal urinary image area of ​​the patient to be diagnosed; wherein, the formula for calculating the diagnostic index of each current corrected urinary image area of ​​the patient to be diagnosed is as follows:

[0071] Among them, ZdS h is the diagnostic index of the hth current corrected urinary image region of the patient to be diagnosed, JyZ′ h is the density feature of the hth current corrected urinary image region of the patient to be diagnosed after standardization, α1 is the density index stored in the database, XtY′ h is the morphological feature of the hth current corrected urinary image region of the patient to be diagnosed after standardization, α2 is the morphological index stored in the database, WtL′ h is the texture feature of the hth current corrected urinary image area of ​​the patient to be diagnosed after standardized processing, α3 is the texture index stored in the database, α1+α2+α3=1, h=1, 2, 3, ..., h0, h0 is the number of current corrected urinary image areas, e is a natural constant, and the value in this implementation example is 2.71.

[0072] It needs to be explained that the specific acquisition process of α1, α2, and α3 is: read the density characteristics, morphological characteristics, and texture characteristics of each current corrected urinary image area of ​​the patient to be diagnosed after standardization, and perform mean analysis to obtain the mean density characteristics, mean morphological characteristics, and mean texture characteristics of the patient to be diagnosed after standardization, and perform sum analysis to obtain the diagnosis and value, and perform proportion analysis on the mean density characteristics, mean morphological characteristics, and mean texture characteristics of the patient to be diagnosed after standardization with the diagnosis and value, and use the proportion results as the corresponding coefficients.

[0073] The domain of the Tanh function (hyperbolic tangent function) is all real numbers, the range is (-1,1), and the expression is Wherein, Tanh(x) is the hyperbolic tangent function, e is a natural constant, and its value is 2.71 in this implementation example.

[0074] The preset diagnostic rules are as follows:

[0075] If the current corrected urinary image region is the kidney region, the diagnostic index of the current corrected urinary image region (i.e., the kidney region) of the patient to be diagnosed is compared and analyzed with the preset (kidney) diagnostic index interval.

[0076] When the diagnostic index of the current corrected urinary image region (i.e., kidney region) of the patient to be diagnosed is lower than the minimum value of the preset (kidney) diagnostic index interval, the current corrected urinary image region (i.e., kidney region) is marked as the current abnormal urinary image region, and the patient to be diagnosed may be suffering from acute renal failure (because it will cause damage or ischemia of kidney tissue, thereby causing decreased kidney density and impaired kidney function, i.e., decreased kidney morphological characteristics and reduced density characteristics) or kidney atrophy (because it will cause decreased kidney volume, i.e., decreased density characteristics and morphological characteristics);

[0077] When the diagnostic index of the current corrected urinary image region (i.e., the kidney region) of the patient to be diagnosed is within the preset (kidney) diagnostic index range, the current corrected urinary image region (i.e., the kidney region) of the patient to be diagnosed is normal;

[0078] When the diagnostic index of the current corrected urinary image region (i.e., kidney region) of the patient to be diagnosed is higher than the maximum value of the preset (kidney) diagnostic index interval, the current corrected urinary image region (i.e., kidney region) is marked as the current abnormal urinary image region, and the patient to be diagnosed may have a kidney tumor (because it will cause the density and morphology of the kidney to increase, i.e., a mass or an enlarged kidney region);

[0079] If the current corrected urinary image region is the bladder region, the diagnostic index of the current corrected urinary image region (i.e., the bladder region) of the patient to be diagnosed is compared and analyzed with the preset (bladder) diagnostic index range.

[0080] When the diagnostic index of the current corrected urinary image region (i.e., bladder region) of the patient to be diagnosed is lower than the minimum value of the preset (bladder) diagnostic index interval, the current corrected urinary image region (i.e., bladder region) is marked as the current abnormal urinary image region, and the patient to be diagnosed may have cystitis (because it will cause edema or thickening of the bladder wall, thereby causing changes in the density and morphology of the bladder, that is, the bladder volume is reduced, the wall is thin, and the density is reduced);

[0081] When the diagnostic index of the current corrected urinary image region (i.e., bladder region) of the patient to be diagnosed is within the preset (bladder) diagnostic index range, the current corrected urinary image region (i.e., bladder region) of the patient to be diagnosed is normal;

[0082] When the diagnostic index of the current modified urinary image region (i.e., bladder region) of the patient to be diagnosed is higher than the maximum value of the preset (bladder) diagnostic index interval, the current modified urinary image region (i.e., bladder region) is marked as the current abnormal urinary image region, and the patient to be diagnosed may have bladder stones (because it appears as a higher density image, and the stones appear more dense on the image, and the stones in the bladder can cause the bladder wall to thicken and increase the density) or bladder tumors (because it can cause the bladder wall to thicken or form a mass, thereby increasing the density);

[0083] If the current corrected urinary image region is the ureter region, the diagnostic index of the current corrected urinary image region (ie, the ureter region) of the patient to be diagnosed is compared and analyzed with the preset (ureter) diagnostic index interval.

[0084] When the diagnostic index of the current corrected urinary image region (i.e., ureter region) of the patient to be diagnosed is lower than the minimum value of the preset (ureter) diagnostic index interval, the current corrected urinary image region (i.e., ureter region) is marked as the current abnormal urinary image region, and the patient to be diagnosed may have ureteral calculi (because it will cause local dilation of the ureter and decrease in density) or urinary tract infection (because it will cause inflammatory reaction, usually manifested as local swelling, congestion and edema, thereby reducing density);

[0085] When the diagnostic index of the current corrected urinary image region (i.e., the ureter region) of the patient to be diagnosed is within the preset (ureter) diagnostic index range, the ureter region of the patient to be diagnosed is normal;

[0086] When the diagnostic index of the current corrected urinary image area (i.e., ureteral area) of the patient to be diagnosed is higher than the maximum value of the preset (ureteral) diagnostic index range, the current corrected urinary image area (i.e., ureteral area) is marked as the current abnormal urinary image area, and the patient to be diagnosed may have a ureteral tumor (because it will cause local hyperplasia of the ureter, thereby increasing the density).

[0087] An example of the lesion type of each current abnormal urinary image area of ​​the patient to be diagnosed is as follows:

[0088] The current corrected urinary imaging areas of the patients to be diagnosed are: kidney area, bladder area, and ureter area.

[0089] The density characteristics, morphological characteristics, and texture characteristics of the kidney area of ​​the current corrected urinary image of the patient to be diagnosed are 150.00 (HU, Helmholtz unit), 1.21, and 28.00, respectively.

[0090] The density characteristics, morphological characteristics, and texture characteristics of the bladder area of ​​the current modified urological image of the patient to be diagnosed are 60.00 (HU, Helmholtz unit), 1.15, and 33.00, respectively.

[0091] The density characteristics, morphological characteristics, and texture characteristics of the ureteral region of the current revised urological image of the patient to be diagnosed are 90.00 (HU, Helmholtz unit), 1.36, and 21.00, respectively.

[0092] The density characteristics, morphological characteristics, and texture characteristics of the kidney area of ​​the current corrected urinary image of the patient to be diagnosed after standardized processing are 0.67, 0.43, and 0.48, respectively.

[0093] The density characteristics, morphological characteristics, and texture characteristics of the bladder area of ​​the current corrected urinary image of the patient to be diagnosed after standardized processing are 0.38, 0.41, and 0.54, respectively.

[0094] The density characteristics, morphological characteristics, and texture characteristics of the ureteral area of ​​the current corrected urinary images of the patient to be diagnosed after standardized processing are 0.54, 0.49, and 0.41, respectively.

[0095] The density index stored in the database is: 0.39.

[0096] The morphological index stored in the database is: 0.33.

[0097] The texture index stored in the database is: 0.28.

[0098] Substituting the above data into the diagnostic index formula corresponding to the current corrected urinary image area of ​​the patient to be diagnosed, we obtain:

[0099] The diagnostic index of the current corrected urinary image region (kidney region) of the patient to be diagnosed=Tanh((0.39*0.67+0.33*0.43+0.28*0.48) / (2.71-1))≈0.31.

[0100] Current modified urinary image area (bladder area) of the patient to be diagnosed = Tanh

[0101] ((0.39*0.38+0.33*0.41+0.28*0.54) / (2.71-1))≈0.25.

[0102] Current corrected urinary imaging area (ureter area) of the patient to be diagnosed = Tanh

[0103] ((0.39*0.54+0.33*0.49+0.28*0.41) / (2.71-1))≈0.28.

[0104] And the preset renal diagnostic index interval is: [0.30,0.60].

[0105] The preset bladder diagnostic index interval is: [0.28, 0.55].

[0106] The preset ureteral diagnostic index interval is: [0.25, 0.5].

[0107] The diagnostic index of each currently corrected urinary image region of the patient to be diagnosed is respectively compared with the preset corresponding diagnostic index interval,

[0108] The diagnostic index of the current corrected urinary image region (kidney region) of the patient to be diagnosed is within the preset kidney diagnostic index range, that is, the kidneys of the patient to be diagnosed are normal;

[0109] The diagnostic index of the current corrected urinary image region (bladder region) of the patient to be diagnosed is lower than the preset minimum value of the bladder diagnostic index interval, that is, the current corrected urinary image region (bladder region) is marked as the current abnormal urinary image region, and the patient to be diagnosed may suffer from cystitis;

[0110] The diagnostic index of the current corrected urinary image region (ureter region) of the patient to be diagnosed is within the preset ureter diagnostic index range, that is, the ureter of the patient to be diagnosed is normal.

[0111] In this embodiment, by comprehensively analyzing the density, morphology and texture features of the image area, the potential abnormal information in the image is fully captured, rather than relying solely on a single feature, thereby increasing the accuracy and robustness of the diagnosis. Secondly, the features of each image area are standardized (de-unitized), which can effectively avoid the scale differences that may exist between different images, making the comparison between different areas more fair and scientific. By calculating the diagnostic index and comparing and analyzing it with the preset diagnostic rules, the possible lesion type of the patient can be determined based on the patient's specific image data, thereby more specifically identifying the lesion area and thus improving the accuracy of the diagnosis. In addition, by By setting clear diagnostic rules and calculation formulas, the system can be automated to improve diagnostic efficiency and reduce manual intervention and judgment errors. The preset diagnostic rules are classified and analyzed according to different image areas (such as kidneys, bladders, ureters, etc.), helping doctors to quickly identify abnormal areas and infer potential lesion types. Finally, the diagnostic process can be flexibly adjusted according to the imaging characteristics of different patients to adapt to different types of lesions, thereby enhancing universality and adaptability. By judging the level of the diagnostic index, the potential risks of lesions can be identified earlier, such as kidney atrophy, bladder stones, etc., which can help early intervention and treatment and reduce the risk of further disease development.

[0112] Specifically, the specific steps for obtaining the regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed are as follows: obtaining image data of a standard urinary system anatomical atlas, wherein the image data is specifically the normal pixel value and normal two-dimensional coordinates of each normal pixel point in the standard urinary system anatomical atlas, and performing regional division and feature extraction in sequence to obtain a normal urinary feature vector of each normal urinary image area of ​​the patient to be diagnosed; performing a comprehensive analysis on the urinary feature vector of each current abnormal urinary image area of ​​the patient to be diagnosed and the normal urinary feature vector of the corresponding normal urinary image area to obtain the regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed; wherein the formula for calculating the regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed is as follows: Among them, ZdS i is the regional lesion index of the i-th abnormal urinary image of the patient to be diagnosed, XzT i YsT is the urinary feature vector of the i-th current abnormal urinary image region of the patient to be diagnosed, i is the normal urinary feature vector of the normal urinary image area corresponding to the i-th current abnormal urinary image area of ​​the patient to be diagnosed, i=1, 2, 3, ..., i0, i0 is the number of current abnormal urinary image areas, e is a natural constant, and its value in this implementation example is 2.71.

[0113] The steps for obtaining the normal urinary feature vector and the urinary feature vector of each normal urinary image region of the patient to be diagnosed are consistent, the calculation logic is consistent, and the calculation formula is also consistent.

[0114] In this embodiment, by comparing and analyzing the urinary feature vector of each abnormal urinary image area of ​​the patient to be diagnosed with the feature vector of the normal area, the degree of the lesion is accurately calculated, thereby effectively eliminating the natural variation and error in the image data. Secondly, using the standard urinary system anatomical atlas as a reference provides a scientific and standardized benchmark for the diagnosis of abnormal areas, thereby avoiding the deviation of the doctor's subjective judgment, thereby improving the reliability and consistency of lesion detection, and by calculating the regional lesion index of each abnormal area, a quantitative index can be provided for each lesion area, thereby objectively reflecting the severity and type of the lesion, thereby helping the doctor to formulate a personalized treatment plan based on the lesion index. In addition, by accurately dividing the image area and extracting the corresponding urinary feature vector, each lesion area is analyzed in detail, thereby ensuring the accuracy of the diagnosis, especially for the detection of small lesions. Finally, since the calculation formula and calculation logic of the regional division and feature extraction of the normal area and the abnormal area are consistent, the uniformity and repeatability of the entire diagnostic process are ensured, thereby reducing the complexity of the calculation and facilitating subsequent optimization and adjustment.

[0115] See also Figure 3 The embodiment of the present invention provides a technical solution: an intelligent digital urinary image processing and diagnosis system, comprising a data acquisition module, a restoration analysis module, a feature extraction module, a lesion analysis module, a comprehensive analysis module, and a comparison analysis module; the data acquisition module is used to acquire the current urinary image data and current physiological data of the patient to be diagnosed; the restoration analysis module is used to input the current urinary image data and current physiological data of the patient to be diagnosed into a pre-established restoration model for restoration analysis, obtain the current corrected urinary image data of the patient to be diagnosed, and perform regional division to obtain a plurality of current corrected urinary image regional data of the patient to be diagnosed; the feature extraction module is used to extract the current corrected urinary image regional data of each patient to be diagnosed Perform feature extraction to obtain the urinary feature vector of each current corrected urinary image area of ​​the patient to be diagnosed; a lesion analysis module is used to perform lesion analysis on the urinary feature vector of each current corrected urinary image area of ​​the patient to be diagnosed to obtain the lesion type of each current abnormal urinary image area of ​​the patient to be diagnosed; a comprehensive analysis module is used to perform comprehensive analysis on the urinary feature vector of each current abnormal urinary image area of ​​the patient to be diagnosed to obtain the regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed; a comparison analysis module is used to compare and analyze the regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed with a preset regional lesion index range, and give corresponding treatment suggestions based on the comparison analysis results.

[0116] In summary, this application has at least the following effects:

[0117] By introducing a restoration model to correct the urinary imaging data, the influence of the patient's physiological data on the urinary imaging data is eliminated, thereby ensuring the authenticity and accuracy of the urinary imaging data during analysis, especially in the complex lesion areas of the urinary system. The corrected data can better reflect the patient's actual pathological condition, and the regional division further refines the analysis of each corrected urinary regional image, thereby ensuring that the lesion of each corrected urinary regional image can be independently and accurately diagnosed.

[0118] By performing lesion analysis on the urinary feature vectors of each corrected urinary image area, the specific lesion type of each area can be identified, thereby improving the accuracy of diagnosis. Then, through comprehensive analysis of the lesion feature vectors, the lesion degree of the lesion type can be further identified, and more targeted treatment recommendations can be provided based on the different lesion degrees of the lesion type. In addition, the comprehensive analysis process also enhances the ability to handle complex lesions, especially for the identification of early lesions.

[0119] By comparing the regional lesion index corresponding to the lesion type of each corrected urinary imaging area with the preset lesion index range, doctors can be assisted in making better treatment decisions. Through detailed analysis of the regional lesion index, the severity and development trend of the lesion can be accurately judged, so as to formulate corresponding intervention measures, making the treatment plan more personalized and targeted, thereby improving the treatment effect and the patient's overall recovery process.

[0120] A highly automated diagnostic process is achieved through modular design. Steps such as data acquisition, restoration analysis, feature extraction, and lesion analysis are all completed through independent modules, reducing the need for manual intervention and improving diagnostic efficiency. In addition, through the cooperation of the comprehensive analysis module and the comparison analysis module, the regional lesion index can be quickly given, and treatment recommendations can be provided based on the comparison results, thereby speeding up the time for diagnostic decision-making, thereby ensuring diagnostic accuracy, and then improving overall medical work efficiency and saving time costs.

[0121] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0122] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. An intelligent digital urology image processing and diagnosis method, characterized in that: The following steps are involved: Acquire current urinary imaging data and current physiological data of the patient to be diagnosed; Inputting the current urinary image data and current physiological data of the patient to be diagnosed into a pre-established restoration model for restoration analysis to obtain the current corrected urinary image data of the patient to be diagnosed, and performing regional division to obtain a plurality of current corrected urinary image regional data of the patient to be diagnosed; Perform feature extraction on each current corrected urinary image region data of the patient to be diagnosed, and obtain a urinary feature vector of each current corrected urinary image region of the patient to be diagnosed; Performing lesion analysis on the urinary feature vector of each current corrected urinary image region of the patient to be diagnosed, and obtaining the lesion type of each current abnormal urinary image region of the patient to be diagnosed; And comprehensively analyzing the urinary feature vector of each current abnormal urinary image region of the patient to be diagnosed, to obtain the regional lesion index of each current abnormal urinary image region of the patient to be diagnosed; The regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed is compared and analyzed with the preset regional lesion index interval, and corresponding treatment recommendations are given based on the comparison and analysis results.

2. The intelligent digital urology image processing and diagnosis method according to claim 1, characterized in that: The current urinary image data specifically includes the pixel value and two-dimensional coordinates of each pixel point in the current urinary image, the current corrected urinary image data specifically includes the corrected pixel value and corrected two-dimensional coordinates of each corrected pixel point in the current corrected urinary image, the urinary feature vector specifically includes density feature, morphological feature, and texture feature, and the current physiological data specifically includes heart rate value, pulse pressure value, body fat percentage value, bladder filling value, pulse wave conduction velocity value, and plasma amino acid concentration value.

3. The intelligent digital urology image processing and diagnosis method according to claim 2, characterized in that: The restoration model is specifically a variational autoencoder, which includes an encoder and a decoder, and the pre-establishment process of the variational autoencoder is as follows: Acquire several groups of image data and current physiological data of several patients to be diagnosed, wherein the image data include pixel values ​​and two-dimensional coordinates of each pixel point in the urinary image, and the current physiological data include heart rate value, pulse pressure value, body fat rate value, bladder filling value, pulse wave velocity value, and plasma amino acid concentration value; Divide each set of imaging data and current physiological data of each patient to be diagnosed to generate a urinary imaging training set and a urinary imaging verification set; Based on the urinary image training set, backpropagation is performed on each training cycle in the set number of training cycles to generate latent variables, and the reconstruction loss function and KL divergence function are calculated respectively, and weighted analysis is performed to generate the final loss function; After each training cycle, an evaluation analysis is performed based on the urinary image validation set, and the parameters of the encoder and decoder are updated until the output of the variational autoencoder meets the expected standards.

4. The intelligent digital urology image processing and diagnosis method according to claim 2, characterized in that: The specific steps for obtaining the current corrected urinary imaging data of the patient to be diagnosed are as follows: Preprocessing the pixel value, two-dimensional coordinates, heart rate value, pulse pressure value, body fat rate value, bladder filling value, pulse wave velocity value, and plasma amino acid concentration value of each pixel point in the current urinary image of the patient to be diagnosed; Combined with the encoder in the variational autoencoder, the pixel value, two-dimensional coordinates, heart rate value, pulse pressure value, body fat rate value, bladder filling value, pulse wave velocity value, and plasma amino acid concentration of each pixel point in the preprocessed current urinary image of the patient to be diagnosed are encoded to generate the mean and standard deviation of the latent space of the patient to be diagnosed, and sampling analysis is performed to obtain the latent variables of the patient to be diagnosed; The decoder in the variational autoencoder is combined to decode the latent variables of the patient to be diagnosed, so as to obtain the current corrected urinary image data of the patient to be diagnosed, that is, the corrected pixel value and corrected two-dimensional coordinates of each corrected pixel point in the current corrected urinary image.

5. The intelligent digital urology image processing and diagnosis method according to claim 2, characterized in that: The specific steps of obtaining several current corrected urinary image region data of the patient to be diagnosed are as follows: Performing grayscale processing on the corrected pixel value of each corrected pixel point in the current corrected urology image of the patient to be diagnosed, to obtain a corrected grayscale pixel value of each corrected pixel point in the current corrected urology image of the patient to be diagnosed; and converting the corrected grayscale pixel value of each corrected pixel point in the current corrected urology image of the patient to be diagnosed to obtain the density information value of each corrected pixel point in the current corrected urology image of the patient to be diagnosed; The density information value of each corrected pixel point in the current corrected urinary image of the patient to be diagnosed is respectively judged and analyzed with a plurality of preset density information threshold intervals to obtain a plurality of pixel points in the current corrected urinary image of the patient to be diagnosed; Performing regional connectivity processing on each pixel point in the current corrected urinary image of the patient to be diagnosed, obtaining a number of current corrected urinary image regional data of the patient to be diagnosed, that is, the corrected pixel value and corrected two-dimensional coordinate of each corrected pixel point in each current corrected urinary image region of the patient to be diagnosed.

6. The intelligent digital urology image processing and diagnosis method according to claim 2, characterized in that: The specific steps of obtaining the urinary feature vector of each current corrected urinary image region of the patient to be diagnosed are as follows: Read the density information value of each corrected pixel point in each current corrected urinary image area of ​​the patient to be diagnosed, and perform comprehensive analysis to obtain the density uniformity index, i.e., density feature, in each current corrected urinary image area of ​​the patient to be diagnosed; Performing statistical analysis on each corrected pixel point of each current corrected urinary image region of the patient to be diagnosed to obtain the area of ​​each current corrected urinary image region of the patient to be diagnosed, reading the regional two-dimensional coordinates of a number of boundary corrected pixel points in each current corrected urinary image region of the patient to be diagnosed, and performing comprehensive analysis to obtain the perimeter of each current corrected urinary image region of the patient to be diagnosed; Comprehensively analyzing the area and perimeter of each current corrected urinary image region of the patient to be diagnosed, and obtaining the morphological index, i.e., morphological characteristics, of each current corrected urinary image region of the patient to be diagnosed; Read the corrected grayscale pixel value of each corrected pixel point of each current corrected urinary image area of ​​the patient to be diagnosed and the corrected grayscale pixel values ​​of several neighborhood pixel points in the corresponding preset neighborhood, and perform comprehensive analysis to obtain the LBP value of each current corrected urinary image area of ​​the patient to be diagnosed, that is, the texture feature; A urinary feature vector is established based on the density feature, morphological feature and texture feature of each current corrected urinary image region of the patient to be diagnosed.

7. The intelligent digital urology image processing and diagnosis method according to claim 6, characterized in that: The formula for calculating the density uniformity index and LBP value in each current urinary image area to be diagnosed of the patient to be diagnosed is as follows: Among them, JyZ h MdX is the density uniformity index of the hth current corrected urinary image area of ​​the patient to be diagnosed. hm is the density information value of the mth corrected pixel point in the hth current corrected urinary image area of ​​the patient to be diagnosed, WtL h is the LBP value of the hth current corrected urinary image area of ​​the patient to be diagnosed, DxS hmn is the corrected grayscale pixel value of the nth neighboring pixel point in the preset neighborhood of the mth corrected pixel point in the hth current corrected urinary image area of ​​the patient to be diagnosed, DxS hm is the corrected grayscale pixel value of the mth corrected pixel point in the hth current corrected urinary image area of ​​the patient to be diagnosed, S(DxS) is the sign function, h=1, 2.3, …, h0, h0 is the number of current corrected urinary image areas, m=1, 2, 3, …, m0, m0 is the number of corrected pixels, n=1, 2, 3, …, n0, n0 is the number of pixels in the preset neighborhood corresponding to the corrected pixel point.

8. The intelligent digital urology image processing and diagnosis method according to claim 2, characterized in that: The specific steps of obtaining the lesion type of each current corrected urinary image region of the patient to be diagnosed are as follows: Reading density features, morphological features, and texture features in the urinary feature vector of each current corrected urinary image region of the patient to be diagnosed, and performing standardization processing; A comprehensive analysis is performed on the density features, morphological features, and texture features of each current corrected urinary image region of the patient to be diagnosed after the standardized processing, so as to obtain a diagnostic index for each current corrected urinary image region of the patient to be diagnosed; and analyzing the diagnostic index of each current corrected urinary image region of the patient to be diagnosed with the preset corresponding diagnostic rules to obtain the lesion type of each current abnormal urinary image region of the patient to be diagnosed; The formula for calculating the diagnostic index of each current corrected urinary image area of ​​the patient to be diagnosed is as follows: Among them, ZdS h is the diagnostic index of the hth current corrected urinary image region of the patient to be diagnosed, JyZ′ h is the density feature of the hth current corrected urinary image region of the patient to be diagnosed after standardization, α1 is the density index stored in the database, XtY′ h is the morphological feature of the hth current corrected urinary image region of the patient to be diagnosed after standardization, α2 is the morphological index stored in the database, WtL′ h is the texture feature of the hth current corrected urinary image area of ​​the patient to be diagnosed after standardized processing, α3 is the texture index stored in the database, α1+α2+α3=1, h=1, 2, 3, …, h0, h0 is the number of current corrected urinary image areas, and e is a natural constant.

9. The intelligent digital urology image processing and diagnosis method according to claim 1, characterized in that: The specific steps of obtaining the regional lesion index of each current abnormal urinary image region of the patient to be diagnosed are as follows: Acquire image data of a standard urinary system anatomical atlas, wherein the image data specifically includes a normal pixel value and a normal two-dimensional coordinate of each normal pixel point in the standard urinary system anatomical atlas, and sequentially perform region division and feature extraction to obtain a normal urinary feature vector of each normal urinary image region of the patient to be diagnosed; Comprehensively analyzing the urinary feature vector of each current abnormal urinary image region of the patient to be diagnosed with the normal urinary feature vector of the corresponding normal urinary image region to obtain the regional lesion index of each current abnormal urinary image region of the patient to be diagnosed; The formula for calculating the regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed is as follows: Among them, ZdS i is the regional lesion index of the i-th abnormal urinary image of the patient to be diagnosed, XzT i YsT is the urinary feature vector of the i-th current abnormal urinary image region of the patient to be diagnosed, i is the normal urinary feature vector of the normal urinary image area corresponding to the i-th current abnormal urinary image area of ​​the patient to be diagnosed, i=1,2,3,…,i0, i0 is the number of current abnormal urinary image areas, and e is a natural constant.

10. An intelligent digital urology image processing and diagnosis system, using the intelligent digital urology image processing and diagnosis method according to any one of claims 1 to 9, characterized in that: include: Data acquisition module, restoration analysis module, feature extraction module, lesion analysis module, comprehensive analysis module, comparison analysis module; The data acquisition module is used to acquire current urinary imaging data and current physiological data of the patient to be diagnosed; The restoration analysis module is used to input the current urinary image data and current physiological data of the patient to be diagnosed into a pre-established restoration model for restoration analysis, obtain the current corrected urinary image data of the patient to be diagnosed, and perform regional division to obtain a plurality of current corrected urinary image regional data of the patient to be diagnosed; The feature extraction module is used to extract features from each current corrected urinary image region data of the patient to be diagnosed, and obtain a urinary feature vector for each current corrected urinary image region of the patient to be diagnosed; The lesion analysis module is used to perform lesion analysis on the urinary feature vector of each current corrected urinary image region of the patient to be diagnosed, and obtain the lesion type of each current abnormal urinary image region of the patient to be diagnosed; The comprehensive analysis module is used to comprehensively analyze the urinary feature vector of each current abnormal urinary image region of the patient to be diagnosed, and obtain the regional lesion index of each current abnormal urinary image region of the patient to be diagnosed; The comparison and analysis module is used to compare and analyze the regional lesion index of each current abnormal urinary image area of ​​the patient to be diagnosed with a preset regional lesion index interval, and to provide corresponding treatment suggestions based on the comparison and analysis results.

Images