Liver disease image diagnosis method and system based on cloud computing

Abstract

The present invention relates to the field of intelligent medical treatment, and discloses a liver disease image diagnosis method and system based on cloud computing, which are used to solve the problem of inaccurate diagnosis through cloud computing when interference factors exist in liver disease image imaging. The method comprises the following steps: obtaining medical image data, and obtaining an intelligent diagnosis model based on training of the medical image data; users upload liver disease images and set a standard image format; obtaining the image format of the liver disease images uploaded by users, and standardizing the liver disease images based on the image format of the liver disease images uploaded by users; obtaining the lesion clarity data of the patient after the standardization, and obtaining the lesion fuzziness index based on the lesion clarity data; and selecting a method for diagnosing the disease based on the lesion fuzziness index, thereby effectively improving the accuracy of the diagnosis result and improving the cure rate of the patient.

Description

Technical Field

[0001] The present invention relates to the field of intelligent medical treatment, and more specifically to a liver disease image diagnosis method and system based on cloud computing. Background Art

[0002] Cloud computing is a technology that provides computing resources and services through the Internet. Users can obtain remote computing power, storage, network, application and other resources on demand without having local hardware and infrastructure.

[0003] Cloud computing-based liver disease image diagnosis is a diagnostic method that combines medical image processing with cloud computing technology. The patient's liver images (such as CT, MRI, ultrasound, etc.) are uploaded to the cloud via the Internet, and automated image analysis and lesion detection are performed with the help of the cloud's powerful computing resources and intelligent algorithms. Cloud computing can provide efficient storage, computing power and collaborative diagnosis support, allowing doctors to remotely view and analyze diagnostic results, helping to determine the type, severity and location of liver diseases, thereby improving diagnostic accuracy and speed. It is especially suitable for large-scale medical scenarios and remote diagnosis and treatment.

[0004] For example, the pathological image diagnosis system and diagnosis method disclosed in the invention patent announcement with announcement number: CN111787174A include: a transfer program, a detection program, a retrieval program and a re-transmission program, wherein the transfer program is used to transfer the data frame encapsulated by the pathological image transmitted by the laptop computer one to the laptop computer two via a pre-constructed connection with the laptop computer two; the detection program is used to detect the response of the laptop computer two; the retrieval program is used to detect the response of the laptop computer two with an error message, and when the error message is detected, the retrieval program is used to detect the data frame encapsulated by the pathological image stored in the temporary storage space two connected to the laptop computer two by relying on the data flag in the error message; the defect of the prior art method of re-transmitting the data frame is very unfavorable to the transmission speed of the data frame is effectively avoided.

[0005] For example, the system and method for rapid intraoperative remote pathology diagnosis published in the invention patent announcement with the announcement number CN110660476A includes: the primary hospital initiates a consultation appointment; the expert receives a text message reminder; the expert receives the consultation; the application end transmits the real-time image of the digital slice to the expert end interface via the Internet; and the client software is operated. During the operation, the clinician can quickly and timely know the patient's pathological results; the remote pathology expert does not need to travel to the primary hospital; as long as there is a network, the remote expert can check the digital pathology image at any time and make a pathological diagnosis in time, which greatly improves the diagnostic efficiency of intraoperative pathology.

[0006] However, in the process of implementing the technical solution of the invention in the embodiments of the present application, the present application found that the above technology has at least the following technical problems:

[0007] In the prior art, liver disease-related indication images automatically uploaded by consumer users may be received. When an automated liver disease image diagnosis is performed on the liver disease-related indication images automatically uploaded by users through an intelligent algorithm, the system may be unable to accurately diagnose based on the images due to possible differences in the formats of the images automatically uploaded by users. Furthermore, the method of diagnosis through the intelligent algorithm is based on analyzing image imaging. If there are interference factors in the image imaging, errors will occur when the intelligent algorithm performs diagnosis, resulting in inaccurate diagnostic results and a reduction in the patient's cure rate. Summary of the invention

[0008] In order to overcome the above-mentioned defects of the prior art, the present invention provides a liver disease image diagnosis method and system based on cloud computing to solve the problems existing in the above-mentioned background technology.

[0009] To achieve the above object, the present invention provides the following technical solutions:

[0010] The liver disease image diagnosis method based on cloud computing includes the following steps: Step 1: obtaining medical image data, and obtaining an intelligent diagnosis model based on the medical image data training; Step 2: the user uploads the liver disease image and sets a standard image format, the image format including standard resolution, standard image size and standard bit depth; Step 3: using image recognition technology to identify the liver disease image uploaded by the user, and obtaining the image format of the liver disease image uploaded by the user, if the image format of the liver disease image is consistent with the standard image format, then the liver disease image is not processed; if the image format of the liver disease image is inconsistent with the standard image format, then the image standard degree is calculated according to the image format of the liver disease image, and the liver disease image is standardized according to the image standard degree; Step 4: for the liver disease image after the standardization processing, the patient's lesion clarity data is obtained, the lesion clarity data includes fat deposition, lesion edge irregularity and lesion location, and the lesion blur index is obtained according to the lesion clarity data; Step 5: selecting a method for diagnosing the condition according to the lesion blur index, and the method for diagnosing the condition includes diagnosis through an intelligent diagnosis model and diagnosis by a doctor.

[0011] Preferably, the steps of calculating the image standard degree according to the image format of the liver disease image are: obtaining the resolution of the liver disease image to obtain the resolution matching degree; obtaining the size of the liver disease image to obtain the file size matching degree; obtaining the bit depth of the liver disease image to obtain the bit depth matching degree; the method of obtaining the image standard degree according to the resolution matching degree, the file size matching degree and the bit depth matching degree is Among them, IS represents the image standard, F represents the resolution matching, J represents the file size matching, and W represents the bit depth matching.

[0012] Preferably, the step of standardizing the liver disease image according to the image standard degree is: comparing the image standard degree with the standard degree threshold; if the image standard degree is greater than the standard degree threshold, adjusting the liver disease image uploaded by the user; if the image standard degree is less than the standard degree threshold, determining that the image is unqualified, and prompting the user to re-upload the image.

[0013] Preferably, the step of obtaining the lesion fuzzy index based on the lesion clarity data is as follows: according to the liver disease image after standardization, the gray value of the image is obtained, and the fat deposition coefficient is obtained according to the gray value of the image; the lesion edge information is obtained through the liver disease image after standardization, and the edge morphology complexity coefficient is calculated according to the lesion edge information; the lesion position of the patient is obtained through the liver disease image after standardization, and the position influence coefficient is calculated according to the lesion position; the method for obtaining the lesion fuzzy index according to the fat deposition coefficient, the edge morphology complexity coefficient and the position influence coefficient is BL=a1×FD+a2×LE+a3×LI, wherein BL represents the lesion fuzzy index, FD represents the fat deposition coefficient, LE represents the edge morphology complexity coefficient, LI represents the position influence coefficient, and a1, a2, and a3 represent the weight coefficients of the fat deposition coefficient, the edge morphology complexity coefficient and the position influence coefficient.

[0014] Preferably, the step of obtaining the fat deposition coefficient according to the gray value of the image is: obtaining the gray value of each pixel in the image to obtain the average gray value of the image; obtaining the standard deviation gray value of the image; and obtaining the fat deposition coefficient according to the average gray value of the image and the standard deviation gray value of the image. Where FD represents the fat deposition coefficient, σ represents the standard deviation gray value of the image, and μ represents the average gray value of the image.

[0015] Preferably, the steps of obtaining lesion edge information through standardized liver disease images and calculating the edge morphology complexity coefficient based on the lesion edge information are as follows: according to the standardized liver disease images, using a threshold segmentation algorithm to extract the lesion area; using an edge detection algorithm to extract the edge of the lesion area to obtain an edge pixel point set; based on the edge pixel point set, extracting the number of edge pixels to obtain the total length of the lesion edge; based on the segmented lesion area, obtaining the area of ​​the lesion area by calculating the number of pixels; using a convex hull algorithm to calculate the minimum convex hull perimeter of the lesion area; and calculating the edge morphology complexity coefficient based on the total length of the lesion edge, the lesion area and the minimum convex hull perimeter.

[0016] Preferably, the step of calculating the edge morphology complexity coefficient according to the total length of the lesion edge, the area of ​​the lesion region and the minimum convex hull perimeter is: obtaining the edge complexity according to the total length of the lesion edge and the minimum convex hull perimeter; obtaining the edge irregularity according to the total length of the lesion edge and the area of ​​the lesion region; and obtaining the edge morphology complexity coefficient according to the edge complexity and the edge irregularity. Where LE represents the edge morphology complexity coefficient.

[0017] Preferably, the steps of obtaining the lesion position of the patient through the standardized liver disease image and calculating the position influence coefficient according to the lesion position are: using a threshold segmentation algorithm to extract the lesion area and obtain a pixel point set in the lesion area; using an edge detection algorithm to identify the edges of blood vessels and organs; calculating the distance between each pixel point in the lesion area and all pixel points on the edges of blood vessels and organs through the Euclidean distance; using a minimum function to find the blood vessel closest to the lesion area and obtain the shortest blood vessel distance, find the organ edge closest to the lesion area and obtain the shortest edge distance; the method for obtaining the position influence coefficient according to the shortest blood vessel distance and the shortest edge distance is: Where LI represents the position influence coefficient, l1 represents the shortest blood vessel distance, and l2 represents the shortest edge distance.

[0018] Preferably, the method steps for selecting a diagnosis of a disease based on the lesion fuzzy index are: comparing the lesion fuzzy index with a preset threshold; if the lesion fuzzy index is less than the preset threshold, it is determined that the patient's current lesion condition is clear, and the intelligent diagnosis model is used to diagnose the disease; if the lesion fuzzy index is greater than the preset threshold, it is determined that the patient's current lesion condition is relatively vague, and the patient's information is transmitted to the doctor, and the disease diagnosis is performed through the doctor's collaboration.

[0019] Preferably, a liver disease image diagnosis system based on cloud computing includes: an intelligent diagnosis model construction module, which is used to obtain an intelligent diagnosis model through training of liver disease medical imaging data, and transmit the intelligent diagnosis model to the diagnosis module; an image calibration module, which is used to calibrate the format of liver disease images uploaded by users, and transmit the calibrated liver disease images to the lesion image analysis module; a lesion image analysis module, which is used to receive the calibrated liver disease images transmitted by the image calibration module, analyze and obtain the lesion fuzzy index, and transmit the lesion fuzzy index to the diagnosis module; a diagnosis module, which is used to receive the intelligent diagnosis model and the lesion fuzzy index transmitted by the intelligent diagnosis model construction module and the lesion image analysis module, and select a method for diagnosing the disease according to the lesion fuzzy index, and the method for diagnosing the disease includes diagnosis through the intelligent diagnosis model and diagnosis by a doctor.

[0020] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0021] 1. Obtain a large amount of high-quality medical imaging data and train an intelligent diagnosis model based on the medical imaging data. High-quality, large-scale medical imaging data can provide rich learning samples for the model, effectively improving the accuracy of the intelligent diagnosis model.

[0022] 2. Use image recognition technology to identify liver disease images uploaded by client users, and standardize liver disease images that do not meet the standards, effectively ensuring the consistency of quality of data measured by consumers, thereby reducing interference caused by factors such as image resolution, contrast or noise. Even images collected by professional equipment from non-medical institutions can be standardized, thereby improving the accuracy and reliability of the intelligent diagnosis model in an environment where consumers transmit data.

[0023] 3. Obtain the patient's lesion clarity data, and obtain the lesion blur index based on the lesion clarity data. The lesion blur index effectively evaluates the difficulty of diagnosing liver disease images uploaded by users;

[0024] 4. By selecting a diagnostic method based on the lesion fuzzy index and combining the intelligent diagnostic model with the doctor's professional judgment, a more comprehensive diagnostic plan can be formed. This combination can improve the overall diagnostic effect and reduce the occurrence of misdiagnosis and missed diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A flow chart of a liver disease image diagnosis method based on cloud computing provided in an embodiment of the present application.

[0026] Figure 2 This is a structural diagram of the liver disease image diagnosis system based on cloud computing provided in an embodiment of the present application. DETAILED DESCRIPTION

[0027] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. In addition, the forms of the various structures recorded in the following embodiments are merely illustrative. The cloud computing-based liver disease image diagnosis method and system involved in the present invention are not limited to the various structures recorded in the following embodiments. All other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0028] The present invention provides a liver disease image diagnosis method based on cloud computing, comprising the following steps:

[0029] Step 1: Obtain a large amount of high-quality medical imaging data and train an intelligent diagnosis model based on the medical imaging data;

[0030] In this embodiment, it should be specifically explained that the steps of obtaining a large amount of high-quality medical imaging data and training an intelligent diagnosis model based on the medical imaging data are as follows:

[0031] Collect a large number of CT, MRI or ultrasound image data of patients with liver diseases. These image data can be obtained from the hospital's medical imaging database or public medical data sets. The data should include liver diseases of different types of patients, such as fatty liver, cirrhosis, liver cancer, etc., and different types of data should be as uniform as possible to ensure that the model can identify different lesions;

[0032] Professional doctors also annotate the image data, including the patient type of each image data. These annotated data are the labels for model training in supervised learning.

[0033] Image preprocessing is performed on the image data. Image preprocessing includes image enhancement: expanding the size of the data set through data enhancement techniques such as rotation, scaling, translation, flipping, etc., which helps the model to be more robust when facing different types of image deformations; image normalization: normalizing the image pixel values ​​to the range of 0-1 so that the neural network can converge faster; image segmentation: before training the model, the liver area is segmented through an automatic segmentation algorithm to ensure that the model only analyzes the area of ​​interest; image alignment: aligning the images of different patients to ensure the relative position and size of the liver are consistent;

[0034] The labeled image data is divided into a training set, a validation set, and a test set in a ratio of 8:1:1. The training set is used to train the model, the validation set is used to adjust the parameters, and the test set is used to evaluate the generalization ability of the model.

[0035] Use the image data in the training set to train the model;

[0036] After model training, the validation set is used to evaluate model performance and the model hyperparameters are adjusted based on the results. Hyperparameters include learning rate, batch size, and convolution kernel size to optimize model performance. K-fold cross-validation technology is used to ensure the robustness of the model on different data through multiple training and validation.

[0037] K-fold cross validation is a model evaluation method used to verify the generalization ability and robustness of machine learning models. Specifically, the data set is randomly divided into K subsets of equal size, and K-1 subsets are used for training each time, and the remaining 1 subset is used for validation. This process is repeated K times, and each subset is used in the validation set. Finally, the average performance indicators (such as accuracy, F1 score, etc.) of K validations are calculated, which can effectively reduce the randomness caused by data partitioning and provide more reliable model performance evaluation.

[0038] Use the test set to evaluate the final performance of the model. The test set is data that has not been seen during the training process to verify the generalization ability of the model, analyze the performance on the test set, and check whether the model has overfitting or underfitting problems. The test results should be highly consistent with the clinician's diagnosis.

[0039] The model tested by the test set is integrated into the cloud computing platform or local server. Through the API interface, the model can be connected to the existing medical imaging system (such as PACS) to realize real-time diagnosis function.

[0040] In this embodiment, it should be specifically explained that the steps of training the model using the image data in the training set are as follows:

[0041] The image data is input into the model, and the output of the model is calculated through the convolution layer, the pooling layer, etc., that is, the prediction result, and the loss value is calculated according to the prediction result and the label data. The loss value calculation is a prior art, and the specific steps are not described in detail in this embodiment;

[0042] Through the back-propagation algorithm, the model calculates the gradient of each parameter according to the size of the loss value, and uses the optimizer to update the model parameters. The back-propagation algorithm is an optimization algorithm used to train artificial neural networks. It updates the parameters by calculating the gradient of the loss function relative to the model parameters. Specifically, the algorithm calculates the output and obtains the loss value through the forward propagation stage, and then passes the loss value through the network layer by layer in the back-propagation stage, using the chain rule to calculate the gradient of each parameter. Finally, these gradients and optimizers are used to adjust the model parameters to minimize the loss function, thereby improving the performance of the model on the task.

[0043] Step 2: The user uploads the liver disease image and sets a standard image format, which includes a standard resolution, a standard image size, and a standard bit depth;

[0044] Resolution refers to the width and height of the image in pixels; bit depth refers to the number of bits used for each pixel in the image, indicating the richness of color or grayscale.

[0045] Step 3: Use image recognition technology to identify the liver disease image uploaded by the user, and obtain the image format of the liver disease image uploaded by the user. If the image format of the liver disease image is consistent with the standard image format, the liver disease image is not processed;

[0046] If the image format of the liver disease image is inconsistent with the standard image format, the image standard degree is calculated according to the image format of the liver disease image, and the liver disease image is standardized according to the image standard degree.

[0047] In this embodiment, it should be specifically explained that the steps of calculating the image standard degree according to the image format of the liver disease image are as follows:

[0048] The method to obtain the resolution of liver disease images and the degree of resolution matching is as follows: Where F represents the resolution matching degree, R u is the resolution of liver disease images, R s For standard resolution, the value of the method is between 0 and 1, where 1 means that the resolution matches exactly, and the closer the value is to 0, the greater the difference;

[0049] The method to obtain the liver disease image size and the file size matching degree is as follows: Where J represents the file size matching degree, S u is the liver disease image size, S s is the standard image size, and the value of the method is between 0 and 1, where 1 means that the resolution matches exactly and 0 means that the difference is greater;

[0050] The method to obtain the bit depth of the liver disease image and the bit depth matching degree is as follows: Where W represents the bit depth matching, B u is the bit depth of the liver disease image, B s For standard bit depth, the method has a value between 0 and 1, where 1 means the resolution matches exactly and 0 means the difference is greater;

[0051] The method for obtaining the image standard degree according to the resolution matching degree, file size matching degree and bit depth matching degree is as follows: Among them, IS represents the image standard, F represents the resolution matching, J represents the file size matching, and W represents the bit depth matching.

[0052] Ensure that the obtained matching value is between 0 and 1. By taking the minimum value, ensure that the matching value is always between 0 and 1, avoid extreme results caused by too large or too small ratios, and provide stable matching evaluation; no matter whether the resolution of the image uploaded by the user is higher or lower than the standard resolution, taking the minimum value can balance it, ensuring that the difference between high and low resolutions can be reasonably evaluated, and will not cause inappropriate matching values ​​due to high or low resolutions; taking the minimum value is simple and efficient, and can directly quantify the similarity between two resolutions without complex calculations, which is convenient for rapid implementation and application in actual systems.

[0053] In this embodiment, it should be specifically explained that the steps of standardizing the liver disease image according to the image standard degree are as follows:

[0054] The image standard degree is compared with the standard degree threshold. If the image standard degree is greater than the standard degree threshold, it is determined that the current image can be adjusted for use, and the liver disease image uploaded by the user is adjusted; if the image standard degree is less than the standard degree threshold, the image is determined to be unqualified, and the user is prompted to re-upload a standardized image.

[0055] Image adjustment is to adjust the image resolution, image size and bit depth in steps through existing image processing libraries, such as OpenCV.

[0056] OpenCV is an open source computer vision and image processing library that supports multiple programming languages ​​and platforms. It provides a large number of efficient functions for tasks such as image processing, video analysis, machine learning, target detection and tracking. There are many benefits to image adjustment through OpenCV: first, it has efficient performance and is suitable for processing large-scale image data; second, it supports a variety of image formats and operations, such as adjusting resolution, file size, bit depth, color enhancement, etc., making image processing more flexible; in addition, OpenCV has extensive community support and rich documentation resources, is easy to develop and expand, and can meet a variety of needs from simple image editing to complex computer vision tasks.

[0057] Step 4: For the liver disease images after standardization, obtain the patient's lesion clarity data, which includes fat deposition, lesion edge irregularity, and lesion location, and obtain the lesion blur index based on the lesion clarity data;

[0058] In this embodiment, it should be specifically explained that the steps of obtaining the lesion fuzzy index according to the lesion clear data are:

[0059] According to the liver disease image after the standardization process, the gray value of the image is obtained, and the fat deposition coefficient is obtained according to the gray value of the image;

[0060] The lesion edge information is obtained through the standardized liver disease image, and the edge morphology complexity coefficient is calculated based on the lesion edge information;

[0061] The lesion position of the patient is obtained through the standardized liver disease image, and the position influence coefficient is calculated according to the lesion position;

[0062] The method to obtain the lesion fuzziness index based on the fat deposition coefficient, edge morphology complexity coefficient and position influence coefficient is BL = a1 × FD + a2 × LE + a3 × LI, where BL represents the lesion fuzziness index and FD represents the fat deposition coefficient. The accumulation of fat may cause significant changes in tissue structure, especially in organs such as the liver, where excessive fat will accumulate in the intercellular space, changing the normal tissue density and morphology. This change not only affects the function of the organ, but also reduces the tissue contrast presented in the medical image, thereby making the boundary of the lesion area blurred and unclear. In medical imaging diagnosis, clear boundaries and contrast are important bases for doctors to determine the nature, scope and severity of the lesion. Due to the accumulation of fat, the characteristics of these lesions may be obscured or confused with normal tissues, especially in the early stages of diseases such as fatty liver or cirrhosis. Fat deposition will greatly reduce the recognition of the image. LE represents the edge morphology complexity coefficient. As the irregularity of the edge of the lesion area increases, the blurriness of the lesion also increases accordingly. When the edge of the lesion is more irregular, the outline of the lesion area becomes more difficult to clearly define in medical images. This irregularity may be caused by the spread of the lesion, uneven growth patterns, or changes in surrounding tissues, which can cause the boundaries to become blurred. LI is expressed as the position influence coefficient. The more complex the specific location of the lesion or the closer the relationship with the surrounding structures, the higher the degree of blur. This relationship means that when the lesion is located near important anatomical structures (such as blood vessels, organ edges, etc.), the clarity of the lesion in the medical image may be affected, causing the visualization of the lesion to become blurred. Specifically, lesions located in deeper layers or more complex anatomical locations often interact more with surrounding tissues. This interaction may cause interference with the image signal, making it more challenging to judge the nature and boundaries of the lesion. In addition, the relationship between the lesion and the adjacent blood vessels or organs may cause artifacts in the imaging technology, making it difficult to accurately determine the boundary and shape of the lesion. a1, a2, and a3 are expressed as weight coefficients of the fat deposition coefficient, the edge morphology complexity coefficient, and the position influence coefficient, and a1+a2+a3=1. The specific values ​​of a1, a2, and a3 are determined by professionals based on the actual situation. For example, a1, a2, and a3 can be 0.4, 0.4, and 0.2.

[0063] In this embodiment, it should be specifically explained that the steps of obtaining the fat deposition coefficient according to the gray value of the image are:

[0064] Get the gray value of each pixel in the image and the method to get the average gray value of the image is: Where μ represents the average gray value of the image, n is the total number of pixels in the image, and I(i) is the gray value of the i-th pixel;

[0065] The method to obtain the standard deviation gray value of the image is Where σ represents the standard deviation gray value of the image, i(i) is the gray value of the i-th pixel, and μ represents the average gray value of the image;

[0066] The method for obtaining the fat deposition coefficient based on the average gray value of the image and the standard deviation gray value of the image is: Where FD represents the fat deposition coefficient, σ represents the standard deviation gray value of the image, and μ represents the average gray value of the image.

[0067] Grayscale values ​​in medical images reflect the density or absorption characteristics of different tissues. The density of fat tissue is usually low, so it presents a lower grayscale value in the image. This makes it possible to directly capture the distribution of fat through grayscale value analysis; grayscale value is the basic statistical feature of the image, which is simple to calculate and has a relatively small amount of calculation. Simple statistical features such as average grayscale value and standard deviation can be used to preliminarily judge the degree of fat deposition.

[0068] In this embodiment, it should be specifically explained that the steps of obtaining the lesion edge information through the liver disease image after the standardization process and calculating the edge morphology complexity coefficient according to the lesion edge information are as follows:

[0069] Based on the standardized liver disease images, the threshold segmentation algorithm is used to extract the lesion area. The threshold segmentation method is a commonly used image processing technology used to separate the foreground from the background in the image. The core idea is to divide the image pixels into different areas by setting one or more gray value thresholds. Usually, the part with a pixel value higher than the threshold is regarded as the foreground, and the part below the threshold is regarded as the background. This method is simple and computationally efficient, and is particularly suitable for images with obvious contrast, such as liver disease lesion extraction in medical images. The lesion area can be accurately separated by a suitable threshold;

[0070] Use edge detection algorithms to extract the edges of the lesion area and obtain a set of edge pixel points. Edge detection algorithms are an image processing technique used to identify areas in an image where brightness or color changes greatly, thereby extracting the edges of the image. These edges usually represent the boundaries of different objects or areas in the image. Common edge detection algorithms include the Sobel operator and the Canny operator, which calculate the gradient or difference of image pixels to find places where pixel values ​​change dramatically (i.e., edges). In medical image processing, such as liver disease images, edge detection algorithms can be used to accurately extract the edges of lesion areas and help doctors analyze the shape and size of lesions.

[0071] According to the edge pixel set, the number of edge pixels is extracted to obtain the total length of the lesion edge, that is, the perimeter of the lesion edge;

[0072] According to the segmented lesion area, the area of ​​the lesion area is obtained by calculating the number of pixels;

[0073] Use convex hull algorithms, such as Graham scan or Andrew's monotone chain algorithm, to calculate the minimum convex hull perimeter of the lesion area. The minimum convex hull refers to the smallest convex polygon that surrounds a set of points, including all points, and there is no concave part in the convex boundary. In other words, the minimum convex hull is a convex area that can completely surround these points, similar to the shape of a rubber band surrounding all the points scattered on a plane and pulling it tight. In medical image processing, the minimum convex hull can be used to surround the lesion area, thereby calculating the convex hull perimeter of the lesion, that is, the minimum closed boundary;

[0074] The edge morphology complexity coefficient was calculated based on the total length of the lesion edge, the lesion area, and the minimum convex hull perimeter.

[0075] In this embodiment, it should be specifically explained that the steps for calculating the edge morphology complexity coefficient according to the edge length, the area of ​​the lesion region and the perimeter of the convex hull are:

[0076] The method to obtain edge complexity based on the total length of the lesion edge and the minimum convex hull perimeter is: Among them C e Expressed as edge complexity, L e Expressed as the total length of the lesion edge, L ch Expressed as the minimum convex hull perimeter, if C e Close to 1, indicating that the edge is relatively smooth. e Greater than 1, indicating more irregular and complex edges;

[0077] The method to obtain edge irregularity based on the total length of the lesion edge and the area of ​​the lesion region is: Among them I e It is expressed as the edge irregularity, and A is expressed as the area of ​​the lesion;

[0078] The method to obtain the edge morphology complexity coefficient based on edge complexity and edge irregularity is: Where LE represents the edge morphology complexity coefficient.

[0079] In this embodiment, it should be specifically explained that the steps of obtaining the lesion position of the patient through the standardized liver disease image and calculating the position influence coefficient according to the lesion position are as follows:

[0080] The lesion area is extracted using a threshold segmentation algorithm, and a pixel point set of the lesion area is obtained;

[0081] Use edge detection algorithms to identify the edges of blood vessels and organs;

[0082] The distance between each pixel in the lesion area and all the pixels at the edge of the blood vessels and organs is calculated by Euclidean distance. Euclidean distance is a commonly used method to measure the distance between two points. It is based on the principle of plane geometry and is defined as the shortest straight-line distance between two points by calculating the straight-line distance between the two points in the coordinate system.

[0083] Through the minimum function, the blood vessel closest to the lesion area is found to obtain the shortest blood vessel distance, and the organ edge closest to the lesion area is found to obtain the shortest edge distance;

[0084] The method for obtaining the position influence coefficient based on the shortest blood vessel distance and the shortest edge distance is: Where LI represents the position influence coefficient, l1 represents the shortest blood vessel distance, and l2 represents the shortest edge distance.

[0085] Step 5: Select a method for diagnosing the disease according to the fuzzy index of the lesion. The method for diagnosing the disease includes diagnosis through an intelligent diagnosis model and diagnosis by a doctor;

[0086] In this embodiment, it should be specifically explained that the method steps for selecting a diagnosis of a disease condition according to the lesion fuzzy index are as follows:

[0087] The lesion fuzziness index is compared with the preset threshold. If the lesion fuzziness index is less than the preset threshold, it is determined that the patient's current lesion condition is relatively clear, and the intelligent diagnosis model is used to diagnose the disease. If the lesion fuzziness index is greater than the preset threshold, it is determined that the patient's current lesion condition is relatively vague. If the diagnosis of the disease using the intelligent diagnosis model may be inaccurate, the patient's information is transmitted to the doctor, and the disease diagnosis is performed through the doctor's collaboration.

[0088] In this embodiment, it should be specifically explained that the liver disease image diagnosis system based on cloud computing includes:

[0089] An intelligent diagnosis model building module is used to obtain an intelligent diagnosis model through training of liver disease medical imaging data and transmit the intelligent diagnosis model to the diagnosis module;

[0090] An image calibration module is used to calibrate the format of the liver disease image uploaded by the user and transmit the calibrated liver disease image to the lesion image analysis module;

[0091] A lesion image analysis module is used to receive the calibrated liver disease image transmitted by the image calibration module, analyze and obtain the lesion fuzzy index, and transmit the lesion fuzzy index to the diagnosis module;

[0092] The diagnosis module is used to receive the intelligent diagnosis model and lesion fuzzy index transmitted by the intelligent diagnosis model construction module and the lesion image analysis module, and select a method for diagnosing the disease according to the lesion fuzzy index. The method for diagnosing the disease includes diagnosis through the intelligent diagnosis model and diagnosis by a doctor.

[0093] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

[0094] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A liver disease image diagnosis method based on cloud computing, characterized in that: The following steps are involved: Step 1: Obtain medical imaging data and train an intelligent diagnosis model based on the medical imaging data; Step 2: The user uploads the liver disease image and sets a standard image format, which includes a standard resolution, a standard image size, and a standard bit depth; Step 3: Use image recognition technology to identify the liver disease image uploaded by the user, and obtain the image format of the liver disease image uploaded by the user. If the image format of the liver disease image is consistent with the standard image format, the liver disease image is not processed; If the image format of the liver disease image is inconsistent with the standard image format, the image standard degree is calculated according to the image format of the liver disease image, and the liver disease image is standardized according to the image standard degree; Step 4: For the liver disease images after standardization, obtain the patient's lesion clarity data, which includes fat deposition, lesion edge irregularity, and lesion location, and obtain the lesion blur index based on the lesion clarity data; Step 5: Select a method for diagnosing the condition based on the lesion fuzzy index. The methods for diagnosing the condition include diagnosis through an intelligent diagnostic model and diagnosis by a doctor.

2. The method for liver disease image diagnosis based on cloud computing according to claim 1, characterized in that: The step of calculating the image standard degree according to the image format of the liver disease image is as follows: Obtain the resolution of the liver disease image and obtain the resolution matching degree; Get the size of liver disease image and get the file size matching degree; Obtaining the bit depth of the liver disease image and obtaining the bit depth matching degree; The method for obtaining the image standard degree according to the resolution matching degree, file size matching degree and bit depth matching degree is as follows: Among them, IS represents the image standard, F represents the resolution matching, J represents the file size matching, and W represents the bit depth matching.

3. The liver disease image diagnosis method based on cloud computing according to claim 1, characterized in that: The steps of standardizing the liver disease image according to the image standard degree are as follows: The image standard degree is compared with the standard degree threshold. If the image standard degree is greater than the standard degree threshold, the liver disease image uploaded by the user is adjusted; if the image standard degree is less than the standard degree threshold, the image is judged to be unqualified and the user is prompted to re-upload the image.

4. The method for liver disease image diagnosis based on cloud computing according to claim 1, characterized in that: The step of obtaining the lesion fuzzy index according to the lesion clear data is: According to the liver disease image after the standardization process, the gray value of the image is obtained, and the fat deposition coefficient is obtained according to the gray value of the image; The lesion edge information is obtained through the standardized liver disease image, and the edge morphology complexity coefficient is calculated based on the lesion edge information; The lesion position of the patient is obtained through the standardized liver disease image, and the position influence coefficient is calculated according to the lesion position; The method to obtain the lesion fuzzy index based on the fat deposition coefficient, edge morphology complexity coefficient and position influence coefficient is BL=a1×FD+a2×LE+a3×LI, where BL represents the lesion fuzzy index, FD represents the fat deposition coefficient, LE represents the edge morphology complexity coefficient, LI represents the position influence coefficient, and a1, a2, and a3 represent the weight coefficients of the fat deposition coefficient, edge morphology complexity coefficient and position influence coefficient.

5. The method for liver disease image diagnosis based on cloud computing according to claim 4, characterized in that: The step of obtaining the fat deposition coefficient according to the gray value of the image is: Get the gray value of each pixel in the image and get the average gray value of the image; Get the standard deviation gray value of the image; The method for obtaining the fat deposition coefficient based on the average gray value of the image and the standard deviation gray value of the image is: Where FD represents the fat deposition coefficient, σ represents the standard deviation gray value of the image, and μ represents the average gray value of the image.

6. The method for liver disease image diagnosis based on cloud computing according to claim 4, characterized in that: The steps of obtaining the lesion edge information through the liver disease image after the standardization process and calculating the edge morphology complexity coefficient according to the lesion edge information are as follows: According to the standardized liver disease images, the lesion area is extracted using the threshold segmentation algorithm; Use edge detection algorithm to extract the edge of the lesion area and obtain the edge pixel point set; According to the edge pixel point set, the number of edge pixel points is extracted to obtain the total length of the lesion edge; According to the segmented lesion area, the area of ​​the lesion area is obtained by calculating the number of pixels; Using the convex hull algorithm, the minimum convex hull perimeter of the lesion area was calculated; The edge morphology complexity coefficient was calculated based on the total length of the lesion edge, the lesion area, and the minimum convex hull perimeter.

7. The method for liver disease image diagnosis based on cloud computing according to claim 6, characterized in that: The step of calculating the edge morphology complexity coefficient according to the total length of the lesion edge, the lesion area and the minimum convex hull perimeter is: The edge complexity was obtained based on the total length of the lesion edge and the minimum convex hull perimeter; The edge irregularity was obtained based on the total length of the lesion edge and the area of ​​the lesion region; The method to obtain the edge morphology complexity coefficient based on edge complexity and edge irregularity is: Where LE represents the edge morphology complexity coefficient.

8. The method for liver disease image diagnosis based on cloud computing according to claim 4, characterized in that: The steps of obtaining the patient's lesion position through the standardized liver disease image and calculating the position influence coefficient according to the lesion position are as follows: The lesion area is extracted using a threshold segmentation algorithm, and a pixel point set of the lesion area is obtained; Use edge detection algorithms to identify the edges of blood vessels and organs; The distance between each pixel in the lesion area and all the pixels at the edge of the blood vessels and organs is calculated by Euclidean distance; Through the minimum function, the blood vessel closest to the lesion area is found to obtain the shortest blood vessel distance, and the organ edge closest to the lesion area is found to obtain the shortest edge distance; The method for obtaining the position influence coefficient based on the shortest blood vessel distance and the shortest edge distance is: Where LI represents the position influence coefficient, l1 represents the shortest blood vessel distance, and l2 represents the shortest edge distance.

9. The method for liver disease image diagnosis based on cloud computing according to claim 1, characterized in that: The method steps for selecting a diagnosis of a disease condition according to the fuzzy index of the disease are as follows: The lesion fuzziness index is compared with the preset threshold. If the lesion fuzziness index is less than the preset threshold, it is determined that the patient's current lesion condition is clear, and the intelligent diagnosis model is used to diagnose the disease. If the lesion fuzziness index is greater than the preset threshold, it is determined that the patient's current lesion condition is relatively vague, and the patient's information is transmitted to the doctor, and the disease is diagnosed through the doctor's collaboration.

10. A liver disease image diagnosis system based on cloud computing, characterized by: The system includes: An intelligent diagnosis model building module is used to obtain an intelligent diagnosis model through training of liver disease medical imaging data and transmit the intelligent diagnosis model to the diagnosis module; An image calibration module is used to calibrate the format of the liver disease image uploaded by the user and transmit the calibrated liver disease image to the lesion image analysis module; A lesion image analysis module is used to receive the calibrated liver disease image transmitted by the image calibration module, analyze and obtain the lesion fuzzy index, and transmit the lesion fuzzy index to the diagnosis module; The diagnosis module is used to receive the intelligent diagnosis model and lesion fuzzy index transmitted by the intelligent diagnosis model construction module and the lesion image analysis module, and select a method for diagnosing the disease according to the lesion fuzzy index. The method for diagnosing the disease includes diagnosis through the intelligent diagnosis model and diagnosis by a doctor.

Images