CN113243887A - Intelligent diagnosis and treatment instrument for macular degeneration of old people - Google Patents
Intelligent diagnosis and treatment instrument for macular degeneration of old people Download PDFInfo
- Publication number
- CN113243887A CN113243887A CN202110803738.1A CN202110803738A CN113243887A CN 113243887 A CN113243887 A CN 113243887A CN 202110803738 A CN202110803738 A CN 202110803738A CN 113243887 A CN113243887 A CN 113243887A
- Authority
- CN
- China
- Prior art keywords
- diagnosis
- macular degeneration
- treatment
- module
- age
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B3/00—Apparatus for testing the eyes; Instruments for examining the eyes
- A61B3/10—Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions
- A61B3/102—Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions for optical coherence tomography [OCT]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B3/00—Apparatus for testing the eyes; Instruments for examining the eyes
- A61B3/10—Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions
- A61B3/12—Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions for looking at the eye fundus, e.g. ophthalmoscopes
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B3/00—Apparatus for testing the eyes; Instruments for examining the eyes
- A61B3/10—Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions
- A61B3/12—Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions for looking at the eye fundus, e.g. ophthalmoscopes
- A61B3/1225—Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions for looking at the eye fundus, e.g. ophthalmoscopes using coherent radiation
- A61B3/1233—Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions for looking at the eye fundus, e.g. ophthalmoscopes using coherent radiation for measuring blood flow, e.g. at the retina
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B3/00—Apparatus for testing the eyes; Instruments for examining the eyes
- A61B3/10—Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions
- A61B3/12—Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions for looking at the eye fundus, e.g. ophthalmoscopes
- A61B3/1241—Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions for looking at the eye fundus, e.g. ophthalmoscopes specially adapted for observation of ocular blood flow, e.g. by fluorescein angiography
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0059—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence
- A61B5/0062—Arrangements for scanning
- A61B5/0066—Optical coherence imaging
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/145—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
Abstract
The invention belongs to the field of ophthalmologic diagnosis and treatment equipment, and particularly relates to an intelligent diagnosis and treatment system for senile macular degeneration. The invention discloses an intelligent diagnosis and treatment system for age-related macular degeneration, which comprises a detection module, a wireless transmission module, a feature extraction module and a diagnosis and treatment module, wherein the detection module comprises an eye fundus OCT (optical coherence tomography), an eye fundus fluorescein angiography FFA, a choroid angiography, a central vision detection, a blood sugar detection and a blood pressure detection, the feature extraction module automatically extracts features from an eye fundus image, the data obtained by the detection module is transmitted to the diagnosis and treatment module through the wireless transmission module, the diagnosis and treatment module comprises a database, a GA-BP (genetic algorithm-BP) neural network diagnosis model and an interface display module, a personal data file of a patient is further established, the disease data can be dynamically tracked, a targeted treatment scheme can be given, and the delay of the disease condition is avoided. The system has the effects of low misdiagnosis rate, no dependence on manual work, near detection and capability of realizing dynamic tracking of patient data.
Description
Technical Field
The invention belongs to the field of ophthalmologic diagnosis and treatment equipment, and particularly relates to an intelligent diagnosis and treatment instrument for macular degeneration of the elderly.
Background
Along with the rapid development of social informatization, the popularity and the utilization rate of electronic equipment are increasing day by day. People rely on electronic screen equipment such as computers and mobile phones in work and life, so that the incidence of a plurality of ophthalmic diseases is higher and higher, and the diagnosis and treatment of eye diseases are more and more emphasized. Age-related Macular Degeneration (AMD) among retinal diseases has become one of the major diseases affecting the health and quality of life of the elderly, and is the leading cause of common blindness above Age 45, clinically AMD is divided into two types, namely exudative (also known as wet) and non-exudative (also known as collapsed), the exudative clinical features being Macular Choroidal Neovascularization (CNV), serous Retinal Pigment Epithelium (RPE) detachment, exudation, bleeding, and scarring. The retina pathological image is an important standard for doctors to diagnose the stage of eye diseases, the normal fundus image mainly comprises structures such as artery blood vessels, vein blood vessels, yellow spots, optic discs and the like, and the common retina OCT image has three acquisition modes: the center of the macula, the center of the optic nerve head, and a large visual field including the first two regions. In order to realize a high-performance automatic computer-aided diagnosis system, it is necessary to firstly perform acquisition region and normal identification and classification on a retina OCT image, so that subsequent segmentation and classification of the retina OCT image can be more efficient. Accurate classification of retinal images is an important basis for physicians to develop optimal treatment regimens. The existing diagnosis and treatment methods are to judge the pathological change degree manually according to visual inspection retina images, the diagnosis depends on the experience of doctors, the difference between different stages of the retina images is very small, the classification is often unclear, misdiagnosis is easy to cause or effective treatment can not be carried out according to personal difference.
The observation of fluorescence angiography of the vision, fundus and fundus oculi was studied in reference 1 (Long-term follow-up of wet age-related macular degeneration and electronic computer image measurement, Zhang Cheng fen, et al, J.Zhong.EYE-BII.D., 1994(01): 1-3). Changes in subretinal neovascular membranes and pigment epithelium were measured using an electronic computer image analyzer. The method simply measures and compares 20 cases of images of initial diagnosis and follow-up diagnosis, and qualitatively analyzes the area of retinal neovascular membrane, but the research is in the initial stage of image analysis, does not consider how to segment and extract features, and compares the features with big data.
Reference 2 (indocyanine green angiography image characterization of exudative age-related macular degeneration, Chensong et al, ophthalmic research 2003(04): 428: -430) discusses image characterization of exudative age-related macular degeneration (AMD) by comparison of indocyanine green angiography (ICGA) with Fluorescein Fundus Angiography (FFA). Methods fundus colorphotography, FFA and ICGA examinations were performed on 52 65 eye exudation type AMD patients. Results in 65 eyes of exudative AMD, ICGA diagnosed as typical Choroidal Neovascularization (CNV) had 33 eyes accounting for 50.8%, FFA diagnosed as typical CNV had 8 eyes accounting for 11.6%, FFA diagnosed as occult CNV had 35 eyes, ICGA diagnosed as well-bounded or poorly-defined plaque CNV had 22 eyes, 39 eyes with macular bleeding combined, CNV not found by FFA but 5 eyes found by ICGA, FFA diagnosed as scar-stained 7 eyes, 2 eyes found CNV in ICGA, and ICGA found supply blood vessels (feeding vessels) of CNV had 3 eyes. Conclusion ICGA has a higher diagnostic rate than FFA-found CNVs, accurately shows CNVs masked by macular hemorrhage, and contributes to the discovery of blood supply vessels for CNVs. It also indicates that ICGA still has 50% insidious CNV and the typical CNV is not clearly found, and some lesions appear less clearly than FFA. The method mainly discusses the influence of two contrast methods on the diagnosis rate of exudative AMD, does not consider how to utilize artificial intelligence technology to segment and extract features, compares the features with big data, and gives a diagnosis suggestion in a targeted manner.
Document 3 (age-related macular degeneration frequency domain optical coherence tomography segmentation study, zhangqiao, et al, "university of Hunan university (Nature science edition), 44 vol 10, p 10, 24-: a method for segmenting an aged macular degeneration retina frequency domain optical coherence tomography image is provided. Secondly, a curvature calculation method is adopted to detect drusen, and finally, Kalman filtering is utilized to carry out rapid and accurate segmentation according to the correlation between adjacent edge points. And finally, the speed and the accuracy are higher than those of the prior art. The method mainly focuses on improving the efficiency and accuracy of the segmentation technology, does not carry out big data research on actual AMD diagnosis and treatment, and does not consider how to carry out treatment aiming at different disease classifications.
Document 4 (medical image diagnosis review based on deep learning, zhangqiao, etc., 'computer science', volume 44, phase 11A, month 11 in 2017, pages 1-6) can see that similarity exists in the model or method using deep learning at present, and CNN or other commonly used deep learning algorithms or a mode of fusing several algorithms are mostly adopted to perform image classification detection by performing deep learning diagnosis analysis on the medical images of the above diseases; and most models are in the theoretical stage at present and have not been applied to clinics. The unsupervised learning method does provide great convenience for image classification, for example, features can be extracted without manual work, misdiagnosis caused by personal factors of doctors is avoided, but at present, research of the method is mainly focused on disease research with high morbidity, namely large amount of patient data, and research of some rare diseases is less.
Chinese patent application No.: CN 201710793834.6 discloses a method for segmenting map-like atrophic GA lesions of an SD-OCT image based on a depth voting model without depending on layer segmentation, which adopts a depth network model to express a complex data structure in three-dimensional data, breaks through the bottleneck of the traditional method for image layer segmentation dependence on the premise of a small number of training samples, breaks through the sensitivity of the traditional method for data of different sources, obtains a relatively ideal effect, greatly improves the segmentation precision of GA lesions, and has important practical significance for the prevention and diagnosis of age-related macular degeneration diseases. The deep network comprises five layers which are an input layer, a three-layer unsupervised Sparse self-encoder and an output layer (a Softmax classifier which distinguishes GA regions from non-GA regions); regarding three-dimensional data with the size of 512 multiplied by 128 multiplied by 1024, regarding each pixel point on a two-dimensional projection image with the size of 512 multiplied by 128 as a sample, wherein each sample has 1024-dimensional characteristics, and combining a marking result, randomly selecting one hundred thousand positive samples, namely GA lesion pixels, and one hundred thousand negative samples, namely normal tissue pixels, to form a training set of a model, thereby constructing the input of a depth network. The output classifies the depth features into two categories, GA lesions and non-GA lesions. Finally, a voting decision strategy is used to improve the segmentation results of the ten training models. The invention takes an SD-OCT retina image as input, expresses three-dimensional input through a depth network model, and generates a two-dimensional map-like atrophy lesion segmentation image. However, it only classifies GA and non-GA images and does not take into account the different categories of AMD, particularly typical exudative AMD, nor the differentiation of age-related macular degeneration treatment regimens by different patient-related physiological indices.
Chinese patent application No.: CN 202010790150.2 discloses a fundus oculi color image grading method and device, comprising: acquiring an original image, and performing enhancement processing on the original image to obtain a target image; carrying out color processing on the original image and the target image to respectively obtain a first processed image and a second processed image; and processing the first processing image and the second processing image by adopting a pre-trained grading model to obtain a target grading result. The grading label comprises light and medium age-related macular degeneration, severe age-related macular degeneration, light and medium sugar nets, severe sugar nets, leopard-striated fundus oculi and pathological myopia. The specific grading algorithm is that a first convolution network and a second convolution network are respectively adopted for carrying out grading prediction on a first processed image and a second processed image, then the first processed image and the second processed image are subjected to characteristic fusion and then are subjected to grading prediction, and finally grading results obtained by the grading predictions are fused to obtain a target grading result. However, the classification process is complex, the classification is carried out on the whole image scale of the fundus image, the processing efficiency is low, the noise is more, the types of the models are more, the model training precision is to be verified, the classification label does not consider the differentiation of dry or wet age-related macular degeneration, and the differentiation of related physiological indexes of different patients on age-related macular degeneration treatment schemes is not further considered.
Chinese patent application No.: CN 201811516615.4 discloses a retina OCT image classification method based on a three-dimensional convolution neural network, which adopts an improved Vinceptinc 3D network based on a C3D convolution neural network, can automatically classify the normal/abnormal three-dimensional OCT images of the macular center, the optic nerve head center and the large-visual-field retina, and lays a foundation for improving the efficiency of subsequent retina OCT image segmentation and analysis. It is only to preliminarily classify the images by six, but does not consider how to classify the lesion degree of the fundus image and how to give a proper diagnosis opinion according to the characteristics of different people.
Patent publication No.: CN106488738A discloses a fundus imaging system comprising a retinal imaging step and a preliminary diagnosis step, wherein example outputs of an evaluation algorithm are the presence of small red dots, the presence of hemorrhages and the detection of hard exudates in the fundus when using the fundus image as input. The output identifies the diagnostic list and the number of tissues observed. But does not consider how to classify fundus images and give appropriate diagnostic opinions.
Chinese patent application No.: CN201910442076.2 discloses an intelligent auxiliary diagnosis system for fundus laser surgery, which comprises a laser image stabilization and treatment device 1, a data control device 2 and an image display device 3, wherein the data control device 2 comprises a laser control module 21, an imaging control module 22 and an image data acquisition module 23; the fundus images are processed and analyzed by the data processing device 4, for example, disease feature data in the fundus images are extracted by the feature extraction module 42, comparison operation is performed by the data analysis matching module 45, the fundus images are matched with the disease feature data stored in the known case feature template library 44, the result of the matching operation is stored in the second database 43, if the matching degree exceeds a set threshold value, a corresponding auxiliary diagnosis conclusion is given, and then an auxiliary diagnosis report is generated by the diagnosis report generation module 46. The main contents of the auxiliary diagnosis report comprise preoperative diagnosis schemes, intraoperative target determination schemes, postoperative treatment effect prediction results and the like. Preferably, the deep learning module 47 is further included for performing a large amount of data training based on the collected fundus image data of the patient in combination with disease feature data extracted from the fundus image, and providing a matching operation result for reference of medical experts by automatically performing a data analysis matching operation (using a data fuzzy matching algorithm). The specific cases are diabetic retinal degeneration, senile macular degeneration and the like. The fundus image information is extracted and compared according to the data obtained by the imaging system, relevant characteristics of diseases of patients are not considered, the disease can be diagnosed and diagnosis opinions can be given out only by comprehensively summarizing and adopting a data matching algorithm, the algorithm and the matching process are not disclosed, laser treatment is mainly carried out, and differences of treatment modes given out in the development stage of the disease are not considered.
Aiming at the problems that in the prior art, an eye fundus image is photographed by an independent detection device or an OCT (optical coherence tomography) segmentation technology or different fluorescence contrast researches are basically carried out, the analysis results are specific diagnosis conclusions and treatment schemes obtained after the analysis is carried out by a doctor, a small part of researches can simply match the results obtained by an imaging system with a template library to diagnose whether the characteristics accord with the characteristics of the age-related macular degeneration or not, a specific method is not provided, the model accuracy is difficult to verify, individual special cases cannot be covered, a plurality of possible influence factors of AMD such as heredity, immunologic abnormality, hypertension and the like are considered, the disease characteristics and the treatment mode of a patient are difficult to accurately consider from a fundus picture, the influence of an individual on the disease development along with the age is difficult to consider, and the prior art mainly relies on the experience of the doctor for diagnosis, the method is easy to misdiagnose or inaccurate in judgment and the like, is often confused with senile drusen, choroidal melanoma, central exudative chorioretinopathy or traumatic chorioretinopathy, and does not have intelligent diagnosis and treatment equipment specially aiming at the senile macular degeneration in the prior art.
The intelligent diagnosis and treatment system for the senile macular degeneration can extract the focus characteristics according to the image obtained by the detection module, form standard data, directly grade the senile macular degeneration diseases of the user after the diagnosis and treatment module analyzes and processes the standard data and the related detection and manually input data, and directly generate a specific treatment scheme. The eye examination can thus be carried out completely off the doctor, even at home, and the user can follow a targeted treatment plan according to the individual course of the disease. The intelligent diagnosis and treatment system for the age-related macular degeneration can automatically extract focus characteristics from a detection image, perform grading through a diagnosis model and display a corresponding treatment scheme through an interface, supports the detection of a patient in a nearby medical institution, uploads data through wireless transmission, avoids inconvenience caused by medical resource shortage, establishes a personal file of the patient, can dynamically track disease data and give a targeted treatment scheme, and avoids delaying the disease. The system has the advantages of high automation degree, low misdiagnosis rate, no dependence on manual work and capability of realizing dynamic tracking of patient data and rapidly providing a targeted treatment scheme.
Disclosure of Invention
In order to solve the technical problems, the invention provides an intelligent diagnosis and treatment system for age-related macular degeneration, which consists of a detection module, a wireless transmission module, a feature extraction module and a diagnosis and treatment module, wherein the detection module comprises fundus OCT detection, fundus fluorescein angiography FFA, choroidal angiography, central vision detection, blood sugar detection and blood pressure detection, the detection module comprises fundus OCT, fundus fluorescein angiography FFA, choroidal angiography, central vision detection, blood sugar detection and blood pressure detection, the feature extraction module automatically extracts features from the OCT, fundus fluorescein angiography FFA and choroidal angiography, the wireless transmission module transmits data obtained by the detection module to the feature extraction module and guides the feature extraction data into the diagnosis and treatment module, and the diagnosis and treatment module comprises a database, a GA-BP neural network diagnosis model and an interface display module, and a patient personal data file is also established, and the disease data can be dynamically tracked to give a targeted treatment scheme.
Further, fundus OCT detection and choroidal angiography are respectively SD-OCT and indocyanine green angiography ICGA.
Further, the characteristic extraction module comprises preprocessing, image segmentation and characteristic extraction of fundus OCT, fundus fluorescein angiography FFA, choroidal angiography respectively, wherein the preprocessing is median filtering preprocessing, the image segments yellow spots and blood vessel regions, the characteristic extraction obtains drusen, progressive pigment epithelium RPE atrophy or detachment, geographic atrophy, yellow spot choroidal neovascularization typicality CNV, yellow spot choroidal neovascularization occult CNV, serous retinal pigment epithelium RPE detachment, exudation, edema, hemorrhage and scar focus characteristics, focus characteristic vectors are extracted, and the focus characteristic vectors are used as part of input data of the diagnosis and treatment module.
Further, the diagnosis step of the GA-BP neural network diagnosis model is as follows: the GA-BP neural network diagnosis model is trained through a large number of known fundus image samples, the images obtained by the detection module are subjected to feature extraction to obtain focus feature vector data and central vision, blood sugar and blood pressure data, the focus feature vector data and the central vision, blood sugar and blood pressure data are stored in a database, and the aged macular degeneration classification category is obtained through deduction through the trained GA-BP neural network diagnosis model.
Furthermore, a corresponding treatment scheme is given after the disease degree of the patient is diagnosed through a GA-BP neural network diagnosis model, the classification category of the age-related macular degeneration is divided into five grades, wherein the first grade is early dry age-related macular degeneration, the second grade is late dry age-related macular degeneration, the third grade is early wet age-related macular degeneration, the fourth grade is middle wet age-related macular degeneration, and the fifth grade is late wet age-related macular degeneration; the first-level treatment scheme is oral anti-oxidation drugs and vitamins, the second-level treatment scheme is oral anti-oxidation drugs, inflammation inhibiting drugs and low-intensity laser treatment, the third-level treatment scheme is anti-VEGF drugs injection and laser treatment or photodynamic therapy, the fourth-level treatment scheme is anti-VEGF drugs injection and surgical treatment, and the fifth-level treatment scheme is vitreous cutting surgery and oral vitamins.
Furthermore, the database also comprises a knowledge rule formed by regularizing the parameters acquired by the detection module and partially manually input, and storing the knowledge rule into the database.
Further, the partially manually entered parameters include patient age, disease age, and genetic factors, wherein the genetic factors are defined according to the age-related macular degeneration disease proportion in the third generation.
Further, optimizing the parameters of the BP neural network model by a global optimization algorithm GA, wherein the GA parameters are as follows: the population size is 22, the iteration times are 40, the cross probability is 0.42, and the variation probability is 0.20; the BP neural network employs a three-layer network model, namely an input layer, a hidden layer, and an output layer.
Further, in the GA-BP neural network diagnosis model, the BP neural network adopts a three-layer structure with 15 input layer nodes, 25 hidden layer nodes and 1 output layer nodes.
Further, the interface display module comprises a diagnosis result display and a treatment scheme display.
The invention has the beneficial effects that:
(1) aiming at the characteristics of complicated and small difference of characteristics of macular degeneration fundus images focus of old people and numerous influencing factors, the prior art mainly adopts manual judgment to make misdiagnosis easily or to give accurate classification judgment difficultly, the equipment disclosed by the invention can integrate the traditional blood sugar detection, blood pressure monitoring, central vision detection and fundus detection equipment, greatly reduces the tedious, time-consuming and labor-consuming detection processes of patients, is beneficial to the patients to make a re-diagnosis on time, is beneficial to tracking the disease course development of the patients in time, and can give a targeted treatment scheme;
(2) data obtained by the detection module is subjected to feature extraction to obtain focus feature data, the focus feature data is imported into a database by combining blood sugar, blood pressure, central vision detection and part of manually input data of the detection module, the disease degree of a patient is diagnosed by a GA-BP neural network diagnosis model, and a corresponding treatment scheme is given, so that a doctor can be basically separated, even a user can automatically check eyes at home, the nearby detection can be realized by the wireless transmission module, the data can reach the system last time, the medical congestion is reduced, and the on-time re-diagnosis is facilitated;
(3) the characteristic extraction module is used for preprocessing the fundus image, segmenting the fundus image and automatically extracting characteristics, focus characteristic vectors can be automatically and quickly extracted and used as input layer data of a BP neural network, the GA-BP neural network diagnosis model automatically classifies the focus characteristic vectors, and compared with the manual classification, the automatic classification method is high in speed and small in error, and avoids the defects caused by insufficient experience of doctors in the classification process and delays in subsequent treatment; the diagnosis is comprehensively carried out by combining the age, blood sugar, blood pressure and genetic factors of different patients, a graded diagnosis result is given, and a targeted treatment scheme is given, so that the accuracy is higher compared with manual judgment or a neural network based on big data, the diagnosis is more comprehensive and scientific compared with the traditional method which mainly carries out diagnosis according to fundus images, and by establishing a personal file of the patient, the timely and effective treatment is facilitated, the delay of the illness state and the delay of the optimal treatment time are avoided; after the return visit, the treatment mode is dynamically adjusted according to the change condition of the disease characteristics of the patient.
Detailed Description
An intelligent diagnosis and treatment system for age-related macular degeneration is composed of a detection module, a wireless transmission module, a feature extraction module and a diagnosis and treatment module, wherein the detection module comprises fundus OCT detection, fundus fluorescein angiography FFA, choroidal angiography, central vision detection, blood sugar detection and blood pressure detection, the detection module comprises fundus OCT, fundus fluorescein angiography FFA, choroidal angiography, central vision detection, blood sugar detection and blood pressure detection, the features are automatically extracted from the OCT, fundus fluorescein angiography FFA and choroidal angiography by the feature extraction module, the data obtained by the detection module is transmitted to the feature extraction module by the wireless transmission module and the feature extraction data is led into the diagnosis and treatment module, the diagnosis and treatment module comprises a database, a GA-BP neural network diagnosis model and an interface display module, and a personal patient data document is established, can dynamically track the disease data and give a targeted treatment scheme.
Further, fundus OCT detection and choroidal angiography are respectively SD-OCT and indocyanine green angiography ICGA.
Further, the characteristic extraction module comprises preprocessing, image segmentation and characteristic extraction of fundus OCT, fundus fluorescein angiography FFA, choroidal angiography respectively, wherein the preprocessing is median filtering preprocessing, the image segments yellow spots and blood vessel regions, the characteristic extraction obtains drusen, progressive pigment epithelium RPE atrophy or detachment, geographic atrophy, yellow spot choroidal neovascularization typicality CNV, yellow spot choroidal neovascularization occult CNV, serous retinal pigment epithelium RPE detachment, exudation, edema, hemorrhage and scar focus characteristics, focus characteristic vectors are extracted, and the focus characteristic vectors are used as part of input data of the diagnosis and treatment module.
Further, the diagnosis step of the GA-BP neural network diagnosis model is as follows: the GA-BP neural network diagnosis model is trained through a large number of known fundus image samples, the images obtained by the detection module are subjected to feature extraction to obtain focus feature vector data and central vision, blood sugar and blood pressure data, the focus feature vector data and the central vision, blood sugar and blood pressure data are stored in a database, and the aged macular degeneration classification category is obtained through deduction through the trained GA-BP neural network diagnosis model.
Furthermore, a corresponding treatment scheme is given after the disease degree of the patient is diagnosed through a GA-BP neural network diagnosis model, the classification category of the age-related macular degeneration is divided into five grades, wherein the first grade is early dry age-related macular degeneration, the second grade is late dry age-related macular degeneration, the third grade is early wet age-related macular degeneration, the fourth grade is middle wet age-related macular degeneration, and the fifth grade is late wet age-related macular degeneration; the first-level treatment scheme is oral anti-oxidation drugs and vitamins, the second-level treatment scheme is oral anti-oxidation drugs, inflammation inhibiting drugs and low-intensity laser treatment, the third-level treatment scheme is anti-VEGF drugs injection and laser treatment or photodynamic therapy, the fourth-level treatment scheme is anti-VEGF drugs injection and surgical treatment, and the fifth-level treatment scheme is vitreous cutting surgery and oral vitamins.
Furthermore, the database also comprises a knowledge rule formed by regularizing the parameters acquired by the detection module and partially manually input, and storing the knowledge rule into the database.
Further, the partially manually entered parameters include patient age, disease age, and genetic factors, wherein the genetic factors are defined according to the age-related macular degeneration disease proportion in the third generation.
Further, optimizing the parameters of the BP neural network model by a global optimization algorithm GA, wherein the GA parameters are as follows: the population size is 22, the iteration times are 40, the cross probability is 0.42, and the variation probability is 0.20; the BP neural network employs a three-layer network model, namely an input layer, a hidden layer, and an output layer.
Further, in the GA-BP neural network diagnosis model, the BP neural network adopts a three-layer structure with 16 input layer nodes, 25 hidden layer nodes and 1 output layer nodes.
Further, the interface display module comprises a diagnosis result display and a treatment scheme display.
The operation process is as follows: first, initialization of knowledge rules is performed. The database regularizes the existing clinical results of the age-related macular degeneration which are proved by expert practice and scientific analysis, forms knowledge rules and stores the knowledge rules into the database, simultaneously performs sampling processing on the input knowledge rules, adopts forward reasoning to realize diagnosis results and targeted treatment schemes through certain algorithms, and displays the results to users.
The main data in the database include drusen, progressive pigmented epithelial RPE atrophy or detachment, geographic atrophy, macular choroidal neovascularization typicality CNV, macular choroidal neovascularization occupational CNV, serous retinal pigment epithelial RPE detachment, exudation, edema, hemorrhage, scar, age, central vision, genetic factors, blood sugar, blood pressure. Preprocessing fundus images, segmenting images and extracting characteristics by a characteristic extraction module through a large amount of clinical data which are authenticated by experts, wherein the preprocessing is median filtering preprocessing, the image segmentation adopts segmentation technology in the prior art, such as a segmentation method based on a convolutional neural network and the like, so as to segment drusen, progressive pigment epithelium atrophy or detachment, geographic atrophy, macular choroidal neovascularization typicality CNV, macular choroidal neovascularization occupational neovascularization CNV, serous retinal pigment epithelium RPE detachment, exudation, edema, hemorrhage and scar focus characteristics, extract focus characteristic vectors, establish a knowledge regularization table of a database, use the focus characteristic vectors as input layer data of a GA-BP neural network, train the neural network, adopt a forward reasoning strategy, match with knowledge rules according to data of a user through a detection unit and manually input data, and deducing a grading result belonging to the age-related macular degeneration through a GA-BP neural network diagnosis model and giving a targeted treatment scheme.
For example, a patient has central vision test of 0.5, 55 years old, age of age-related macular degeneration of 10 years, fasting glucose: 7.0mmol/L, blood pressure: 150/90mmHg, the genetic factor is 0.4, after obtaining the fundus SD-OCT test, fundus fluorescein angiography FFA, choroid angiography ICGA image, obtaining the focus characteristic vector through the characteristic extraction module, deducing to obtain the macular degeneration belonging to the third-level old age through the GA-BP neural network diagnosis model, the treatment scheme is injecting anti-VEGF medicine + laser treatment or photodynamic therapy; specifically, the method further comprises the steps that the first-level treatment scheme is oral anti-oxidation drugs and vitamins, the second-level treatment scheme is oral anti-oxidation drugs, inflammation inhibiting drugs and low-intensity laser treatment, the fourth-level treatment scheme is anti-VEGF drug injection and surgical treatment, and the fifth-level treatment scheme is vitreous cutting surgery and oral vitamins. .
The method also comprises the steps of establishing a complete personal file for the patient, effectively monitoring the development trend of the age-related macular degeneration, and enabling the patient to clearly know the condition of the patient so as to provide a more accurate treatment scheme.
The specific operation mechanism of the GA-BP neural network diagnosis model is as follows:
genetic Algorithm (GA) was proposed in 1962 by Holland of Michigan university in the United states according to the principle mechanism of Darwin 'excellence and disadvantage elimination, survival of the fittest', and it is a global optimization search Algorithm for seeking a better structure by randomly exchanging chromosome information in a population according to Mendelian Genetic variation theory. The general flow of the genetic algorithm is as follows: (1) determining a coding mode, an initial population (including population scale, selection probability, cross probability and variation probability) and a fitness function, and randomly generating the initial population; (2) calculating individual fitness, judging whether convergence standards are met or not, if yes, finishing the calculation, and outputting to obtain optimal weight and threshold of the neural network; if not, turning to the step (3); (3) selecting individuals according to the fitness, wherein the probability of selecting the individuals with high fitness is higher; (4) generating new individuals by crossing and mutating the individuals; (5) and (3) returning a new population consisting of new individuals generated by the cross mutation to the step (2) and continuing optimization. The basic GA parameters and operating modes obtained from repeated comparative studies are shown in Table 1.
TABLE 1 GA basic parameters and modes of operation
TABLE 2 basic parameter Table of BP neural network
As shown in Table 3, the GA-BP neural network has the diagnosis accuracy rate of 91% for age-related macular degeneration, and compared with the doctor manually diagnosing 63% and the BP neural network 73%, the diagnosis accuracy is greatly improved, and the conclusion can be drawn: the GA-BP neural network has the highest diagnosis precision, can be used for providing a good reference for the actual age-related macular degeneration, and can provide the most accurate guidance for the subsequent treatment.
TABLE 3 GA-BP neural network and comparison of artificial diagnosis, BP neural network diagnosis accuracy
The system also comprises a patient personal file which is established, wherein the patient personal file comprises data which is acquired by the detection module and manually input by the patient, specific diagnosis and treatment suggestions are given by the characteristic extraction module and the diagnosis and treatment module, the data are displayed by the interface display module, the re-diagnosis information is stored in the personal file in real time, the development trend of diseases is effectively monitored, and the patient can clearly know the state of an illness so as to provide a more accurate treatment scheme.
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that changes may be made in the embodiments and/or equivalents thereof without departing from the spirit and scope of the invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. The utility model provides an old macular degeneration intelligence system of diagnosing which characterized in that: the system consists of a detection module, a wireless transmission module, a characteristic extraction module and a diagnosis and treatment module, wherein the detection module comprises fundus OCT detection, fundus fluorescein angiography FFA, choroidal angiography, central vision detection, blood sugar detection and blood pressure detection, the detection module comprises fundus OCT, fundus fluorescein angiography FFA, choroidal angiography, central vision detection, blood sugar detection and blood pressure detection, the characteristic extraction module automatically extracts focus characteristics for the OCT, fundus fluorescein angiography FFA and choroidal angiography, the wireless transmission module transmits data obtained by the detection module to the characteristic extraction module, and the characteristic extraction data is led into a diagnosis and treatment module, the diagnosis and treatment module comprises a database, a GA-BP neural network diagnosis model and an interface display module, and a personal data document of a patient is also established, so that disease data can be dynamically tracked to give a targeted treatment scheme.
2. The intelligent diagnosis and treatment system for age related macular degeneration according to claim 1, wherein: the fundus OCT test and the choroid angiography are respectively SD-OCT and indocyanine green angiography ICG.
3. The intelligent diagnosis and treatment system for age related macular degeneration according to claim 1, wherein: the characteristic extraction module comprises preprocessing, image segmentation and characteristic extraction of fundus OCT, fundus fluorescein angiography FFA, choroidal angiography, wherein the preprocessing is median filtering preprocessing, images are used for segmenting yellow spots and blood vessel regions, characteristics are extracted to obtain drusen, progressive pigment epithelium RPE atrophy or detachment, geographic atrophy, macular choroidal neovascularization typicality CNV, macular choroidal neovascularization occult CNV, serous retinal pigment epithelium RPE detachment, exudation, edema, hemorrhage and scar focus characteristics, focus characteristic vectors are extracted, and the focus characteristic vectors are used as part input data of the diagnosis and treatment module.
4. The intelligent diagnosis and treatment system for age related macular degeneration according to claim 3, wherein: the diagnosis steps of the GA-BP neural network diagnosis model are as follows: the GA-BP neural network diagnosis model is trained through a large number of known fundus image samples, the images obtained by the detection module are subjected to feature extraction to obtain focus feature vector data and central vision, blood sugar and blood pressure data, the focus feature vector data and the central vision, blood sugar and blood pressure data are stored in a database, and the aged macular degeneration classification category is obtained through deduction through the trained GA-BP neural network diagnosis model.
5. The intelligent diagnosis and treatment system for age related macular degeneration according to claim 3, wherein: providing a corresponding treatment scheme after diagnosing the disease degree of a patient through a GA-BP neural network diagnosis model, wherein the classification category of the age-related macular degeneration is divided into five grades, wherein the first grade is early dry age-related macular degeneration, the second grade is late dry age-related macular degeneration, the third grade is early wet age-related macular degeneration, the fourth grade is middle wet age-related macular degeneration, and the fifth grade is late wet age-related macular degeneration; the first-level treatment scheme is oral anti-oxidation drugs and vitamins, the second-level treatment scheme is oral anti-oxidation drugs, inflammation inhibiting drugs and low-intensity laser treatment, the third-level treatment scheme is anti-VEGF drugs injection and laser treatment or photodynamic therapy, the fourth-level treatment scheme is anti-VEGF drugs injection and surgical treatment, and the fifth-level treatment scheme is vitreous cutting surgery and oral vitamins.
6. The intelligent diagnosis and treatment system for age related macular degeneration according to claim 3, wherein: the database also comprises a step of regularizing the parameters acquired by the detection module and partially manually input to form a knowledge rule and storing the knowledge rule into the database.
7. The intelligent diagnosis and treatment system for age related macular degeneration according to claim 3, wherein: the parameters partially inputted manually include the age of the patient, the age of the disease, and genetic factors, wherein the genetic factors are defined according to the age-related macular degeneration disease proportion in the third generation.
8. The intelligent diagnosis and treatment system for age related macular degeneration according to claim 1, wherein: optimizing BP neural network model parameters by a genetic algorithm GA (global optimization algorithm), wherein the GA parameters are as follows: the population size is 22, the iteration times are 40, the cross probability is 0.42, and the variation probability is 0.20; the BP neural network employs a three-layer network model, namely an input layer, a hidden layer, and an output layer.
9. The intelligent diagnosis and treatment system for age related macular degeneration according to any one of claims 1 to 7, wherein: in the GA-BP neural network diagnosis model, a BP neural network adopts a three-layer structure with 15 input layer nodes, 25 hidden layer nodes and 1 output layer nodes.
10. The intelligent diagnosis and treatment system for age related macular degeneration according to claim 1, wherein: the interface display module comprises a diagnosis result display and a treatment scheme display.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110803738.1A CN113243887B (en) | 2021-07-16 | 2021-07-16 | Intelligent diagnosis and treatment instrument for macular degeneration of old people |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110803738.1A CN113243887B (en) | 2021-07-16 | 2021-07-16 | Intelligent diagnosis and treatment instrument for macular degeneration of old people |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113243887A true CN113243887A (en) | 2021-08-13 |
| CN113243887B CN113243887B (en) | 2021-10-22 |
Family
ID=77180476
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110803738.1A Active CN113243887B (en) | 2021-07-16 | 2021-07-16 | Intelligent diagnosis and treatment instrument for macular degeneration of old people |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113243887B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113768461A (en) * | 2021-09-14 | 2021-12-10 | 北京鹰瞳科技发展股份有限公司 | Fundus image analysis method and system and electronic equipment |
| TWI786862B (en) * | 2021-10-01 | 2022-12-11 | 明達醫學科技股份有限公司 | Automatic eye test device and automatic eye test method |
| CN116434960A (en) * | 2023-06-13 | 2023-07-14 | 中国人民解放军总医院 | Cardiogenic shock clinical decision support system, equipment and storable medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120184845A1 (en) * | 2010-11-11 | 2012-07-19 | University Of Pittsburgh - Of The Commonwealth System Of Higher Education | Automated macular pathology diagnosis in threedimensional (3d) spectral domain optical coherence tomography (sd-oct) images |
| US20150029464A1 (en) * | 2013-07-26 | 2015-01-29 | The Regents Of The University Of Michigan | Automated Measurement of Changes in Retinal, Retinal Pigment Epithelial, or Choroidal Disease |
| CN108272434A (en) * | 2017-12-07 | 2018-07-13 | 江威 | The method and device that eye fundus image is handled |
| CN108305241A (en) * | 2017-09-06 | 2018-07-20 | 南京理工大学 | SD-OCT image GA lesion segmentation methods based on depth Voting Model |
| CN109903848A (en) * | 2019-03-04 | 2019-06-18 | 北京大学人民医院(北京大学第二临床医学院) | Fracture around joint clinic intelligent decision support system |
| CN109920547A (en) * | 2019-03-05 | 2019-06-21 | 北京工业大学 | A kind of diabetes prediction model construction method based on electronic health record data mining |
| CN110033861A (en) * | 2019-03-26 | 2019-07-19 | 西安交通大学 | Suitable for the blood vessel of OCTA image and the quantitative analysis method of macula lutea avascular area and system |
| CN110037650A (en) * | 2019-03-26 | 2019-07-23 | 中国科学院宁波工业技术研究院慈溪生物医学工程研究所 | Retina OCT abnormal image classification method based on convolutional neural networks |
| CN110378877A (en) * | 2019-06-24 | 2019-10-25 | 南京理工大学 | SD-OCT image CNV lesion detection method based on depth convolutional network model |
| CN110400289A (en) * | 2019-06-26 | 2019-11-01 | 平安科技（深圳）有限公司 | Eye fundus image recognition methods, device, equipment and storage medium |
-
2021
- 2021-07-16 CN CN202110803738.1A patent/CN113243887B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120184845A1 (en) * | 2010-11-11 | 2012-07-19 | University Of Pittsburgh - Of The Commonwealth System Of Higher Education | Automated macular pathology diagnosis in threedimensional (3d) spectral domain optical coherence tomography (sd-oct) images |
| US20150029464A1 (en) * | 2013-07-26 | 2015-01-29 | The Regents Of The University Of Michigan | Automated Measurement of Changes in Retinal, Retinal Pigment Epithelial, or Choroidal Disease |
| CN108305241A (en) * | 2017-09-06 | 2018-07-20 | 南京理工大学 | SD-OCT image GA lesion segmentation methods based on depth Voting Model |
| CN108272434A (en) * | 2017-12-07 | 2018-07-13 | 江威 | The method and device that eye fundus image is handled |
| CN109903848A (en) * | 2019-03-04 | 2019-06-18 | 北京大学人民医院(北京大学第二临床医学院) | Fracture around joint clinic intelligent decision support system |
| CN109920547A (en) * | 2019-03-05 | 2019-06-21 | 北京工业大学 | A kind of diabetes prediction model construction method based on electronic health record data mining |
| CN110033861A (en) * | 2019-03-26 | 2019-07-19 | 西安交通大学 | Suitable for the blood vessel of OCTA image and the quantitative analysis method of macula lutea avascular area and system |
| CN110037650A (en) * | 2019-03-26 | 2019-07-23 | 中国科学院宁波工业技术研究院慈溪生物医学工程研究所 | Retina OCT abnormal image classification method based on convolutional neural networks |
| CN110378877A (en) * | 2019-06-24 | 2019-10-25 | 南京理工大学 | SD-OCT image CNV lesion detection method based on depth convolutional network model |
| CN110400289A (en) * | 2019-06-26 | 2019-11-01 | 平安科技（深圳）有限公司 | Eye fundus image recognition methods, device, equipment and storage medium |
Non-Patent Citations (2)
| Title |
|---|
| 王冠方: "基于N-Gram模型的医学影像分类算法研究", 《中国优秀硕士学位论文全文数据库 (医药卫生科技辑)》 * |
| 陈迪等: "《实用眼科诊疗学》", 31 March 2019 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113768461A (en) * | 2021-09-14 | 2021-12-10 | 北京鹰瞳科技发展股份有限公司 | Fundus image analysis method and system and electronic equipment |
| CN113768461B (en) * | 2021-09-14 | 2024-03-22 | 北京鹰瞳科技发展股份有限公司 | Fundus image analysis method, fundus image analysis system and electronic equipment |
| TWI786862B (en) * | 2021-10-01 | 2022-12-11 | 明達醫學科技股份有限公司 | Automatic eye test device and automatic eye test method |
| CN116434960A (en) * | 2023-06-13 | 2023-07-14 | 中国人民解放军总医院 | Cardiogenic shock clinical decision support system, equipment and storable medium |
| CN116434960B (en) * | 2023-06-13 | 2023-10-13 | 中国人民解放军总医院 | Cardiogenic shock clinical decision support system, equipment and storable medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113243887B (en) | 2021-10-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113243887B (en) | Intelligent diagnosis and treatment instrument for macular degeneration of old people | |
| CN107423571B (en) | Diabetic retinopathy identifying system based on eye fundus image | |
| Li et al. | Deep learning based early stage diabetic retinopathy detection using optical coherence tomography | |
| Zhang et al. | A survey on computer aided diagnosis for ocular diseases | |
| Kauppi | Eye fundus image analysis for automatic detection of diabetic retinopathy | |
| Huang et al. | Glaucoma detection using adaptive neuro-fuzzy inference system | |
| Girard et al. | Artificial intelligence and deep learning in glaucoma: current state and future prospects | |
| Hallett et al. | Deep learning based unsupervised and semi-supervised classification for keratoconus | |
| Ovreiu et al. | Early detection of glaucoma using residual networks | |
| Hassan et al. | CDC-Net: Cascaded decoupled convolutional network for lesion-assisted detection and grading of retinopathy using optical coherence tomography (OCT) scans | |
| Singh et al. | Collaboration of features optimization techniques for the effective diagnosis of glaucoma in retinal fundus images | |
| Kumar et al. | Automated detection of eye related diseases using digital image processing | |
| Zeng et al. | Automated detection of diabetic retinopathy using a binocular siamese-like convolutional network | |
| Gupta et al. | An efficient model for detection and classification of internal eye diseases using deep learning | |
| Khalil et al. | A study to identify limitations of existing automated systems to detect glaucoma at initial and curable stage | |
| Yadav et al. | Automatic Cataract Severity Detection and Grading Using Deep Learning | |
| CN113273959B (en) | Portable diabetic retinopathy diagnosis and treatment instrument | |
| Thanh et al. | A Real-Time Classification Of Glaucoma from Retinal Fundus Images Using AI Technology | |
| Nugroho et al. | Convolutional neural network for classifying retinal diseases from OCT2017 dataset | |
| Santos-García et al. | Using artificial neural networks to identify glaucoma stages | |
| Chaudhary et al. | Glaucoma detection using cross validation algorithm: A comparitive evaluation on rapidminer | |
| Lee et al. | Using Deep Learning Models to Characterize Major Retinal Features on Color Fundus Photographs | |
| CN113781380B (en) | System for distinguishing neuromyelitis optica and primary open angle glaucoma | |
| Kaur et al. | Diabetic retinopathy detection system (DRDS); A novel GUI based approach for diabetic retinopathy detection | |
| Shuxian | Design of Postoperative Visual Acuity Prediction Program Based on Machine Learning before Cataract Surgery |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |