GB2593824A

GB2593824A - Analysis method and system for feature data change of diabetic retinopathy fundus, and storage device

Info

Publication number: GB2593824A
Application number: GB2104798.0A
Authority: GB
Inventors: Yu Lun; cao Xin-rong; Wang Li-Na; Lin Jia-Wen
Original assignee: Fuzhou Yiying Health Tech Co Ltd
Current assignee: Fuzhou Yiying Health Tech Co Ltd
Priority date: 2018-11-21
Filing date: 2018-12-27
Publication date: 2021-10-06
Also published as: WO2020103288A1; CN111292286A; CN111292286B; GB202104798D0

Abstract

An analysis method and system for feature data change of a diabetic retinopathy fundus image, and a storage device, relating to the field of fundus image processing. The analysis method for the feature data change of the diabetic retinopathy fundus comprises the steps: obtaining the fundus image of a diabetic retinopathy patient (S101); extracting and storing the feature data of the fundus image (S102); analyzing and comparing the feature data of the fundus images of different stages of the diabetic retinopathy patient, and obtaining the fundus screening feature data change of the diabetic retinopathy patient for this time (S103); and analyzing and processing the fundus screening feature data change (S104). During a whole process, the diabetic retinopathy patient can go to any place where a fundus image obtaining device is provided so as to collect the fundus image of the patient, and then uploads the corresponding fundus image to a fundus image processing terminal; the fundus image processing terminal stores the fundus image, and automatically compares and analyzes the fundus image feature data change obtained at different stages.

Description

METHOD AND SYSTEM FOR ANALYZING CHANGES IN FEATURE DATA OF

FUNDUS OF DIABETIC RETINOPATHY, AND A STORAGE DEVICE

Technical Field

The invention relates to the field of fundus image processing, especially relates to a method and systcm for analyzing changes in feature data of ffindus of diabetic retinopathy, and a storage device.

Background

For chronic diseases such as diabetes, there are hundreds of mobile medical APPs available in China, but most of them are focusing on blood sugar and unable to obtain personalized information such as systemic blood vessels and nerves. Without knowing the time of diabetes, it is difficult to know the damage to the health.

The awareness rate of diabetic patients, the proportion of diabetic patients who are aware of and treated, and the proportion of treated patients whose blood glucose level is effectively controlled are all less than 35%. One main reason is that a large number of patients with type 2 diabetes have no feeling before their serious complications, so it is difficult to achieve,'preventive treatment". Another main reason is that so far there is no incentive or method to encourage patients to realize the most important or necessary lifestyles such as diet and exercise with basic treatment, so it is difficult to achieve good therapeutic effects. These problems or difficulties need to be solved or overcome.

Diabetic retinopathy is one of the main complications of diabetes and the leading cause of irreversible blindness in working-age people. As the course of diabetes grows, the chance of suffering from diabetic retinopathy continues to increase, which may eventually lead to irreversible blindness.

Without proper lifestyle with basic treatment and drug treatment, the pathological features of the fundus retina of patients with diabetic retinopathy, such as the number of micro-angiomas, the size of bleeding points, and distribution of hard exudates. will definitely continue to develop.

Therefore, by taking advantage of the above-mentioned features in the disease development of patients with diabetic retinopathy, the obtained retinopathy features obtained after annual fundus screening can be compared and analyzed through the comparison and analysis of the annual data changes in diabetic retinopathy to further evaluate the changes in the fundus condition of patients with diabetic retinopathy, so as to get the evaluation data of related prevention and treatment effects, including the damage of diabetes to the eyes, the control of overall blood glucose level and the treatment effect over a period of time, for providing a great incentive to encourage the diabetes to live a well lifestyle with basic treatment.

The fundus is the only part of the human body where blood vessels and nerves can be directly observed without surgery. Fundus photography allows us to obtain personalized health medical information in precision medicine, if we can process, storage and follow the structured feature data for each fundus screening of diabetic patients, and combine the structured feature data with interrogation data, a new method to achieve the health management or health big data service of diabetic patients can be realized.

Summary

Thus, it is necessary to provide a method for analyzing the changes of the feature data of the fundus of patients with diabetic retinopathy to solve the technical problems mentioned above. The specific technical solutions are as follows: The advantages of the present invention arc: storing the fundus image and the structurally processed feature data of fimdus image by automatically obtaining the fundus images of the patient with diabetic retinopathy at different periods; obtaining changes in the feature data of the patient with diabetic retinopathy by automatically analyzing and comparing the feature data of the fundus images of the patient with diabetic retinopathy at different periods; analyzing and processing the changes in the feature data of the patient with diabetic retinopathy. Throughout the above process, the patient with diabetic retinopathy can obtain his/her own fundus image at any place where the fundus image obtaining device is installed, and then upload the corresponding fimdus image to the fundus image processing terminal which will store the filndus images, and automatically compare and analyze the changes of the fundus images of the patient with diabetic retinopathy at different periods.

Since diabetic retinopathy is one of the major serious complications of diabetic patients, it is also the leading blinding eye disease in China. If patients do not take proper lifestyle with basic treatment and other effective blood glucose level control measures, their conditions must be getting worse. Therefore, by observing the related conditions of fundus retinopathy, under certain circumstances, the data on the evaluation of diabetes prevention and treatment effects, the damage of diabetes to the eyes and even systemic vascular changes, including the overall effect of blood glucose level control over time with the changes in fundus image feature data, can be obtained. Combined with the changes in fundus image feature data, the professional health managers can also combine the effect of blood glucose level control of the patient with diabetic retinopathy in the recent period of time and the analysis of their physical health. By giving relevant health services or recommendations for the prevention of major complications of diabetes, the effect of prevention treatment is achieved.

The patient with diabetic retinopathy can go to any place where the image obtaining terminal is set, to obtain and upload his own fundus image immediately to the fundus image processing tenninal, and the fundus image processing terminal automatically reads the previous fundus images of the patient with diabetic retinopathy at different periods and analyzes and processes these images. The whole process is high in efficiency, low in cost, and good in the experience of the patient with diabetic retinopathy. Furthermore, the data after analysis and processing can be sent to the patient with diabetic retinopathy himself/herself, so that the patient with diabetic retinopathy can understand the development of their own eyes and fundus retinopathy. The above-mentioned analysis report and health service recommendation implemented by this system can continuously improve the patient's overall health by making the patient having a better understanding of the effects of blood glucose control treatment, enhancing the consciousness of taking proper lifestyle. Thus, the system realizes personalized health big data services.

BRIEF DESCRIPTION OF DRAWINGS

FIG. 1 is a flowchart of a method for analyzing changes in feature data of fundus of diabetic retinopathy according to a specific embodiment; FIG. 2 is a schematic diagram of modules of a storage device according to a specific embodiment; FIG. 3 is a schematic view of a system for analyzing changes in feature data of fundus of diabetic retinopathy according to a specific embodiment; Embodiment To describe the technical contents, the stmctural features, the achieved objective and effect in detail, the following will perform description in detail with reference to the specific embodiments and the accompanying drawings.Plcase refer to FIG. 1. In the embodiment, a method for analyzing changes in feature data of fundus of diabetic retinopathy can be applied to a storage device, which includes but is not limited to: a personal computer, a server, a general-purpose computer, and a dedicated computer, a network equipment, intelligent mobile terminals, etc. In the embodiment, a general-purpose computer is taken as an example. The general-purpose computer is installed with a system or a remote center for analyzing changes in feature data of fundus, or provides a browser which can open a webpage to log in a center of cloud health service system for analyzing changes in feature data of fundus. In the embodiment, a specific embodiment of a method for analyzing changes in feature data of fundus is described as follows: The step S10 1 of obtaining a fundus image of a patient with diabetic retinopathy is executed as follows. The fundus image of the patient with diabetic retinopathy is obtained through the fundus camera of a primary health center (such as the primary medical institutions, the health checkups, the health management or primary community clinics), and the obtained fundus image is transmitted to the PC through the data cable and processed by the fimdus image data analysis workstation software, or sent to the PC through the network, and the PC send the fundus image to the fundus image data analysis center. The patient with diabetic retinopathy can also upload the fundus image through the mobile terminal device. It should be noted that, in the embodiment, the primary health center may be located in a remote area where there is no professional ophthalmologist staff or where the cost of hiring professional ophthalmologists is very high.

After the fimdus image data analysis center or the above-mentioned the fimdus image data analysis workstation software obtains the fundus image of the patient with diabetic retinopathy, the step S102 of extracting and identifying feature data of the fundus image and storing the feature data of the fimdus image is executed as follows.

Through automatic or semi-automatic human-computer interactive method, the micro-angiomas, the bleeding points, and the hard exudates of fundus image of patient with diabetic retinopathy are obtained and stored. The fundus image and the feature data thereof can be stored to a cloud server, a specific server, or a local storage device. The purpose of storing the fundus image and the feature data thereof is to facilitate the same patient with diabetic retinopathy comes to check or screen regularly, the currently obtained data can be compared with the previously obtained data each time for better data analysis.

After the feature data of the fimdus image is stored, the step S103 of "determining whether previous feature data of the fundus image of the patient with diabetic retinopathy is stored, and if previous feature data of the fundus image of' the patient with diabetic retinopathy is stored, analyzing and comparing the feature data of the fundus images of the patient with diabetic retinopathy at different periods to obtain changes in the feature data of the fimdus image of the patient with diabetic retinopathy" is executed as follows. According to the patient's name and ID, check the database to see if the previously fimdus image and its feature data of the patient with diabetic retinopathy is stored. If the previously fundus image and its feature data of die patient with diabetic retinopathy is stored, the fundus images and their feature data of patients with diabetic retinopathy at different periods are analyzed and compared, to obtain the changes in the feature data of the fundus image. The details are described as follows.

A brightness histogram equalization is established to preprocess the fundus image; a morphological filter is established to determine positions of a macula and a optic disc in the preprocessed fimdus image; blood vessels of the preprocessed fimdus image are segmented; the fundus image is aligned according to fundus parameters and the identification of the feature data of fundus image is corrected. The fundus parameters include the position of the macula, the position of the optic disc, and the information of the blood vessel; the changes in the feature data of the fundus image are automatically analyzed.

The "identifying feature data of the fundus image" further includes steps of: marking a micro-angiomas and its relative position to a fovea; marking size of a bleeding point and its relative position to the fovea; identifying and analyzing the range of hard exudate and its minimum distance from the fovea.

The details are as follows: In the image to be analyzed, a clear fundus image in which the optic disc, macula and blood vessels can be clearly observed is selected as the standard reference image, and the standard reference image is processed to generate a standard grayscale histogram. According to the grayscale distribution of the standard grayscale histogram, the grayscales of the remaining fimdus images to be analyzed are mapped to obtain a fimdus image with the same grayscale distribution as the standard reference image.

According to the macular brightness and optic disc brightness in thc preprocessed fimdus image. the shapes of the macula and the optic disc, and the distance between the macula and optic disc, a morphological filter is established to determine the positions of the macula and optic disc. in other words, in the preprocessed fundus image, the macula has extremely low brightness, the optic disc has extremely high brightness, the shapes of the two tends to be circular, and the relative distance and position of the two are fixed, so as to establish the morphological filter. The circular areas with extremely low brightness and extremely high brightness in the fundus image are detected and used as the candidate areas of the macula and optic disc. According to the distance and position of the two, the wrong candidate areas are filtered out, and then the center positions of the macula and optic disc are determined.

In the preprocessed fundus image, the blood vessels of the fundus have similar gray-scale information and have a higher contrast with the background. The blood vessels can be segmented through the above features. In the embodiment, the blood vessels are segmented by using a threshold segmentation method. After segmenting the main blood vessels, according to the position of the macula, the position of the optic disc, and the information of the main blood vessels, the change area of the fiandus image is identified on the fundus image. Through the change area, the professionals can quickly see whether the number of microvascular tumors, bleeding spots, or hard exudation has increased, whether the scope of hard exudation has expanded or is getting closer to the macula.

In other embodiments, a rectangle is used to mark the feature data of the relevant fundus image in the fimdus image: micro-angioma area, bleeding point area, hard exudation area (at the same time, record the size of these feature areas and their relative position to the fovea in the database). Different colors represent different features and areas of diabetic retinopathy. For example, white represents hard exudation, pink represents m cro-angioma, and green represents bleeding points. Then, the fundus images are aligned according to the fundus parameters, the fundus parameters include: the position of the macula, the position of the optic disc, and the blood vessel information; and the change area of the fundus image is identified.

The step S104 of analyzing and processing the changes in the feature data in the fundus screening can be executed as follows. The control effect of blood glucose level and the health status of the patient with diabetic retinopathy in a preset period are analyzed and calculated. A corresponding health service recommendation based on the analysis results is given. A report of the control effect of blood glucose level, the health status and the health service recommendation is generated, and the report is sent to the mobile terminal device of the patient with diabetic retinopathy.

The details can be as follows: After data processing and analysis, the obtained data can be sent directly to the patients with diabetic retinopathy, so that the patients with diabetic retinopathy themselves can better understand their own health conditions. It can also be sent to a professional medical institution with the consent of the patients with diabetic retinopathy, assisting the medical staff to quickly understand the condition of the patients with diabetic retinopathy, and provide the follow-up disease control recommendations for the patients with diabetic retinopathy. In this embodiment, the results of data processing and analysis include: whether the number or size of micro-angiomas is increased or not, whether the area of hard exudation is enlarged or involves the area of the macular. If possible, it can be referred for auxiliary judgments with the relevant medical inquiry data and other physical index data sent by the primary health center with the fundus image, and the data is such as a significant increase or decrease in body weight and waist circumference, diet, exercise, no smoking and less alcohol, etc. If there is an increase in the number of micro-angiomas and the area of hard exudation, it indicates that the control of blood glucose level during this period is poor, and the condition of retinopathy is still developing, and further control is needed to ensure a good lifestyle. If the number of the micro-angiomas and the bleeding points increases significantly, the area of hard exudation expands and affects the macular area, it is recommended to do further follow-up diagnosis and treatment. If there is no obvious change in micro-angioma and hard exudation, it indicates that the control level is good, and a good lifestyle and basic treatment should be continued in accordance with the doctor's advice. In particular, when the number of micro-angiomas decreases or some micro-angiomas disappear, it means that the blood supply of the capillaries in the area is severely insufficient or disappears. When the extent of hard exudation expands or approaches the macular area, it means that the symptoms of macular edema are serious or there is a danger of blindness. Special attention should be paid or it is recommended to do further review or go to the hospital for examination. The augmented reality (AR) technology can be used to make simple demonstration animations of changes in these fundus features and conditions that may affect vision or general health if they continue to develop, and superimpose them on real fundus photos to achieve visual education effects, and encourage the patients to live a good lifestyle with basic treatment and have awareness of timely screening and timely preventive treatment.

After obtaining the fundus images of patients with diabetic retinopathy at different periods, the fundus images and the feature data of the fiindus images after structured processing is stored. Through automatic analysis and comparison of features of the fundus images at different periods, die changes in the features of the fundus images of the patients with diabetic retinopathy can be obtained. The changes in the features of the fundus images of the patients with diabetic retinopathy are analyzed and processed. Throughout the above process, the patients with diabetic retinopathy can obtain their own fundus image at any place where the fundus image obtaining device is installed, and then upload the corresponding fundus image to the fundus image processing terminal which will store the fundus images, and automatically compare and analyze the fundus images and their feature data of patients with diabetic retinopathy at different periods, to obtain the changes in the features of the fundus image. And the changes in the features of the fundus images of the patients with diabetic retinopathy are analyzed and processed. The whole process is high in efficiency, low in cost, and good in the experience of the patient with diabetic retinopathy. Furthermore, the data after analysis and processing can be sent to the patients with diabetic retinopathy themselves, so that the patients with diabetic retinopathy can understand the health status of their own ffindus. Furthermore, the professional medical staff can give the patients with diabetic retinopathy better treatment recommendations based on the analysis results.

In the embodiment, it also includes the steps of generating an analysis report and sending the analysis report to a mobile terminal device of the patient with diabetic retinopathy. The information related to the analysis report includes one or more of an analysis report, an analysis report download link, and a recommendation report. By generating the analysis report and sending it to the patient with diabetic retinopathy, the patient with diabetic retinopathy can know the result at the first time, so that the patient with diabetic retinopathy can adjust their living habits or do further treatment or observation, etc., effectively helping the patient with diabetic retinopathy to control their diabetes well and in real time.

Please refer to FIG. 2, in the embodiment, a specific embodiment of a storage device 200 is described as follows: A storage device 20, storing an instruction set, wherein the instruction set is used to execute steps of: obtaining a fundus image of a patient with diabetic retinopathy by a fundus screening; extracting and identifying feature data of the fundus image, storing the feature data of the fundus image; determining whether previous feature data of the fundus image of the patient with diabetic retinopathy is stored, and if previous feature data of the fundus image of the patient with diabetic retinopathy is stored, analyzing and comparing the feature data of the fundus images of the patient with diabetic retinopathy at different periods to obtain changes in the feature data of the patient with diabetic retinopathy in the fundus screening; analyzing and processing the changes in the feature data in the fundus screening.

The above-mentioned steps are described in detail as follows: The fundus image of the patient with diabetic retinopathy is obtained through the fundus camera of a primary health center (such as the primary medical institutions, the health checkups, the health management or primary community clinics), and the obtained fundus image is transmitted to the PC through the data cable and processed by the fundus image data analysis workstation software, or sent to the PC through the network, and the PC send the fundus image to the fundus image data analysis center. The patient with diabetic retinopathy can also upload the fundus image through the mobile terminal device. It should be noted that, in the embodiment, the primary health center may be located in a remote area where there is no professional ophthalmologist staff or where the cost of hiring professional ophthalmologists is very high.

After the fundus image data analysis center or the above-mentioned the fundus image data analysis workstation software obtains the fundus image of the patient with diabetic retinopathy, the feature data of the fundus image and storing the feature data of the fundus image is extracted, identified, and stored. The step is described in detail as follows. Through automatic or semi-automatic human-computer interactive method, the micro-angiomas, the bleeding points, and the hard exudates of fimdus image of patient with diabetic retinopathy are obtained and stored. The fundus image and the feature data thereof can be stored to a cloud server, a specific server, or a local storage device. The purpose of storing the fundus image and the feature data thereof is to facilitate the same patient with diabetic retinopathy comes to check or screen regularly, the currently obtained data can be compared with the previously obtained data each time for better data analysis.

After the feature data of the fundus image is stored, whether previous feature data of die fimdus image of the patient with diabetic retinopathy is stored is determined. If previous feature data of the fundus image of the patient with diabetic retinopathy is stored, the feature data of the fimdus images of the patient with diabetic retinopathy at different periods is analyzed and compared to obtain changes in the feature data of the fundus image of the patient with diabetic retinopathy-. The step is described in detail as follows. According to the patient's name and ID, check the database to see if the previously fundus image and its feature data of the patient with diabetic retinopathy is stored. If the previously fundus image and its feature data of the patient with diabetic retinopathy is stored, the fundus images and their feature data of patients with diabetic retinopathy at different periods are analyzed and compared, to obtain the changes in the feature data of the fimdus image The details are described as follows.

A brightness histogram equalization is established to preprocess the fundus image; a morphological filter is established to determine positions of a macula and a optic disc in the preprocessed fundus image; blood vessels of the preprocessed fundus image are segmented; the fundus image is aligned according to fundus parameters and the identification of the feature data of fundus image is corrected. The fiindus parameters include the position of the macula, the position of the optic disc, and the information of the blood vessel; the changes in the feature data of the thndus image are automatically analyzed.

The above-mentioned steps are described in detail as follows: In the image to be analyzed, a clear fundus image in which the optic disc, macula and blood vessels can be clearly observed is selected as the standard reference image, and the standard reference image is processed to generate a standard grayscale histogram. According to the grayscalc distribution of the standard grayscale histogram, the grayscalcs of the remaining fundus images to be analyzed are mapped to obtain a fundus image with the same grayscale distribution as the standard reference image.

According to the macular brightness and optic disc brightness in the preprocessed fundus image, the shapes of the macula and the optic disc, and the distance between the macula and optic disc, a morphological filter is established to determine the positions of the macula and optic disc. in other words, in the preprocessed fundus image, the macula has extremely low brightness, die optic disc has extremely high brightness, the shapes of the two tends to be circular, and the relative distance and position of the two are fixed, so as to establish the morphological filter. The circular areas with extremely low brightness and extremely high brightness in the fundus image are detected and used as the candidate areas of the macula and optic disc. According to the distance and position of the two, the wrong candidate areas are filtered out, and then die center positions of the macula and optic disc are determined.

In the preprocessed fundus image, the blood vessels of the fundus have similar gray-scale information and have a higher contrast with the background. The blood vessels can be segmented through the above features. in the embodiment, the blood vessels are segmented by using a threshold segmentation method. After segmenting die main blood vessels, according to the position of the macula, the position of the optic disc, and the information of the main blood vessels, the change area of the ftindus image is identified on the fundus image. Through the change area, the professionals can quickly see whether the number of microvascular tumors, bleeding spots, or hard exudation has increased, whether the scope of hard exudation has expanded or is getting closer to the macula.ln other embodiments, a rectangle is used to mark the feature data of the relevant fundus image in the ftindus image: micro-angioma area, bleeding point area, hard exudation area (at the same time, record die size of these feature areas and their relative position to the fovea in the database). Different colors represent different features and areas of diabetic retinopathy. For example, white represents hard exudation, pink represents micro-angioma, and green represents bleeding points. Then, the fundus images are aligned according to the fundus parameters, the fundus parameters include: the position of the macula, the position of the optic disc, and the blood vessel information; and the change area of the fundus image is identified.

The step of analyzing and processing the changes in the feature data in the thndus screening can be executed as follows. The control effect of blood glucose level and the health status of the patient with diabetic retinopathy in a preset period are analyzed and calculated. A corresponding health service recommendation based on the analysis results is given. A report of the control effect of blood glucose level, the health status and the health service recommendation is generated, and the report is sent to the mobile terminal device of the patient with diabetic retinopathy.

The details can be as follows: After data processing and analysis, the obtained data can be sent directly to the patients with diabetic retinopathy, so that the patients with diabetic retinopathy themselves can better understand their own health conditions. It can also be sent to a professional medical institution with the consent of the patients with diabetic retinopathy, assisting the medical staff to quickly understand the condition of the patients with diabetic retinopathy, and provide the follow-up disease control recommendations for the patients with diabetic retinopathy. In this embodiment, the results of data processing and analysis include: whether the number or size of micro-angiomas is increased or not, whether the area of hard exudation is enlarged or involves die area of the macular. If possible, it can be referred for auxiliary judgments with the relevant medical inquiry data and other physical index data sent by the primary health center with the fundus image, and the data is such as a significant increase or decrease in body weight and waist circumference, diet, exercise, no smoking and less alcohol, etc. if there is an increase in the number of micro-angiomas and the area of hard exudation, it indicates that the control of blood glucose level during this period is poor, and the condition of retinopathy is still developing, and thither control is needed to ensure a good lifestyle. If the number of the micro-angiomas and the bleeding points increases significantly, the area of hard exudation expands and affects the macular area, it is recommended to do further follow-up diagnosis and treatment. If there is no obvious change in micro-angioma and hard exudation, it indicates that the control level is good, and a good lifestyle and basic treatment should be continued in accordance with the doctor's advice. In particular, when the number of micro-angiomas decreases or some micro-angiomas disappear, it means that the blood supply of the capillaries in the area is severely insufficient or disappears. When the extent of hard exudation expands or approaches the macular area, it means that the symptoms of macular edema are serious or there is a danger of blindness. Special attention should be paid or it is recommended to do further review or go to the hospital for examination. The augmented reality (AR) technology can be used to make simple demonstration animations of changes in these fundus features and conditions that may affect vision or general health if they continue to develop, and superimpose them on real fundus photos to achieve visual education effects, and encourage the patients to live a good lifestyle with basic treatment and have awareness of timely screening and timely preventive treatment.

The following steps are executed through the instruction set on the storage device 200. After obtaining the fundus images of patients with diabetic retinopathy at different periods, the fundus images and the feature data of the fundus images after structured processing is stored. Through automatic analysis and comparison of features of the fundus images at different periods, the changes in the features of the fundus images of the patients with diabetic retinopathy can be obtained. The changes in the features of the fundus images of the patients with diabetic retinopathy are analyzed and processed. Throughout the above process, the patients with diabetic retinopathy can obtain their own fundus image at any place where the fundus image obtaining device is installed, and then upload the corresponding fundus image to the fundus image processing terminal which will store the fundus images, and automatically compare and analyze the fundus images and their feature data of patients with diabetic retinopathy at different periods, to obtain the changes in the features of the fundus image. And the changes in the features of the fundus images of the patients with diabetic retinopathy are analyzed and processed. The whole process is high in efficiency, low in cost, and good in the experience of the patient with diabetic retinopathy. Furthermore, the data after analysis and processing can be sent to the patients with diabetic retinopathy themselves, so that the patients with diabetic retinopathy can understand the health status of their own fundus. Furthermore, the professional medical staff can give the patients with diabetic retinopathy better treatment recommendations based on the analysis results.

Please refer to FIG. 2, in the embodiment, a specific embodiment of a system 300 for analyzing changes in feature data of fundus of diabetic retinopathy is described as follows: A system 300 for analyzing changes in feature data of fundus of diabetic retinopathy includes a fundus image obtaining terminal 301 and a fundus image processing terminal 302. The fundus image processing terminal 302 includes a data storage module 3021, a fundus image analysis and comparison module 3022 and a result analysis module 3023. The fundus image obtaining terminal 301 is connected to the fundus image processing terminal 302. The fundus image obtaining terminal 301 is used for obtaining a fundus image of a patient with diabetic retinopathy and sending the fundus image to the fundus image processing terminal 302. The fundus image processing terminal 302 is used for extracting and identifying feature data of the fundus image. The data storage module 3021 is used for storing the feature data of the fimdus image. The fundus image analysis and comparison module 3022 is used for determining whether previous feature data of the fundus image of the patient with diabetic retinopathy is stored; and if previous feature data of the fundus image of the patient with diabetic retinopathy is stored, analyzing and comparing the feature data of the fundus images of the patient with diabetic retinopathy at different periods to obtain changes in the feature data of the patient with diabetic retinopathy in the fundus screening. The result analysis module 3023 is used for analyzing and processing the changes in the feature data in the fundus screening In the embodiment, the fundus image obtaining terminal 301 includes a fimdus image acquisition terminal and PC, and the fundus image acquisition terminal is a fundus camera in some embodiment. The fundus image processing terminal 302 includes but is not limited to: a personal computer, a server, a general-purpose computer, and a dedicated computer, a network equipment, intelligent mobile terminals, smart home devices, etc. The details are as follows: The fundus image of the patient with diabetic retinopathy is obtained through the fiindus camera and the obtained fundus image is transmitted to the PC through the data cable and processed by thc fiindus image data analysis workstation software, or sent to the PC through the network, and the PC send the fiindus image to the fundus image data analysis center. The patient with diabetic retinopathy can also upload the fundus image through the mobile terminal device. It should be noted that, in the embodiment, the primary health center may be located in a remote area where there is no professional ophthalmologist staff or where the cost of hiring professional ophthalmologists is very high.According to the patient's name and ID, check the data storage module 3021 to see if the previously fundus image and its feature data of the patient with diabetic retinopathy is stored. If the previously fundus image and its feature data of the patient with diabetic retinopathy is stored. The fundus images and their feature data of patients with diabetic retinopathy at different periods are analyzed and compared by the fundus image analysis and comparison module 3022, to obtain the changes in the feature data of the fundus image.

The fundus image analysis and comparison module 3022 is further used for: establishing brightness histogram equalization to preprocess the fundus image; establishing a morphological filter to determine positions of a macula and a optic disc in the preprocessed fiindus image; segmenting blood vessels of the preprocessed fundus image; aligning the fimdus image according to fundus parameters and correcting the identification of the feature data of fundus image, wherein the fundus parameters include the position of the macula, the position of the optic disc, and the information of the blood vessel; automatically analyzing the changes in the feature data of the fundus image. The details are as follows: In the image to be analyzed, a clear fundus image in which the optic disc, macula and blood vessels can be clearly observed is selected as the standard reference image, and the standard reference image is processed to generate a standard grayscale histogram. According to the grayscale distribution of the standard grayscale histogram, the grayscales of the remaining fundus images to be analyzed are mapped to obtain a fundus image with the same grayscale distribution as the standard reference image.

According to the macular brightness and optic disc brightness in the preprocessed fundus image, the shapes of the macula and the optic disc, and the distance between the macula and optic disc, a morphological filter is established to determine the positions of the macula and optic disc. In other words, in the preprocessed fundus image, the macula has extremely low brightness, the optic disc has extremely high brightness, the shapes of the two tends to be circular, and the relative distance and position of the two are fixed, so as to establish the morphological filter. The circular areas with extremely low brightness and extremely high brightness in the fundus image are detected and used as the candidate areas of the macula and optic disc. According to the distance and position of the two, the wrong candidate areas are filtered out, and then the center positions of the macula and optic disc arc determined.

In the preprocessed fundus image, the blood vessels of the fundus have similar gray-scale information and have a higher contrast with the background. The blood vessels can be segmented through the above features. In the embodiment, the blood vessels are segmented by using a threshold segmentation method. After segmenting the main blood vessels, according to the position of the macula, the position of the optic disc, and the information of the main blood vessels, the change area of the fundus image is identified on the fundus image. Through the change area, the professionals can quickly see whether the number of microvascular tumors, bleeding spots, or hard exudation has increased, whether the scope of hard exudation has expanded or is getting closer to the macula.

In other embodiments, a rectangle is used to mark the feature data of the relevant fundus image in the fundus image: micro-angioma area, bleeding point area, hard exudation area (at the same time, record the size of these feature areas and their relative position to the fovea in the database). Different colors represent different features and areas of diabetic retinopathy. For example, white represents hard exudation, pink represents micro-angioma, and green represents bleeding points. Then, the fundus images are aligned according to the fundus parameters, the fimdus parameters include: the position of the macula, the position of the optic disc, and the blood vessel information; and the change area of the fundus image is identified.

In the embodiment, the result analysis module 3023 is further used for: Analyzing and calculating control effect of blood glucose level and the health status of the patient with diabetic retinopathy in a preset period giving a corresponding health service recommendation based on the analysis results; generating a report of the control effect of blood glucose level, the health status and the health service recommendation, and sending the report to the mobile terminal device of the patient with diabetic retinopathy.

The details can be as follows: After data processing and analysis, the obtained data can be sent directly to the patients with diabetic retinopathy, so that the patients with diabetic retinopathy themselves can better understand their own health conditions. It can also be sent to a professional medical institution with the consent of the patients with diabetic retinopathy, assisting the medical staff to quickly understand the condition of the patients with diabetic retinopathy, and provide the follow-up disease control recommendations for the patients with diabetic retinopathy. In this embodiment, the results of data processing and analysis include: whether the number or size of micro-angiomas is increased or not, whether the area of hard exudation is enlarged or involves the area of the macular. If possible, it can be referred for auxiliary judgments with the relevant medical inquiry data and other physical index data sent by the primary health center with the fundus image, and the data is such as a significant increase or decrease in body weight and waist circumference, diet, exercise, no smoking and less alcohol, etc. if there is an increase in the number of micro-angiomas and the area of hard exudation, it indicates that the control of blood glucose level during this period is poor, and the condition of retinopathy is still developing, and further control is needed to ensure a good lifestyle. If the number of the micro-angiomas and the bleeding points increases significantly, the area of hard exudation expands and affects the macular area, it is recommended to do further follow-up diagnosis and treatment. If there is no obvious change in micro-angioma and hard exudation, it indicates that the control level is good, and a good lifestyle and basic treatment should be continued in accordance with the doctor's advice, in particular, when the number of micro-angiomas decreases or some micro-angiomas disappear, it means that the blood supply of the capillaries in the area is severely insufficient or disappears. When the extent of hard exudation expands or approaches the macular area, it means that the symptoms of macular edema are serious or there is a danger of blindness. Special attention should be paid or it is recommended to do further review or go to the hospital for examination. The augmented reality (AR) technology can be used to make simple demonstration animations of changes in these fundus features and conditions that may affect vision or general health if they continue to develop, and superimpose them on real fundus photos to achieve visual education effects, and encourage the patients to live a good lifestyle with basic treatment and have awareness of timely screening and timely preventive treatment.

After the fundus images of patients with diabetic retinopathy at different periods are obtained automatically by the fundus image obtaining terminal 301, the fundus images is stored by the data storage module 3021. Through automatic analysis and comparison of features of the fundus images at different periods by the fundus image analysis and comparison module 3022, the changes in the features of the fundus images of the patients with diabetic retinopathy can be obtained. The changes in the features of the fundus images of the patients with diabetic retinopathy are analyzed and processed by the result analysis module 3023. Throughout the above process, the patients with diabetic retinopathy can obtain their own fundus image at any place where the fundus image obtaining device is installed, and then upload the corresponding fundus image to the fundus image processing tenninal 302 which will store the fundus images, and automatically compare and analyze the fundus images and their feature data of patients with diabetic retinopathy at different periods. The whole process of data analyzing and processing is high in efficiency, low in cost, and good in the experience of the patient with diabetic retinopathy. Furthermore, the data after analysis and processing can be sent to the patients with diabetic retinopathy themselves, so that the patients with diabetic retinopathy can understand the health status of their own fundus. Furthermore, the professional medical staff can give the patients with diabetic retinopathy better treatment recommendations based on the analysis results.

In the embodiment, for the analysis results, the result analysis module 3023 is further used for generating an analysis report and sending the analysis report to a mobile terminal device of the patient with diabetic retinopathy. The information related to the analysis report includes one or more of an analysis report, an analysis report download link, and a recommendation report. By generating the analysis report and sending it to the patient with diabetic retinopathy, the patient with diabetic retinopathy can know the result at the first time, so that the patient with diabetic retinopathy can adjust their living habits or do further treatment or observation, etc., effectively helping the patient with diabetic retinopathy to control their diabetes well and in real time.

Claims

Claims 1. A method for analyzing changes in feature data of fundus of diabetic retinopathy, comprising steps of obtaining a fundus image of a patient with diabetic retinopathy; extracting and identifying feature data of the fundus image; storing the feature data of the fundus image; determining whether previous feature data of the fundus image of the patient with diabetic retinopathy is stored, and if previous feature data of the fundus image of the patient with diabetic retinopathy is stored, analyzing and comparing the feature data of the fundus images of the patient with diabetic retinopathy at different periods to obtain changes in the feature data of die fundus image of the patient with diabetic retinopathy in a fundus screening; analyzing and processing the changes in the feature data in the fundus screening.
2. The method for analyzing changes in feature data of fundus of diabetic retinopathy according to claim 1, wherein the step of "analyzing and comparing the feature data of die fundus image of the patient with diabetic retinopathy at different periods to obtain changes in the feature data of the patient with diabetic retinopathy in the fimdus screening" further comprises steps of establishing brightness histogram equalization to preprocess the fundus image; establishing a morphological filter to determine positions of a macula and a optic disc in the preprocessed fundus image; segmenting blood vessels of the preprocessed fundus image; aligning the fundus image according to fundus parameters and correcting the identification of the feature data of fimdus image, wherein the fundus parameters include the position of die macula, the position of the optic disc, and the information of the blood vessel; automatically analyzing die changes in die feature data of die fundus image.
3. The method for analyzing changes in feature data of fundus of diabetic retinopathy according to claim I, wherein the step of "analyzing and processing the changes in the feature data in the fundus screening" further comprises steps of: analyzing and calculating control effect of blood glucose level and the health status of the patient with diabetic retinopathy in a preset period; giving a corresponding health service recommendation based on the analysis results; generating a report of the control effect of blood glucose level, the health status and the health service recommendation, and sending the report to a mobile terminal device of the patient with diabetic retinopathy.
4. The method for analyzing changes in feature data of fundus of diabetic retinopathy according to claim 1, wherein the "identifying feature data of the fundus image" further comprises steps of: marking a micro-angiomas and its relative position to a fovea; marking size of a bleeding point and its relative position to the fovea; identifying and analyzing the range of hard exudate and its minimum distance from the fovea.
5. A storage device, storing an instruction set, wherein the instruction set is used to execute steps of obtaining a fundus image of a patient with diabetic retinopathy; extracting and identifying feature data of the fundus image; storing the feature data of the thndus image; determining whether previous feature data of the fundus image of the patient with diabetic retinopathy is stored, and if previous feature data of the fundus image of the patient with diabetic retinopathy is stored, analyzing and comparing the feature data of the fundus images of the patient with diabetic retinopathy at different periods to obtain changes in the feature data of the patient with diabetic retinopathy in a fundus screening; analyzing and processing the changes in the feature data in the fundus screening.
6. The storage device according to claim 5, wherein the instruction set is further used to execute following steps: the step of "analyzing and comparing the feature data of the fundus image of the patient with diabetic retinopathy at different periods to obtain changes in the feature data of the patient with diabetic retinopathy in the fundus screening further comprises steps of establishing brightness histogram equalization to preprocess the fundus image; establishing a morphological filter to determine positions of a macula and a optic disc in the preprocessed fundus image; segmenting blood vessels of the preprocessed finadus image; aligning the fimdus image according to fiindus parameters and correcting the identification of the feature data of fimdus image, wherein the fundus parameters include the position of the macula, the position of the optic disc, and the information of the blood vessel; automatically analyzing the changes in the feature data of the fundus image.
7. The storage device according to claim 5, wherein the instruction set is further used to execute following steps: the step of "analyzing and processing the changes in the feature data in the fundus screening" further comprises steps of analyzing and calculating control effect of Hood glucose level and the health status of the patient with diabetic retinopathy in a preset period; giving a corresponding health service recommendation based on the analysis results; generating a report of the control effect of blood glucose level, the health status and the health service recommendation, and sending the report to the mobile terminal device of the patient with diabetic retinopathy.
8. The storage device according to claim 5, wherein the instruction set is further used to execute following steps: the "identifying feature data of the fundus image" further comprises steps of: marking a micro-angiomas and its relative position to a fovea; marking size of a bleeding point and its relative position to the fovea; identifying and analyzing the range of hard exudate and its minimum distance from the fovea.
9. A system for analyzing changes in feature data of fundus of diabetic retinopathy, comprising a fundus image obtaining terminal and a fundus image processing terminal, the fundus image processing terminal comprising a data storage module a fundus image analysis and comparison module and a result analysis module, wherein the fimdus image obtaining terminal is connected to the fundus image processing terminal; the fundus image obtaining terminal is used for obtaining a fundus image of a patient with diabetic retinopathy; the fundus image processing terminal is used for extracting mid identifying feature data of the fundus image; the data storage module is used for storing the feature data of the fundus image; the fundus image analysis and comparison module is used for determining whether previous feature data of the fundus image of the patient with diabetic retinopathy is stored, and if previous feature data of the fundus image of the patient with diabetic retinopathy is stored, analyzing and comparing the feature data of the fundus images of the patient with diabetic retinopathy at different periods to obtain changes in the feature data of the patient with diabetic retinopathy in a fundus screening; the result analysis module is used for analyzing and processing the changes in the feature data in the fundus screening.
10. The system for analyzing changes in feature data of fundus of diabetic retinopathy according to claim 9, wherein the fundus image analysis and comparison module is further used for establishing brightness histogram equalization to preprocess the fundus image; establishing a morphological filter to determine positions of a macula and a optic disc in the preprocessed fundus image; segmenting blood vessels of the preprocessed fundus image; aligning the fimdus image according to fimdus parameters and correcting the identification of the feature data of fundus image, wherein the fimdus parameters include the position of the macula, the position of the optic disc, and the information of the blood vessel; automatically analyzing the changes in the feature data of the fundus image; the fundus image obtaining terminal is further used for marking a micro-angiomas and its relative position to a fovea; marking size of a bleeding point and its relative position to the fovea; identifying and analyzing the range of hard exudate and its minimum distance from the fovea.