CN113889267A - Method for constructing diabetes diagnosis model based on eye image recognition and electronic equipment - Google Patents
Method for constructing diabetes diagnosis model based on eye image recognition and electronic equipment Download PDFInfo
- Publication number
- CN113889267A CN113889267A CN202111347064.5A CN202111347064A CN113889267A CN 113889267 A CN113889267 A CN 113889267A CN 202111347064 A CN202111347064 A CN 202111347064A CN 113889267 A CN113889267 A CN 113889267A
- Authority
- CN
- China
- Prior art keywords
- constructing
- diabetes
- image
- images
- samples
- 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.)
- Pending
Links
- 206010012601 diabetes mellitus Diseases 0.000 title claims abstract description 102
- 238000003745 diagnosis Methods 0.000 title claims abstract description 77
- 238000000034 method Methods 0.000 title claims abstract description 75
- 238000012549 training Methods 0.000 claims abstract description 41
- 210000003786 sclera Anatomy 0.000 claims abstract description 32
- 201000010099 disease Diseases 0.000 claims abstract description 26
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims abstract description 26
- 238000007781 pre-processing Methods 0.000 claims abstract description 14
- 210000001508 eye Anatomy 0.000 claims description 84
- 238000012545 processing Methods 0.000 claims description 22
- 239000008280 blood Substances 0.000 claims description 15
- 210000004369 blood Anatomy 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 13
- 238000010606 normalization Methods 0.000 claims description 10
- WQZGKKKJIJFFOK-GASJEMHNSA-N Glucose Natural products OC[C@H]1OC(O)[C@H](O)[C@@H](O)[C@@H]1O WQZGKKKJIJFFOK-GASJEMHNSA-N 0.000 claims description 8
- 239000008103 glucose Substances 0.000 claims description 8
- 238000013526 transfer learning Methods 0.000 claims description 8
- 238000013528 artificial neural network Methods 0.000 claims description 7
- 238000010276 construction Methods 0.000 claims description 5
- 241000406668 Loxodonta cyclotis Species 0.000 claims description 2
- 230000000007 visual effect Effects 0.000 abstract description 15
- 238000010586 diagram Methods 0.000 description 13
- 230000008569 process Effects 0.000 description 11
- 230000036541 health Effects 0.000 description 8
- 230000006870 function Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 238000001514 detection method Methods 0.000 description 4
- 239000003814 drug Substances 0.000 description 4
- 230000011218 segmentation Effects 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000011002 quantification Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 230000004075 alteration Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 210000000795 conjunctiva Anatomy 0.000 description 2
- 238000003066 decision tree Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 230000004089 microcirculation Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000003908 quality control method Methods 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 235000021124 short-term dietary change Nutrition 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 230000009885 systemic effect Effects 0.000 description 2
- 241000270295 Serpentes Species 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000003556 assay Methods 0.000 description 1
- 210000005252 bulbus oculi Anatomy 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000002405 diagnostic procedure Methods 0.000 description 1
- 230000010339 dilation Effects 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000003628 erosive effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 210000000720 eyelash Anatomy 0.000 description 1
- 238000007429 general method Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000003702 image correction Methods 0.000 description 1
- 238000003706 image smoothing Methods 0.000 description 1
- 208000015181 infectious disease Diseases 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000005295 random walk Methods 0.000 description 1
- 238000013077 scoring method Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Image Analysis (AREA)
Abstract
The application discloses a method for constructing a diabetes diagnosis model based on eye image recognition and electronic equipment, wherein the method for constructing the diabetes diagnosis model is used for acquiring a plurality of groups of binocular eye image images of people based on the electronic equipment; preprocessing each group of binocular eye image images to obtain sclera images and iris images of each group of binocular eye image images, and constructing a training sample set based on all the sclera images and the iris images; constructing a first step classifier by using the non-healthy samples and the healthy samples in the training sample set; constructing a classifier of a second step by using other disease samples and diabetes samples in the training sample set; and integrating the first step classifier and the second step classifier to obtain the diabetes diagnosis model. The diagnosis tool for the visual diagnosis of the diabetes diagnosis model can be obtained through the scheme, and the diabetes diagnosis is performed on the examinee based on objective standards without depending on visual observation and personal experience of doctors, so that the stability of the visual diagnosis of the diabetes is improved.
Description
Technical Field
The application relates to the technical field of medical equipment, in particular to a method for constructing a diabetes diagnosis model based on eye image recognition and electronic equipment.
Background
Diabetes is a systemic disease which seriously harms human health, and in order to treat the disease, the diabetes must be scientifically and accurately diagnosed so as to obtain an accurate diagnosis result. At present, the diagnosis of diabetes is mainly realized by blood drawing assay, and multiple blood drawing tests are needed in the process to detect blood sugar, so that the time and the economic cost of a subject and medical personnel are increased, and the method belongs to invasive diagnosis, so that certain infection risk exists. Furthermore, the subject must participate in the test at a specific time (hospital hours) and at a specific location (hospital).
The traditional Chinese medicine believes that the tiny color change of the human body due to diseases can be known by observing two eyes; modern medicine also believes that changes in the bulbar conjunctiva microcirculation may reflect systemic conditions, with corresponding changes in the bulbar conjunctiva microcirculation in the eye regardless of the disease. Therefore, the eye diagnosis is not only an important technology for 'getting up the work and treating the disease' but also an excellent path for combining Chinese and western medicine. Therefore, the diagnosis of diabetes can be realized through the research on the eye images.
However, the current diabetic eye diagnosis lacks a carrier for presenting an eye image entity and lacks an informationized and visualized presentation form, so that the diabetic eye diagnosis must be completed by visual observation and personal experience of doctors, great ambiguity, subjectivity and instability exist in the diabetic eye diagnosis, great inconvenience is brought to clinic, teaching and scientific research, and the effect of the diabetic eye diagnosis is severely limited.
Disclosure of Invention
In view of this, the present application provides a method for constructing a diabetes diagnosis model based on eye image recognition and an electronic device, which are used to provide a diagnosis tool for the eye diagnosis of diabetes, so as to improve the stability of the eye diagnosis of diabetes.
In order to achieve the above object, the following solutions are proposed:
a method for constructing a diabetes diagnosis model based on eye image recognition is applied to electronic equipment, and comprises the following steps:
acquiring a plurality of groups of binocular image images and demographic information of people based on the electronic equipment, wherein the people comprise non-healthy people and healthy people, and the non-healthy people comprise other disease people and diabetes people;
preprocessing each group of binocular eye image images to obtain sclera images and iris images of each group of binocular eye image images, and constructing a training sample set based on all the sclera images and the iris images;
constructing a first step classifier by using the non-healthy samples and the healthy samples in the training sample set;
constructing a second-step classifier by using other disease samples and diabetes samples in the training sample set;
and integrating the first step classifier and the second step classifier to obtain a diabetes diagnosis model.
Optionally, the binocular image includes eye images of two eyes at a plurality of viewing angles.
Optionally, the preprocessing the images of the binocular images of each group includes:
carrying out white balance normalization processing on the brightness channel of each binocular image;
extracting a sclera part image and an iris part image of the binocular eye image subjected to white balance normalization processing;
and accurately extracting and processing the sclera partial image and the iris partial image to obtain the sclera image and the iris image.
Optionally, the preprocessing is performed on each group of the binocular eye image, and the method further includes the steps of:
and performing enhancement processing on the sclera image and/or the iris image.
Optionally, the constructing a first step classifier by using the non-healthy samples and the healthy samples in the training sample set includes:
and taking the unhealthy sample as a case group sample, taking the healthy sample as a control group sample, and training a deep residual error neural network by using an age-hierarchical, multivariate, multi-view and multi-stage transfer learning method to obtain the first-step classifier.
Optionally, the constructing a second-step classifier by using other disease samples and diabetes samples in the training sample set includes the steps of:
and taking the diabetes sample as a case group sample, taking the other disease samples as a control group sample, and training the deep residual error neural network by using an age-hierarchical, multivariate, multi-view and multi-stage transfer learning method to obtain the second-step classifier.
Optionally, the method further comprises the steps of:
and constructing a regression model according to the binocular elephant image and the diabetes diagnosis model, wherein the regression model is used for predicting fasting blood glucose.
An electronic device comprising at least one processor and a memory coupled to the processor, wherein:
the memory is for storing a computer program or instructions;
the processor is used for acquiring and executing the computer program or the instructions so as to enable the electronic equipment to realize the method for constructing the diabetes diagnosis model based on the eye image recognition.
According to the technical scheme, the application discloses a method for constructing a diabetes diagnosis model based on eye image recognition and electronic equipment, and the method comprises the steps of acquiring a plurality of groups of binocular eye image images of people based on the electronic equipment; preprocessing each group of binocular eye image images to obtain sclera images and iris images of each group of binocular eye image images, and constructing a training sample set based on all the sclera images and the iris images; constructing a first step classifier by using the non-healthy samples and the healthy samples in the training sample set; constructing a classifier of a second step by using other disease samples and diabetes samples in the training sample set; and integrating the first step classifier and the second step classifier to obtain the diabetes diagnosis model. The diagnosis tool for the visual diagnosis of the diabetes diagnosis model can be obtained through the scheme, and the diabetes diagnosis is performed on the examinee based on objective standards without depending on visual observation and personal experience of doctors, so that the stability of the visual diagnosis of the diabetes is improved.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a block diagram of one embodiment provided herein;
FIG. 2 is an exemplary diagram of a binocular eye image according to an embodiment of the present application;
FIG. 3 is a flowchart illustrating a pre-processing procedure of a binocular image according to an embodiment of the present disclosure;
FIG. 4 is an exemplary illustration of a sclera partial image and an iris image for each set of binocular eye images according to an embodiment of the present application;
FIG. 5 is a flow chart of another embodiment of the present application for pre-processing binocular imagery;
FIG. 6 is a schematic view of a diagnostic process of a diabetes diagnostic model according to an embodiment of the present application;
FIG. 7 is a schematic diagram of a transfer learning process according to an embodiment of the present application;
FIG. 8 is a flowchart of another method for constructing a diabetes diagnosis model based on eye image recognition according to an embodiment of the present application
Fig. 9 is a block diagram of an electronic device according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
Example one
Fig. 1 is a flowchart of a method for constructing a diabetes diagnosis model based on eye image recognition according to an embodiment of the present application.
As shown in fig. 1, the method for constructing a diabetes diagnosis model provided in this embodiment is implemented based on an electronic device, which can be understood as an electronic computer or a server having computing and data processing functions, and is implemented based on an eye image recognition technology, and the method for constructing a diabetes diagnosis model specifically includes the following steps:
and S1, acquiring a plurality of groups of binocular image images and demographic information of the crowd.
The population includes both non-healthy and healthy populations, with healthy populations being used as control populations for non-healthy populations. The non-healthy population specifically comprises the diabetic population and other disease population, and the other disease population is used as the control population of the diabetic population. At the time of collection, a plurality of groups of binocular imagery of all people are collected, wherein the binocular imagery refers to eye imagery of each person's eyes at a plurality of viewing angles, as shown in fig. 2, wherein the plurality of viewing angles include, but are not limited to, up-view, down-view, left-view, right-view, and the like.
And after the acquisition, the acquired eye image is subjected to quality control processing, and quality control items comprise but are not limited to eye image integrity, eye surface exposed area, focusing state, hue, brightness and saturation. Meanwhile, the human demographic information is collected while the eye image is collected, and the eye image is labeled based on the middle health information of the demographic information, namely the corresponding eye image is labeled to belong to healthy people, other disease people or diabetes people.
And S2, preprocessing each group of binocular image images.
And preprocessing each group of binocular eye image to obtain a sclera image and an iris image of each binocular eye image. And constructing a training sample set by using the obtained multiple groups of sclera images and iris images. The training sample set comprises diabetes samples of diabetes people, other disease samples of other disease people and health samples of health people, wherein the diabetes samples and the other disease samples are non-health samples.
As shown in fig. 3, the specific pretreatment process is as follows:
s201, performing white balance normalization on the brightness channel of the eye image.
Namely, the white balance normalization processing is carried out on the brightness channel of each group of binocular image images.
S202, extracting and processing the foreground of each group of binocular eye image images.
Through foreground extraction processing. The skin, eyelashes, the edge of the eye image acquisition device and other areas of each eye image are removed, and only partial images of the surface and sclera of the eye and partial images of the iris are reserved. And normalized to 512x512x3 pixels for the image size, as shown in fig. 4, while maintaining the image scale. Methods used therein include, but are not limited to, active contour models (SNAKE, level set, etc.), graph theory segmentation (GraphCut, GrabCut and Random Walk, etc.), threshold segmentation (Otsu), etc.
And S203, extracting the sclera image and the iris image of each group of binocular eye image.
Namely, based on the result of the processing in the above steps, the iris image and the sclera image are accurately extracted. Firstly, judging an iris external rectangular frame by using a YOLO-v3 network, and then detecting the iris in the external rectangular frame according to a MiniBatchKMeans method. And finally, carrying out ellipse fitting on the current iris edge to obtain a smooth iris edge and an iris center coordinate. And separating the iris part from the foreground image to obtain the sclera. Specifically, the image processing algorithms include, but are not limited to, image enhancement, erosion, dilation, edge detection, and segmentation of connected regions.
In addition, the present embodiment may further include the following steps, as shown in fig. 5.
And S204, enhancing the sclera image and the iris image.
Methods of enhancement processing include, but are not limited to, image correction (histogram equalization, normalization, white balance, gamma correction and transformation, image smoothing, etc.), illumination distortion (images randomly change brightness, contrast, saturation, etc.) and geometric distortion (images randomly scale, crop, flip, rotate), local random occlusion of images (block occlusion, mesh occlusion), etc. The image is finally scaled to 224x224 pixel size. Therefore, the quality of the sclera image and the iris image is better, and the subsequent modeling processing is convenient.
S3, constructing a first step classifier by using the non-health sample and the health sample.
The method comprises the steps of using unhealthy samples in the training sample set as case group samples, using healthy samples as control group samples, and training a deep residual error neural network by using an age-hierarchical, multivariate, multi-view and multistage transfer learning method to obtain a first-step classifier.
S4, constructing a second step classifier by using other disease samples and diabetes samples.
The method comprises the steps of using diabetes samples in the training sample set as case group samples, using other disease samples as control group samples, and training a deep residual error neural network by using an age-hierarchical, multivariate, multi-view and multistage transfer learning method to obtain a second-step classifier.
And S5, integrating the first step classifier and the second step classifier.
The diabetes diagnosis model in this embodiment is obtained by integrating the first-step classifier and the second-step classifier.
When diagnosis is performed based on the diabetes diagnosis model of the embodiment, for a sample classified as unhealthy in the first step, prediction is performed using the second-step classifier, and the result of the second-step classifier is used as a final model result; for samples classified as healthy in the first step, the classification was directly non-diabetic. The specific flow is shown in FIG. 6.
According to the technical scheme, the embodiment provides the method for constructing the diabetes diagnosis model based on the eye image recognition, and the method is applied to the electronic equipment, and particularly, the method comprises the steps of acquiring a plurality of groups of binocular eye image images of people based on the electronic equipment; preprocessing each group of binocular eye image images to obtain sclera images and iris images of each group of binocular eye image images, and constructing a training sample set based on all the sclera images and the iris images; constructing a first step classifier by using the non-healthy samples and the healthy samples in the training sample set; constructing a classifier of a second step by using other disease samples and diabetes samples in the training sample set; and integrating the first step classifier and the second step classifier to obtain the diabetes diagnosis model. The diagnosis tool for the visual diagnosis of the diabetes diagnosis model can be obtained through the scheme, and the diabetes diagnosis is performed on the examinee based on objective standards without depending on visual observation and personal experience of doctors, so that the stability of the visual diagnosis of the diabetes is improved.
In the method, the deep residual error neural network is trained by using age-hierarchical, multivariate, multi-view and multi-stage transfer learning. Specifically, the subject is divided into three stages by age: young people group (<a1Year of age), middle aged group (not less than a)1Old and young<a2Aged people, and aged people (not less than a)2And (4) performing multivariate, multi-view and multi-stage model training respectively according to the samples of each group, establishing a diabetes image-level classifier, extracting high-dimensional features of the eye image from the model, and establishing the diabetes sample-level classifier. The method comprises the following specific steps:
firstly, a training set and a testing set are randomly divided according to a sample ID. Taking the training set as an example, for each sample, the multivariate eye image graph corresponding to the sample is respectively divided into G and L ═ L according to the shooting visual anglep},p∈[1,2,…,m]A subset of (a). I.e. G is the set of all perspective eye images for the model to find the eye image features of the whole area, and each subset of L contains the eye image (or horizontally mirrored eye image) of a specific inner corner and relative position of the eyeball for finding the local eye image features. Wherein the data partitioning strategy includes, but is not limited to, K-fold cross validation, no-put-back random sampling, hierarchical equal-ratio random sampling, and the like. Wherein the multi-element eye chart bagIncluding iris and sclera images, iris images, sclera images, and their enhanced images obtained in steps 2.2-2.4.
Dividing the process of model training into T stages, in the first stage, using ImageNet to pre-train the model, and respectively using images in G and L to finely tune the model to obtain a model G1And
wherein,
only pictures with view p are trained and predicted. As in the first stage, at each stage thereafter, the model obtained at the previous stage is used as the initial model of the current stage, and the images in G and L are used to perform fine tuning on the model. In particular, in order to improve the practicability of the model, the model can be finely adjusted in the last stages by using only the data of the classes with more errors in the model classification in the previous stage.
E.g. the T th0The classification accuracy of the model of each stage on a certain diabetes and a certain healthy population is low, and then T-T is carried out later0At this stage, the model can be fine-tuned using only the data of these two subclasses in the training set. In particular, if the model obtained in the previous stage consists of m sub-models, i.e.
The i +1 th stage fine-tunes the m models using the data in G, while using the image L with view angle ppFine tuning network
For T stages, this step will yield 2TAnd each group of models consists of 1 or m sub-models. Taking m-8 and T-4, and using the whole eye table image as an example, the transition learning process at each stage is shown in fig. 7.
Specifically, the value of T is an integer greater than 0; the batch size is an integer factoring in 2 and 5; the learning rate value range is (0.0000001, 1); optimization methods include, but are not limited to, Ranger, Adam, SGD, RMSProp, AdaGrad, and the like; the depth residual error network comprises the following modules in sequence: convolutional layer, normalization, activation layer, pooling layer, 8 sub-modules, random discard, softmax layer. Wherein each submodule comprises in turn: convolutional layer, normalization layer, activation layer, convolutional layer, normalization layer, residual connecting layer and activation layer. Convolution kernel sizes of convolutional layers include, but are not limited to, integers of 3,5,7, etc., and probabilities of random discard include, but are not limited to, 0.3,0.5, 0.7.
Specifically, the high-dimensional features include, but are not limited to, feature maps of intermediate layers of the model, feature vectors, image prediction results of the final layer, and the like, and the classifiers include, but are not limited to, voting, Logistic Regression (Logistic Regression), Random Forest (Random Forest), Support Vector Machine (SVM), Decision Tree (Decision Tree), adaptive boosting (AdaBoost), extreme gradient boosting (XGBoost), and the like. In addition, the high-dimensional features of the iris part of the image can be extracted, and the high-dimensional features and other high-dimensional features are modeled together to generate the diabetes sample-level classifier, namely the classifier of the second step.
In addition, in a specific implementation manner of this embodiment, the method further includes the following steps, as shown in fig. 8:
and S6, constructing a regression model according to the binocular eye image.
Namely, a regression model for predicting fasting blood glucose is established according to the obtained binocular image and the above diabetes diagnosis model. I.e. the last layer of the deep residual network of the above diabetes diagnosis model is changed to a fully connected layer with only one output node (last in this section). And training the residual error network by using the binocular eye image and the fasting blood glucose value thereof. And obtaining the fasting blood glucose prediction model based on the eye chart.
The regression model can provide the predicted value of fasting blood glucose for the patient diagnosed with diabetes, and is convenient for the patient to monitor the body.
The application provides a method for constructing a diabetes diagnosis model in steps, so that a health sample and a diabetes sample can be judged at the same time, the specificity of the model is improved, and the method has high application value. Because the aberration of eyes of diabetes and healthy people is large, and the characteristics for classifying the diabetes and the healthy people can be easily learned by the model, when the healthy people and other non-diabetic people are used as control groups in the training process of the model, the healthy control can prevent the model from continuously learning the specific characteristics of the diabetes by reducing the training loss. The two-step approach of the present application can avoid this situation, allowing the model to learn the diabetes-specific characteristics.
The method for constructing the diabetes diagnosis model in steps greatly reduces the workload of subsequent model adjustment. The use of diverse control data is a guarantee of high robustness and specificity of the model, especially when the model is applied in a community screening project. The knowledge of diseases is continuously updated, so that new kinds of data are necessarily required to be added into the training of the model in the long-term application of the model. The two-step method proposed in this application excludes the healthy sample with the largest proportion in the first step. In the later model adjusting process, only the model in the second step needs to be retrained, so that the time cost is greatly saved.
The application establishes the diabetes classifier suitable for people of different ages, and improves the recognition sensitivity of young diabetics and the specificity of old subjects.
The application realizes blood sugar value quantification based on the eye surface image by utilizing the deep learning technology. Compared with the traditional blood sugar measuring method, the blood sugar quantification based on the eye picture is more convenient and noninvasive, and does not depend on short-term dietary change, thereby being more objective and accurate.
The method and the device use the deep learning technology and the traditional machine learning technology to realize automatic segmentation of the eye surface image, the iris and the sclera, can obtain the iris area with smooth edges, and can obtain the central coordinate of the iris.
The application not only provides a novel method for predicting and diagnosing the diabetes risk, but also provides a general method and paradigm based on the modern application and research of the traditional medical eye diagnosis.
Compared with a scale scoring method, the method has the advantages that only eye surface (iris and scleral area) photographing is needed during detection, detection complexity is reduced, detection accuracy is improved, and the method is more suitable for community-oriented large-scale diabetes screening.
In addition, the diabetes diagnosis model based on the application has the advantages of accuracy draft and wide application range when used for diagnosing patients. The ROC-AUC of the model for diagnosing the diabetes reaches up to 0.97, which is higher than the performance of a diabetes scale. Provides a reliable technical means for community-oriented diabetes large-scale screening and provides support for national diabetes prevention, control and management.
Compared with the traditional Chinese medicine eye diagnosis method, the diabetes diagnosis model has the following advantages:
1. the algorithm is strong in universality and objective and accurate in result. The method is based on the large data of the eye image, and an advanced artificial intelligence algorithm is adopted, so that the result is more objective and robust.
2. And (4) objectivity. The diabetes diagnosis model is based on the large data of the eye image, does not depend on the experience and the technical level of a doctor, and is more objective and accurate in result.
3. Full automation and strong practicability are achieved: the diabetes diagnosis model realizes the automation of diabetes diagnosis, does not need the participation of doctors in visual diagnosis, is quick in diagnosis and has higher practicability.
Compared with other diabetes diagnosis methods, the diabetes diagnosis model has the following advantages:
1. the robustness is high. The ROC-AUC of the diagnosis of the diabetes diagnosis model is up to 0.97, which is higher than that of other methods. The method uses the diversified contrast data to construct the model according to the two-step method, so that the robustness and the specificity of the model are improved, and a convenient and low-consumption way is provided for subsequent addition of new data and adjustment of the model.
2. The invention establishes the diabetes classifier suitable for people of different ages, and improves the recognition sensitivity to young diabetics and the specificity to older subjects.
3. Objectivity: the application establishes a blood sugar quantification method based on eye images. Compared with the traditional fasting blood glucose and random blood glucose detection method, the method is non-invasive and convenient, and is not limited by short-term dietary change.
Example two
Fig. 9 is a block diagram of an electronic device according to an embodiment of the present application.
As shown in fig. 9, the electronic device provided in this embodiment can be understood as an electronic computer or a server having computing and data processing functions, and includes at least one processor 101 and a memory 102, which are connected by a data bus 103. The memory is used for storing computer programs or instructions, and the processor is used for acquiring and executing the corresponding computer programs or instructions, so that the electronic equipment realizes the method for constructing the diabetes diagnosis model based on the eye image recognition.
The construction method specifically comprises the steps that multiple groups of binocular eye image images of people are obtained based on electronic equipment; preprocessing each group of binocular eye image images to obtain sclera images and iris images of each group of binocular eye image images, and constructing a training sample set based on all the sclera images and the iris images; constructing a first step classifier by using the non-healthy samples and the healthy samples in the training sample set; constructing a classifier of a second step by using other disease samples and diabetes samples in the training sample set; and integrating the first step classifier and the second step classifier to obtain the diabetes diagnosis model. The diagnosis tool for the visual diagnosis of the diabetes diagnosis model can be obtained through the scheme, and the diabetes diagnosis is performed on the examinee based on objective standards without depending on visual observation and personal experience of doctors, so that the stability of the visual diagnosis of the diabetes is improved.
The embodiments in the present specification are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, apparatus, or computer program product. Accordingly, embodiments of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
Embodiments of the present invention are described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing terminal to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing terminal to cause a series of operational steps to be performed on the computer or other programmable terminal to produce a computer implemented process such that the instructions which execute on the computer or other programmable terminal provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications of these embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the embodiments of the invention.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or terminal. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or terminal that comprises the element.
The technical solutions provided by the present invention are described in detail above, and the principle and the implementation of the present invention are explained in this document by applying specific examples, and the descriptions of the above examples are only used to help understanding the method and the core idea of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (8)
1. A method for constructing a diabetes diagnosis model based on eye image recognition is applied to electronic equipment, and is characterized by comprising the following steps:
acquiring a plurality of groups of binocular image images and demographic information of people based on the electronic equipment, wherein the people comprise non-healthy people and healthy people, and the non-healthy people comprise other disease people and diabetes people;
preprocessing each group of binocular eye image images to obtain sclera images and iris images of each group of binocular eye image images, and constructing a training sample set based on all the sclera images and the iris images;
constructing a first step classifier by using the non-healthy samples and the healthy samples in the training sample set;
constructing a second-step classifier by using other disease samples and diabetes samples in the training sample set;
and integrating the first step classifier and the second step classifier to obtain a diabetes diagnosis model.
2. The construction method according to claim 1, wherein the binocular imagery includes eye imagery of both eyes at a plurality of viewing angles.
3. The construction method according to claim 1, wherein the preprocessing of each set of the binocular image includes the steps of:
carrying out white balance normalization processing on the brightness channel of each binocular image;
extracting a sclera part image and an iris part image of the binocular eye image subjected to white balance normalization processing;
and accurately extracting and processing the sclera partial image and the iris partial image to obtain the sclera image and the iris image.
4. The construction method according to claim 3, wherein the preprocessing is performed on each set of the binocular image, further comprising the steps of:
and performing enhancement processing on the sclera image and/or the iris image.
5. The method of constructing as claimed in claim 1, wherein said constructing a first step classifier using non-healthy samples and healthy samples in said training sample set comprises the steps of:
and taking the unhealthy sample as a case group sample, taking the healthy sample as a control group sample, and training a deep residual error neural network by using an age-hierarchical, multivariate, multi-view and multi-stage transfer learning method to obtain the first-step classifier.
6. The method of constructing as claimed in claim 1, wherein said constructing a second step classifier using other disease samples and diabetes samples in said training sample set comprises the steps of:
and taking the diabetes sample as a case group sample, taking the other disease samples as a control group sample, and training the deep residual error neural network by using an age-hierarchical, multivariate, multi-view and multi-stage transfer learning method to obtain the second-step classifier.
7. The construction method according to any one of claims 1 to 6, further comprising the steps of:
and constructing a regression model according to the binocular elephant image and the diabetes diagnosis model, wherein the regression model is used for predicting fasting blood glucose.
8. An electronic device comprising at least one processor and a memory coupled to the processor, wherein:
the memory is for storing a computer program or instructions;
the processor is used for acquiring and executing the computer program or the instructions so as to enable the electronic equipment to realize the method for constructing the diabetes diagnosis model based on the eye image recognition according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111347064.5A CN113889267A (en) | 2021-11-15 | 2021-11-15 | Method for constructing diabetes diagnosis model based on eye image recognition and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111347064.5A CN113889267A (en) | 2021-11-15 | 2021-11-15 | Method for constructing diabetes diagnosis model based on eye image recognition and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113889267A true CN113889267A (en) | 2022-01-04 |
Family
ID=79018079
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111347064.5A Pending CN113889267A (en) | 2021-11-15 | 2021-11-15 | Method for constructing diabetes diagnosis model based on eye image recognition and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113889267A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114511524A (en) * | 2022-01-21 | 2022-05-17 | 深圳市铱硙医疗科技有限公司 | Method and system for detecting neurological disease |
| CN114566273A (en) * | 2022-01-05 | 2022-05-31 | 中山大学孙逸仙纪念医院 | Method for constructing onychomycosis diagnosis model, diagnosis model and diagnosis device |
| CN116030042A (en) * | 2023-02-24 | 2023-04-28 | 智慧眼科技股份有限公司 | Diagnostic device, method, equipment and storage medium for doctor's diagnosis |
| CN116913508A (en) * | 2023-09-13 | 2023-10-20 | 中国人民解放军总医院第一医学中心 | Method and system for predicting diabetic nephropathy based on white eye characteristics |
| CN117079808A (en) * | 2023-10-16 | 2023-11-17 | 罗麦(北京)营养食品研究有限公司 | Be used for ocular surface periocular image collection and artificial intelligence health analysis system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111640498A (en) * | 2020-06-08 | 2020-09-08 | 博奥生物集团有限公司 | Information processing method and device based on eye image data |
| CN112686855A (en) * | 2020-12-28 | 2021-04-20 | 博奥生物集团有限公司 | Information correlation method for elephant and symptom information |
| CN113611420A (en) * | 2021-08-11 | 2021-11-05 | 季凯 | Disease screening method and system based on blood examination indexes |
-
2021
- 2021-11-15 CN CN202111347064.5A patent/CN113889267A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111640498A (en) * | 2020-06-08 | 2020-09-08 | 博奥生物集团有限公司 | Information processing method and device based on eye image data |
| CN112686855A (en) * | 2020-12-28 | 2021-04-20 | 博奥生物集团有限公司 | Information correlation method for elephant and symptom information |
| CN113611420A (en) * | 2021-08-11 | 2021-11-05 | 季凯 | Disease screening method and system based on blood examination indexes |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114566273A (en) * | 2022-01-05 | 2022-05-31 | 中山大学孙逸仙纪念医院 | Method for constructing onychomycosis diagnosis model, diagnosis model and diagnosis device |
| CN114511524A (en) * | 2022-01-21 | 2022-05-17 | 深圳市铱硙医疗科技有限公司 | Method and system for detecting neurological disease |
| CN116030042A (en) * | 2023-02-24 | 2023-04-28 | 智慧眼科技股份有限公司 | Diagnostic device, method, equipment and storage medium for doctor's diagnosis |
| CN116913508A (en) * | 2023-09-13 | 2023-10-20 | 中国人民解放军总医院第一医学中心 | Method and system for predicting diabetic nephropathy based on white eye characteristics |
| CN116913508B (en) * | 2023-09-13 | 2023-12-12 | 中国人民解放军总医院第一医学中心 | Method and system for predicting diabetic nephropathy based on white eye characteristics |
| CN117079808A (en) * | 2023-10-16 | 2023-11-17 | 罗麦(北京)营养食品研究有限公司 | Be used for ocular surface periocular image collection and artificial intelligence health analysis system |
| CN117079808B (en) * | 2023-10-16 | 2024-02-02 | 罗麦(北京)营养食品研究有限公司 | Be used for ocular surface periocular image collection and artificial intelligence health analysis system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | Automatic detection of diabetic retinopathy in retinal fundus photographs based on deep learning algorithm | |
| US20220076420A1 (en) | Retinopathy recognition system | |
| CN113889267A (en) | Method for constructing diabetes diagnosis model based on eye image recognition and electronic equipment | |
| Khandouzi et al. | Retinal vessel segmentation, a review of classic and deep methods | |
| CN113768461B (en) | Fundus image analysis method, fundus image analysis system and electronic equipment | |
| CN113012093B (en) | Training method and training system for glaucoma image feature extraction | |
| Sakthivel et al. | An automated detection of glaucoma using histogram features | |
| Vimal et al. | RETRACTED ARTICLE: A method of progression detection for glaucoma using K-means and the GLCM algorithm toward smart medical prediction | |
| CN110610480B (en) | MCASPP neural network eyeground image optic cup optic disc segmentation model based on Attention mechanism | |
| Phridviraj et al. | A bi-directional Long Short-Term Memory-based Diabetic Retinopathy detection model using retinal fundus images | |
| Bali et al. | Analysis of deep learning techniques for prediction of eye diseases: A systematic review | |
| Shaik et al. | Glaucoma identification based on segmentation and fusion techniques | |
| Kamal et al. | A comprehensive review on the diabetic retinopathy, glaucoma and strabismus detection techniques based on machine learning and deep learning | |
| Sesikala et al. | A Study on Diabetic Retinopathy Detection, Segmentation and Classification using Deep and Machine Learning Techniques | |
| CN113378794A (en) | Information correlation method for elephant and symptom information | |
| Shraddha et al. | Identification of Suitable Machine Learning or Deep Learning Algorithmfor Diabetic Retinopathy Detection | |
| Akbar et al. | A Novel Filtered Segmentation-Based Bayesian Deep Neural Network Framework on Large Diabetic Retinopathy Databases. | |
| Taranum et al. | Image Based Edge Weighted Linked Segmentation Model Using Deep Learning for Detection of Diabetic Retinopathy. | |
| Santhi et al. | Diagnosis of diabetic retinopathy by exudates detection using clustering techniques | |
| Hussein et al. | Convolutional Neural Network in Classifying Three Stages of Age-Related Macula Degeneration | |
| De Silva et al. | A thickness sensitive vessel extraction framework for retinal and conjunctival vascular tortuosity analysis | |
| Akshita et al. | Diabetic retinopathy classification using deep convolutional neural network | |
| Priya et al. | Automated diagnosis of age-related macular degeneration using machine learning techniques | |
| Hussein et al. | Automatic classification of AMD in retinal images | |
| Li et al. | Retinal OCT image classification based on domain adaptation convolutional neural networks |
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 |