Anterior eye disease diagnostic system and diagnostic method using same
Provided is a diagnosis system for anterior eye diseases. An image obtainer obtains an anterior eye image of an image obtainer configured to obtain an anterior eye image of a subject. A characteristic extractor extracts characteristic information of the obtained anterior eye image on the basis of a machine-learning model. A disease analyzer determines whether or not an anterior eye of the subject has a disease and classifying a disease class according to the extracted characteristic information.
1. A diagnosis system for anterior eye diseases, the diagnosis system comprising:
an image obtainer configured to obtain an anterior eye image of the anterior eye of a subject;
a characteristic extractor configured to extract characteristic information of the obtained anterior eye image including a machine-learning model which learns patterns of changes or aspects of changes in anterior eyes over time from a plurality of anterior eye images, wherein the characteristic extractor extracts the characteristic information by applying a plurality of filters to each of a plurality of areas of the anterior eye image and repeatedly resizing the plurality of areas to generate a feature map of the anterior eye of the subject; and
a disease analyzer configured to determine whether or not the anterior eye of the subject has a disease and classifying a disease class according to the extracted characteristic information and the patterns or changes in the plurality of anterior images over time, and wherein the disease analyzer classifies the disease using a classification model being one of a multi-layer perceptron (MLP) of a support vector maching (SVM).
2. The diagnosis system of claim 1 , further comprising a storage in which the plurality of anterior eye images are stored including previously-obtained anterior eye images of the subject and other subjects,
wherein the anterior eye image of the subject is obtained in real time for a diagnosis or is obtained by loading the anterior eye image stored in the storage.
3. The diagnosis system of claim 1 , further comprising an output unit configured to output medical information regarding whether or not the anterior eye of the subject has the disease and the disease class, which are determined and classified by the disease analyzer.
4. The diagnosis system of claim 3 , wherein the machine-learning model of the characteristic extractor comprises a deep convolutional generative adversarial net, and
the output unit outputs both the anterior eye image and image information regarding a lesion extracted in accordance with the machine-learning model.
5. The diagnosis system of claim 3 , wherein the medical information comprises at least one of current state information of the anterior eye, future state prediction information of the anterior eye, and treatment information of the disease class.
6. The diagnosis system of claim 1 , wherein the characteristic information comprises at least one of color information of a lesion, position information of the lesion, depth information of the lesion, blood vessel information of surrounding portions of cornea, surface information of conjunctiva, and surface information of sclera.
7. The diagnosis system of claim 1 , wherein the image obtainer includes a slit lamp microscope.
8. A diagnosis method using a diagnosis system for anterior eye diseases, the diagnosis method comprising:
obtaining an anterior eye image of the anterior eye of a subject;
extracting characteristic information of the obtained anterior eye image on using a machine-learning model which learns patterns of changes or aspects of changes in anterior eyes over time from a plurality of anterior eye images, wherein extracting the characteristic information includes applying a plurality of filters to each of a plurality of areas of the anterior eye image and repeatedly resizing the plurality of areas to generate a feature map of the anterior eye of the subject; and
determining whether or not the anterior eye of the subject has a disease and classifying a disease class according to the extracted characteristic information and the patterns or changes in the plurality of anterior images over time, and wherein the disease is classified using a classification model being one of a multi-layer perceptron (MLP) or a support vector maching (SVM).
9. The diagnosis method of claim 8 , wherein the anterior eye image of the subject is obtained in real time for a diagnosis or is obtained by loading an anterior eye image previously stored in a storage of the diagnosis system for anterior eye disease.
10. The diagnosis method of claim 8 , further comprising outputting, by an output unit of the diagnosis system for anterior eye diseases, medical information regarding the classified disease class after the determining and classifying.
11. A non-transitory computer readable recording medium in which a program configured to realize the method of claim 8 is recorded.