Apparatus for predicting parameter of corneal refractive correction lens on basis of learning model and operation method thereof

Abstract

A method for predicting a parameter of a corneal refractive correction lens according to one embodiment of the present invention may comprise the steps of: collecting inspection data of a subject; generating a parameter prediction model of the corneal refractive correction lens by using the collected inspection data; receiving input data; and predicting at least one parameter of parameters of the corneal refractive correction lens by applying the input data to the prediction model, wherein the parameters of the corneal refractive correction lens may include overall diameter (OAD), base curve (BC), return zone depth (RZD), and landing zone angle (LZA).

Description

Device for predicting parameters of corneal refractive lens based on learning model and its operating method

[0001] The present invention relates to artificial intelligence technology. Specifically, it relates to a device for predicting parameters of a corneal refractive lens based on artificial intelligence technology, and its operating method.

[0002] Orthokeratology contact lenses (ORKs) are a widely used treatment for myopia correction and suppression in growing children and adolescents. These contact lenses are specially designed, rigid, oxygen-permeable lenses that flatten the central portion of the cornea, temporarily correcting myopia and astigmatism.

[0003] When prescribing corneal refractive lenses, the hospital provides trial lenses to determine the right lens for the patient. The patient is evaluated by wearing the trial lenses several times. After deciding on the right lens for the patient, the patient wears the selected lens for a certain period of time and returns to the hospital for observation. The process is repeated to determine the final lens. Due to the nature of lenses that must be worn at night, the corrective effect can only be seen after a certain period of time. In addition, side effects such as corneal scarring and decreased vision caused by lenses that do not fit the eye must be closely monitored. This means that it can take anywhere from a week to several months to determine the most appropriate lens for the patient.

[0004] It is necessary to develop a technology that can minimize the risk of side effects by providing corneal refractive lenses appropriate for the patient's eye condition, reduce the number of trial lens wears, and improve fitting efficiency.

[0005] The purpose of the present invention is to provide a device for predicting parameters of a corneal refractive correction lens based on a machine learning model and an operating method thereof.

[0006] A method for predicting parameters of a corneal refractive lens according to one embodiment of the present invention comprises the steps of: collecting examination data of a subject; generating a parameter prediction model of a corneal refractive lens using the collected examination data; receiving input data; and predicting at least one parameter of the corneal refractive lens by applying the input data to the prediction model, wherein the parameters of the corneal refractive lens may include an overall diameter (OAD), a base curve (BC), a return zone depth (RZD), and a landing zone angle (LZA).

[0007] According to one embodiment of the present invention, the examination data may include examination data before and after the procedure for the subject.

[0008] According to one embodiment of the present invention, the step of selecting, among the inspection data, inspection data stored in a database in the past rather than a preset time as training data is further included, and the step of generating the prediction model can learn the prediction model by utilizing the training data.

[0009] According to one embodiment of the present invention, the method may further include a step of randomly selecting internal validation data from the training data; and a step of judging the performance of the prediction model using the internal validation data.

[0010] According to one embodiment of the present invention, the method may further include a step of selecting data among the test data that is not selected as the training data as external validation data; and a step of judging the performance of the prediction model using the external validation data.

[0011] According to one embodiment of the present invention, the input data may include at least one of age, gender, uncorrected visual acuity, corrected visual acuity, refractive power by objective examination, refractive power by autorefraction, corneal diameter, corneal vertical / horizontal / average eccentricity, corneal topography image, corneal and corneal epithelial thickness by region according to ocular optical coherence tomography image, axial length according to optical laser ophthalmometry, central corneal thickness, anterior chamber depth, corneal diameter, and pupil size.

[0012] According to one embodiment of the present invention, the step of generating the prediction model may generate the prediction model based on a statistical analysis of the correlation between the pre-treatment inspection data and the post-treatment inspection data.

[0013] In one embodiment of the present invention, an electronic device for predicting parameters of a corneal refractive lens comprises: a memory storing commands for performing various operations; and at least one processor electrically connected to the memory, wherein the processor collects examination data of a subject, generates a parameter prediction model of the corneal refractive lens using the collected examination data, receives input data, and applies the input data to the prediction model to predict at least one parameter of the corneal refractive lens, wherein the parameters of the corneal refractive lens may include an overall diameter (OAD) of the cornea, a base curve (BC), a return zone depth (RZD), and a landing zone angle (LZA).

[0014] According to the device for predicting parameters of a corneal refractive lens based on the machine learning model of the present invention and its operating method, parameters of a corneal refractive lens suitable for the eye condition of a subject can be predicted based on the eye examination data of the subject, thereby reducing the risk of side effects and improving fitting efficiency by reducing the number of times the lens is worn.

[0015] Figure 1 shows the operating environment of a device for predicting parameters of a corneal refractive correction lens based on a machine learning model according to one embodiment.

[0016] FIG. 2 is a block diagram showing the configuration of a device for predicting parameters of a corneal refractive correction lens based on a learning model according to one embodiment.

[0017] Figure 3 illustrates the relationship between devices according to one embodiment.

[0018] Figure 4 illustrates a learning model according to one embodiment of the present invention.

[0019] FIG. 5 conceptually illustrates the operation of an electronic device for predicting parameters of a corneal refractive lens according to one embodiment.

[0020] Figure 6a is a flowchart showing the operation of an electronic device according to one embodiment of the present invention.

[0021] FIGS. 6b to 6d are conceptual diagrams showing the operation of an electronic device according to one embodiment of the present invention.

[0022] FIG. 7 illustrates an architecture for predicting parameters of a corneal refractive lens according to one embodiment.

[0023] FIG. 8 illustrates an architecture for predicting parameters of a corneal refractive lens according to another embodiment.

[0024] Hereinafter, various embodiments of the present invention will be described with reference to the accompanying drawings. It should be understood that the present invention is not limited to specific embodiments, but rather encompasses various modifications, equivalents, and / or alternatives of the embodiments of the present invention. In connection with the description of the drawings, similar reference numerals may be used for similar components.

[0025] In this document, the expressions “has”, “may have”, “includes”, or “may include” indicate the presence of a feature (e.g., a number, function, operation, or component such as a part), but do not exclude the presence of additional features.

[0026] In this document, the expressions "A or B," "at least one of A and / or B," or "one or more of A and / or B" can include all possible combinations of the listed items. For example, "A or B," "at least one of A and B," or "at least one of A or B" can all refer to cases where (1) at least one A is included, (2) at least one B is included, or (3) at least one A and at least one B are included.

[0027] The terms "first," "second," "first," or "second" used herein may describe various components, regardless of order and / or importance, and are used only to distinguish one component from another, without limiting the components. For example, without departing from the scope of the rights set forth in this document, the first component may be renamed the second component, and similarly, the second component may be renamed the first component.

[0028] The expression "configured to" as used herein can be used interchangeably with, for example, "suitable for", "having the capacity to", "designed to", "adapted to", "made to", or "capable of". The term "configured to" does not necessarily mean "specifically designed to".

[0029] In this document, the words "command", "instruction", "control information", "message", "information", "data", "packet", "data packet", "intent" and / or "signal" transmitted and received between the first electronic device(s) and the second electronic device(s) may include or refer to human-perceivable ideas or specific electrical expressions (e.g., digital codes / analog physical quantities) regardless of their expressions. It will be apparent to those skilled in the art to which the invention disclosed in this document pertains that the exemplary expressions listed above may be interpreted in various ways depending on the context in which they are used. In this document, "is greater than B" not only simply means "is greater than B" but also includes the meaning of "is equal to or greater than B."

[0030] The terms used in this document are used only to describe specific embodiments and may not be intended to limit the scope of other embodiments. The singular expression may include the plural expression unless the context clearly indicates otherwise. Terms used herein, including technical or scientific terms, may have the same meaning as commonly understood by those of ordinary skill in the art described in this document. Terms defined in general dictionaries among the terms used in this document may be interpreted as having the same or similar meaning in the context of the relevant technology, and shall not be interpreted in an idealized or overly formal sense unless explicitly defined in this document. In some cases, even if a term is defined in this document, it cannot be interpreted to exclude the embodiments of this document.

[0031] Figure 1 shows the operating environment (100) of a device for predicting parameters of a corneal refractive correction lens based on a machine learning model according to one embodiment.

[0032] The operating environment (100) of a device for predicting parameters of a corneal refractive lens based on a machine learning model according to various embodiments of the present invention may include at least one electronic device (110, 120, 130). In addition, each device may communicate with each other through a network (150).

[0033] According to one embodiment, the first electronic device (110) may be a device that allows a medical professional or the like to input information on an eye examination (cornea, fundus, etc.) of a subject (patient) according to various embodiments of the present invention. The first electronic device (110) may be a device such as a smartphone, tablet PC, laptop, or computer.

[0034] The second electronic device (120) may be a device that performs measurements on a subject. The second electronic device (120) may be various equipment such as tomography, topography, optical coherence tomography (OCT), ultrasound biomicroscopy (UMB), etc., such as Pentacam, CASIA2, AL-Scan, OQAS, etc.

[0035] The second electronic device (120) may be a device for examining the cornea of ​​a subject. The second electronic device (120) may be a device for checking corneal curvature, astigmatism, eccentricity, corneal size, corneal thickness, keratoconus, corneal tension, corneal topography, etc.

[0036] Additionally, the second electronic device (120) may be a device for photographing the eye or retina. The second electronic device (120) may be a device for measuring the anterior segment of the eye using a laser and / or high-frequency ultrasound. For example, the second electronic device (120) may include an OCT (Optical Coherence Tomography) device that takes a tomographic image of the area around the optic nerve and the macula.

[0037] According to various embodiments of the present invention, the operating environment (100) may include a plurality of second electronic devices (120).

[0038] The third electronic device (130) is a device that acquires input data and processes calculations based on a machine learning model according to an embodiment of the present invention, and can perform all or part of the operations described below. The input data may include the patient's age, sex, naked eye, corrected vision, eye examination information (e.g., refractive power, astigmatic axis, corneal refractive power, corneal thickness, corneal eccentricity, etc.), corneal topography), optical coherence tomography, optical biometer extraction information, etc.

[0039] According to one embodiment of the present invention, the third electronic device (130) may be a learning device. The third electronic device (130) may generate a learning model through neural network operations, etc. Specifically, the third electronic device (130) may train the learning model based on a learning data set.

[0040] The third electronic device (130) can build and store learning models according to various embodiments of the present invention. The third electronic device (130) can store the learning models in a database. Upon external request, the third electronic device (130) can transmit the learning models or directly operate the learning models.

[0041] The third electronic device (130) can store and / or drive a learning model according to various embodiments of the present invention. The learning model may include a model for predicting parameters of a corneal refractive lens. The third electronic device (130) can collect and / or store data used to predict parameters of the corneal refractive lens.

[0042] The third electronic device (130) can communicate with the first electronic device (110) and the second electronic device (120) to receive or transmit data such as input data or a learning model.

[0043] The third electronic device (130) can receive learning data from the first electronic device (110) and the second electronic device (120). The third electronic device (130) can transmit a learning model generated based on the learning data to an external device. Alternatively, the third electronic device (130) can perform all or part of the operations of the first electronic device (110) and the second electronic device (120).

[0044] The third electronic device (130) can store the learning model in a database. The third electronic device (130) can store the weights applied to the learned learning model. The third electronic device (130) can collect and / or store data used in the learning model. The third electronic device (130) can transmit the learning model to an external device upon request from the external device.

[0045] The third electronic device (130) may perform calculations directly using the learning model and transmit the results to an external device. For example, the third electronic device (130) may obtain input data from the first electronic device (110) or the second electronic device (120), determine parameters based on a corneal refractive lens parameter prediction model, and in this case, transmit the derived results to the first electronic device (110) or the second electronic device (120).

[0046] The third electronic device (130) may include one or more servers, for example, a cloud of servers.

[0047] According to one embodiment, the network (150) can be directly or indirectly communicatively coupled with the first electronic device (110), the second electronic device (120), and the third electronic device (130). The network (150) can include a wired or wireless communication network. Furthermore, the network (150) can include a short-range or long-range communication network. For example, the network (150) can include a cellular network such as LTE, LTE-A, 5G, or 6G.

[0048] The configuration of FIG. 1 is exemplary and is not intended to limit the scope of the present invention. The environment in which the device according to the present invention operates may include only some of the configurations of FIG. 1, or may include additional or different configurations compared to the configurations illustrated in FIG. 1.

[0049] In the embodiments below, the operation of the third electronic device (130) is mainly described, but some or all of the operations described below may be performed by the first electronic device (110) or the second electronic device (120).

[0050] FIG. 2 is a block diagram showing the configuration of a device for predicting parameters of a corneal refractive correction lens based on a learning model according to one embodiment.

[0051] As illustrated in FIG. 2, the electronic device (200) (e.g., the third electronic device (130) of FIG. 1) may include a bus (210), a display (220), a communication circuit (230), a database (240), a memory (250), an input / output (I / O) interface (260), and a processor (270). In other embodiments, the device (200) may omit at least one of the above components or may additionally include other components.

[0052] For reference, the components (210, 220, 230, 240, 250, 260, 270) of the electronic device (200) illustrated in FIG. 2 are merely exemplary components for explaining a device for predicting parameters of a corneal refractive correction lens based on a learning model according to an embodiment of the present invention. That is, it is clear that the electronic device (200) according to an embodiment of the present invention may additionally include other components in addition to the illustrated components.

[0053] The bus (210) can electrically connect the components (220 to 270) to each other. The bus (210) can include circuitry for communication (e.g., control messages and / or data) between the components (220 to 270).

[0054] The display (220) can display text, images, videos, icons, or symbols that constitute various contents. The display (220) can include a touch screen and can receive touch, gesture, proximity, or hovering input using an electronic pen or a part of the user's body.

[0055] For example, the display (220) may include a liquid crystal display (LCD), a light emitting diode (LED) display, an organic light emitting diode (OLED) display, a microelectromechanical systems (MEMS) display, or an electronic paper display. The display (220) may be implemented as included in the electronic device (200), or may be implemented separately from the electronic device (200) but operatively connected to the electronic device (200).

[0056] The communication circuit (230) can establish a communication channel between the electronic device (200) and external devices. The communication circuit (230) can access a network (280) via wireless or wired communication to communicate with the external devices. In one embodiment, the communication circuit (230) may include a circuit for forming a wide area network connection or peer-to-peer connection, such as a Wi-Fi, Bluetooth, and / or cellular communication circuit.

[0057] The database (240) may be implemented on the memory (250) or on a separate storage medium. The database (240) may store all of the contents and details of data transmitted and received with an external device. The data stored in the database (240) may be regularly updated according to a predetermined cycle.

[0058] According to an embodiment of the present invention, the database (240) may store various information input from an external device. For example, the database (240) may store questionnaire data such as the age and gender of the subject, or examination data of the subject.

[0059] According to an embodiment of the present invention, the database (240) can store the examination data of a subject (patient) in chronological order under the control of the processor (270) described below to build a database for the examination data.

[0060] According to an embodiment of the present invention, the database (240) can separately construct a learning DB for learning data and a verification DB for verification based on a database for inspection data constructed under the control of the processor (270).

[0061] According to an embodiment of the present invention, the database (240) can store a model for predicting parameters of a refractive corneal correction lens under the control of the processor (270).

[0062] According to various embodiments, the data stored in the database (240) may be distributed and stored on a blockchain network to enhance the security of use of the information, as it contains sensitive information about the subject. When the database (240) is distributed and stored on a blockchain network, the history of transmission, modification, deletion, and addition of information contained in the database (240) can be more securely managed on the blockchain network.

[0063] The memory (250) may include volatile and / or non-volatile memory. The memory (250) may store instructions or data related to at least one other component of the electronic device (200). For example, the memory (250) may store instructions that, when executed, cause the processor (270) to perform various operations described herein. For example, the instructions may be included in a package file of an application program.

[0064] The I / O interface (260) can perform a role of transmitting commands or data input from a user or another external device to other components of the electronic device (200). The I / O interface (260) can be implemented in hardware or software, and can be used as a concept encompassing a user interface (UI) and a terminal for communication with other external devices.

[0065] The processor (270) may include at least one of a central processing unit (CPU), an application processor (AP), or a communication processor (CP). The processor (270) is electrically connected to the memory (250), the display (220), and the communication circuit (230) via the bus (210), and during operation, may execute operations or data processing related to control and / or communication of other components according to commands, programs, or software stored in the memory (250). Therefore, the execution of the commands, application programs, or software may be understood as the operation of the processor (270).

[0066] According to an embodiment of the present invention, the processor (270) can perform data processing for performing learning and verification of a model related to parameter prediction of a corneal refractive lens, which will be described later, and a model operation step. According to an embodiment of the present invention, the processor (270) can perform learning of a model related to parameter prediction of a corneal refractive lens, which will be described later, and can predict the parameters of a corneal refractive lens by utilizing a model related to parameter prediction of a corneal refractive lens.

[0067] The network (280) may include at least one of a telecommunications network, a computer network, the Internet, or a telephone network. A wireless communication protocol for accessing the network (280) may use, for example, at least one of LTE (Long-Term Evolution), LTE-A (LTE Advanced), CDMA (Code Division Multiple Access), WCDMA (Wideband CDMA), UMTS (Universal Mobile Telecommunications System), WiBro (Wireless Broadband), GSM (Global System for Mobile communications), or 5G standard communication protocols. However, this is merely exemplary, and various wired and wireless communication technologies applicable in the relevant technical field may be utilized depending on the embodiment to which the present invention is applied.

[0068] Figure 3 illustrates the relationship between devices according to one embodiment.

[0069] Referring to FIG. 3, an electronic device (300) (e.g., the first electronic device (110) of FIG. 1) may interact with an external device (310) (e.g., the second electronic device (120) or server (130) of FIG. 1) to diagnose corneal refraction. Alternatively, the electronic device (300) may interact with an external device (310) to predict parameters of a corneal refractive lens.

[0070] According to one embodiment, the electronic device (300) may include at least one of a processor (302), a memory (304), or a communication circuit (306). The processor (302) may control the communication circuit (306) to communicate with an external device (310). The communication circuit (306) may support wired or wireless communication. Additionally, the communication circuit (306) may support bidirectional or unidirectional communication.

[0071] According to one embodiment, the external device (310) may include at least one of a processor (312), a memory (314), or a communication circuit (316). The processor (312) may control the communication circuit (316) to communicate with the electronic device (310). The communication circuit (316) may support wired or wireless communication. Additionally, the communication circuit (316) may support bidirectional or unidirectional communication.

[0072] In one embodiment, the electronic device (300) and the external device (310) may have a client-server relationship. In one embodiment, the electronic device (300) may request the external device (310) to transmit a learning model (e.g., a model for predicting parameters of a corneal refractive lens) and, in response, may obtain the transmitted learning model.

[0073] Alternatively, the electronic device (300) may transmit input data to an external device (310), and the external device (310) may drive a learning model in response thereto. The external device (310) may obtain output data based on the learning model and transmit the obtained output data to the electronic device (300). For example, the electronic device (300) may transmit inspection data to the external device (310), and the external device (310) may drive a model for predicting parameters of a corneal refractive lens, and transmit information related to parameters of the corneal refractive lens generated using the inspection data to the electronic device (300).

[0074] Figure 4 illustrates a learning model according to one embodiment of the present invention.

[0075] A learning model according to one embodiment of the present invention (e.g., a model related to predicting parameters of a corneal refractive lens) may include a neural network for classifying input data (e.g., a fundus photograph or a retinal tomography photograph of a subject). According to an embodiment of the present invention, the neural network may be composed of a set of interconnected node units. A plurality of nodes may mean a plurality of neurons. According to an embodiment of the present invention, the nodes constituting the neural network may be connected by one or more links. In the neural network, one or more nodes connected by a link may form a relationship between an input node and an output node.

[0076] In a relationship between input nodes and output nodes connected through a single link, the value of the data of the output node can be determined according to the data input to the input node. According to an embodiment of the present invention, the link interconnecting the input node and the output node can have a weight. The weight can be variable. In addition, the neural network according to an embodiment of the present invention can vary the weight by a user or a certain algorithm in order to diagnose corneal refraction after surgery. For example, when one or more input nodes are connected to one output node by a link, the output node can determine the output node value based on the values ​​input to the input nodes connected to the corresponding output node and the weight set for the link corresponding to each input node.

[0077] According to an embodiment of the present invention, the performance of a neural network can be determined based on the number of nodes and links within the neural network, the relationships between the nodes and links, and the weight values ​​assigned to each link. For example, if two neural networks exist with the same number of nodes and links but different link weight values, the two neural networks can be recognized as different from each other.

[0078] A neural network can be composed of a set of multiple nodes, and a subset of the nodes constituting the neural network can be referred to as a layer. Some of the nodes constituting the neural network can form a layer based on the distance they form from a specific input node. For example, a set of multiple nodes that are 'n' distances from a specific input node can form an 'n'-layer. The distance from a specific input node refers to the number of links that must be passed through to reach the node from the specific input node. However, this definition of a layer is only an example and is not limited thereto.

[0079] According to one embodiment of the present invention, the number of nodes included in the input layer of the neural network may be the same as the number of nodes included in the output layer. According to another embodiment of the present invention, the number of nodes in the input layer may be different from the number of nodes in the output layer of the neural network.

[0080] A deep neural network (DNN) is a neural network that includes input and output layers, as well as at least one hidden layer. DNNs can be used to identify latent structures in data. DNNs may include convolutional neural networks (CNNs), recurrent neural networks (RNNs), generative adversarial networks (GANs), autoencoders, and deep belief networks (DBNs). However, these are merely examples and are not intended to be limiting.

[0081] Neural networks can be trained using at least one of the following methods: supervised learning, unsupervised learning, semi-supervised learning, reinforcement learning, or imitation learning. Neural network training can be the process of applying information (data) to the neural network to build a specific model for predicting post-surgical corneal refraction.

[0082] Neural networks can be trained based on training data to minimize output errors. Neural networks repeatedly receive training data and, based on the training data, calculate the error between the network output and target data. To reduce this error, neural networks can backpropagate the error from the output layer (or output node) to the input layer (or input node) to update the weights of each node in the network.

[0083] In supervised learning, a neural network can use training data (labeled data) with the correct answer for each training data point. In unsupervised learning, the neural network may not have the correct answer for each training data point. For example, training data used in supervised learning for data classification may be data with each category labeled. The neural network can input labeled data and compare the output of the neural network with the training data labels to calculate the error. In another example, for training data used in unsupervised learning for data classification, the neural network can compare the output of the neural network with the training data and calculate the error. The calculated error can be propagated backwards through the neural network. Through backpropagation, the neural network can update the weights corresponding to the links of each node in each layer of the neural network. The amount of change in the connection weights of each updated node can be determined by the learning rate. The neural network's calculation of the input data and the backpropagation of the error can constitute a learning cycle. The learning rate can be applied differently depending on the number of iterations of the learning cycle of the neural network.

[0084] In neural network training, the training data can typically be a subset of real-world data (i.e., eye state information to be processed using the trained neural network). While errors on the training data may decrease, errors on the real-world data may increase. Overfitting occurs when the model overtrains on the training data, resulting in increased errors on the real-world data. Overfitting can increase errors in machine learning algorithms.

[0085] To prevent overfitting, training data augmentation methods can be utilized. For example, linear data augmentation methods can be used, including image flipping, random position translation (-10% to 10%), random rotation (-15° to 15°), random magnification (-10% to 10%), or random brightness change (-10% to 10%).

[0086] The learning model may include at least one of a plurality of learning algorithms for predicting postoperative corneal refraction. For example, the algorithm may include at least one of linear regression, logistic regression, K-Nearest Neighbors, a support vector machine, a decision tree, a random forest, and a neural network.

[0087] A learning model can utilize multiple learning algorithms to calculate predicted values. For example, an ensemble method can be used in a learning model, which can achieve better prediction performance than using individual learning algorithms.

[0088] For example, the learning model of the present invention may include a convolutional layer and an artificial neural network. According to one embodiment, the convolutional layer may extract features from input data. In addition, the convolutional layer may include a filter that extracts features and an activation function that converts the value of the filter that extracts features into a non-linear value. The filter may include a function that detects whether the input data includes the feature. Once the feature value is extracted from the convolutional layer, the feature value may be input to the artificial neural network. Through this, the artificial neural network may classify the fundus photograph.

[0089] In one embodiment, the learning model may include a fully connected layer and a softmax function at the end. The output data may include probability values ​​between 0 and 1.

[0090] In one embodiment, a neural network can be used to learn patterns according to embodiments of the present invention, in conjunction with transfer learning. For example, a neural network can be used to learn patterns for classifying or predicting corneal refraction. Transfer learning refers to learning a new model using information from a previously used model. By applying transfer learning, the performance of the neural network can be improved.

[0091] Learning results according to the present invention can generate an attention map. This attention map can be generated, for example, from a softmax function and an activated convolutional layer. Techniques such as Grad-CAM heatmaps can be applied to the attention map.

[0092] FIG. 5 conceptually illustrates the operation of an electronic device (130) for predicting parameters of a corneal refractive lens according to one embodiment. In particular, FIG. 5 conceptually illustrates the operation of an electronic device (130) for predicting parameters of a corneal refractive lens classifying data for learning a parameter prediction model of a corneal refractive lens.

[0093] An electronic device (130, hereinafter, electronic device) for predicting parameters of a corneal refractive lens can acquire training data to build a model for predicting parameters of a corneal refractive lens, and can preprocess the acquired training data. That is, the electronic device (130) can select training data from the acquired data according to predetermined criteria.

[0094] According to an embodiment of the present invention, the electronic device (130) can acquire examination data of a subject (patient). The examination data may include both examination data before and after corneal refractive correction surgery on the subject's eyes.

[0095] Pre-procedure examination data may include data generated by conducting an eye examination on a subject before receiving corneal refractive correction surgery.

[0096] In one embodiment, the pre-procedure examination data may include data obtained from an ocular measurement device for the subject. The ocular measurement device may be a variety of devices, such as tomography, topography, optical coherence tomography (OCT), ultrasound biomicroscopy (UMB), or other devices, such as Pentacam, CASIA2, AL-Scan, or OQAS.

[0097] In one embodiment, the pre-procedure examination data may include information related to corneal refraction actually diagnosed based on data obtained from an ophthalmoscope. For example, the pre-procedure examination data may include information related to corneal refraction diagnosed based on a keratometer obtained using an ophthalmoscope.

[0098] In addition, the pre-procedure examination data may include various types of parameters as multiple examination data. For example, the pre-procedure examination data may include the patient's SE (Spherical Equivalent), Km (mean Keratometry), pupil size, IOP (Intraocular Pressure), WHW (White-to-white), CCT (Central Corneal Thickness), ATA (Angle to Angle Distance), ACD (Anterior Chamber Depth), ACW (Anterior Chamber Width), CLR (Crystalline Lens Rise), Lens Vault, AOD (Angle Opening Distance), or TISA (Trabecular Iris Surface Area) before the procedure.

[0099] Pre-procedure examination data may include questionnaire data. Specifically, this may include data such as the patient's age and gender.

[0100] Pre-procedure examination data may include the patient's medical history. For example, this may include data on past eyeglass prescriptions, vision test results, and refractive errors (myopia, astigmatism, hyperopia).

[0101] Post-procedure data may include parameter values ​​of corneal refractive lenses determined by a medical professional (e.g., an ophthalmologist) based on the subject's pre-procedure data. The examination data of the present invention refers to linked data in which pre-procedure data and post-procedure data are mapped.

[0102] According to an embodiment of the present invention, inspection data can be stored in a database (240) taking time into consideration.

[0103] The electronic device (130) can designate test data stored prior to a preset point in time among test data arranged in chronological order as training data. According to an embodiment of the present invention, the preset point in time can be set by the designer. According to various embodiments of the present invention, the preset point in time can be determined based on the point in time at which the last test data was stored in the database (240).

[0104] The electronic device (130) can individually store designated training data in the database (240). Alternatively, the electronic device (130) can use a separate mark (e.g., a flag shape, etc.) on the previously stored test data to distinguish between training data and non-training data within the database (240).

[0105] According to an embodiment of the present invention, the electronic device (130) may designate relatively recently stored inspection data among the training data as internal verification data. For example, the electronic device (130) may designate inspection data stored after a preset point in time among the training data as internal verification data. According to an embodiment of the present invention, the preset point in time may be set by the designer. According to various embodiments of the present invention, the preset point in time may be determined based on the point in time at which the most recent inspection data among the training data was stored in the database (240).

[0106] According to various embodiments of the present invention, the electronic device (130) can randomly select some of the training data and designate them as internal verification data.

[0107] The electronic device (130) may designate as external verification data the inspection data (i.e., non-training data) that is not selected as training data among the inspection data stored in the database (240). The external verification data may be inspection data stored in the database (240) relatively more recently than the training data.

[0108] FIG. 6A is a flowchart illustrating the operation of an electronic device (130) according to an embodiment of the present invention. In particular, FIG. 6A illustrates an operation of constructing a model for predicting parameters of a corneal refractive lens according to an embodiment of the present invention.

[0109] Meanwhile, FIGS. 6b to 6d are conceptual diagrams illustrating the operation of an electronic device (130) for predicting parameters of a corneal refractive lens according to an embodiment of the present invention. In particular, FIG. 6b illustrates an operation in which the electronic device (130) generates or learns a model for predicting parameters of a corneal refractive lens. In particular, FIGS. 6c and 6d illustrate an operation in which the electronic device (130) evaluates the performance of a model for predicting parameters of a corneal refractive lens.

[0110] The operation of the electronic device (130) described below is an operation performed by the processor (270), but for convenience of explanation, it is initiated as an operation of the electronic device (130).

[0111] Referring to FIG. 6A, in operation S610, the electronic device (130) may collect training data. According to one embodiment of the present invention, the electronic device (130) may receive training data from an external device. Alternatively, the electronic device (130) may collect training data on its own. The training data may include test data. According to various embodiments of the present invention, the electronic device (130) may collect training data by utilizing data augmentation, cycle-GAN, etc. based on test data stored in the database (240).

[0112] In operation S620, the electronic device (130) may process a training data set based on the training data. According to one embodiment, the electronic device (130) may classify each training data set into domains, types, and clusters. The electronic device (130) may classify sample data into domains, types, and clusters based on certain criteria, and generate a learning data set.

[0113] In operation S630, the electronic device (130) can build a model for predicting parameters of the subject's corneal refractive lens based on a training data set.

[0114] According to one embodiment of the present invention, the electronic device (130) may generate or train a model for predicting parameters of a corneal refractive lens using training data so that the model has a criterion for predicting the parameter value from input data. Specifically, the electronic device (130) may generate or train a model for predicting parameters of a corneal refractive lens using training data and statistical analysis of the subject's pre- and post-procedure examination data. Specifically, the electronic device (130) may generate or train the prediction model based on a statistical analysis of the correlation between the pre- and post-procedure examination data.

[0115] The electronic device (130) can train a model to predict the parameters of a corneal refractive lens using various methods, such as supervised learning, unsupervised learning, reinforcement learning, or imitation learning.

[0116] According to one embodiment of the present invention, a model for predicting parameters of a corneal refractive lens of the present invention may include a CNN. As a CNN structure, at least one of EfficientNet, AlexNet, LENET, NIN, VGGNet, ResNet, WideResnet, GoogleNet, FractaNet, DenseNet, FitNet, RitResNet, HighwayNet, MobileNet, or DeeplySupervisedNet may be used. The model for predicting parameters of a corneal refractive lens may also include an RNN.

[0117] In one embodiment, a model for predicting parameters of a corneal refractive lens may be implemented including a classifier. The classifier may include a binary classifier or a multi-classifier. In one embodiment, the classifier may utilize an algorithm such as a decision tree, a support vector machine, or a random forest.

[0118] In one embodiment, a model for predicting parameters of a corneal refractive lens can be implemented by including a regression model. The regression model can derive parameter values ​​in the form of probabilities. Each probability can sum to a value of 1. The regression model can utilize various algorithms, such as linear regression, regression trees, support vector regression, and kernel regression.

[0119] Referring to FIG. 6b, according to an embodiment of the present invention, the electronic device (130) can generate and train a model for predicting parameters of a corneal refractive lens by statistically analyzing the interrelationship between the data before the procedure, such as age, gender, unaided eye, corrected visual acuity, refractive power by objective examination, refractive power by autorefraction, corneal diameter, corneal vertical / horizontal / average eccentricity, corneal topography image, corneal and corneal epithelial thickness by region according to ocular optical coherence tomography image, axial length, central corneal thickness, anterior chamber depth, corneal diameter, pupil size, etc. according to optical laser ophthalmometry, and the parameter data of the final prescription lens among the data after the procedure.

[0120] According to various embodiments of the present invention, the electronic device (130) can individually generate and train a model for predicting the parameters of a corneal refractive lens through statistical analysis of the correlation between the parameter data of the final prescription lens and each of the various data included in the pre-procedure data. For example, the electronic device (130) can generate and train a model for predicting the parameters of a first corneal refractive lens through statistical analysis of the correlation between age and the parameter data of the final prescription lens. As another example, the electronic device (130) can generate and train a model for predicting the parameters of a second corneal refractive lens through statistical analysis of the correlation between gender and the parameter data of the final prescription lens. In this manner, the electronic device (130) can generate and train a model for predicting the parameters of n (where n is a natural number) corneal refractive lenses.

[0121] In operation S640, the electronic device (130) can evaluate the performance of a model for predicting parameters of a corneal refractive correction lens.

[0122] Referring to FIG. 6C, according to one embodiment of the present invention, the electronic device (130) can use internal verification data to determine the performance of a model for predicting parameters of a corneal refractive lens. Specifically, the electronic device (130) can compare data output based on a model for predicting parameters of a corneal refractive lens with internal verification data classified among previously constructed test data to calculate an error. The electronic device (130) can update the model for predicting parameters of a corneal refractive lens by reflecting the calculated error.

[0123] Referring to FIG. 6D, according to various embodiments of the present invention, the electronic device (130) can use external verification data to determine the performance of a model for predicting parameters of a corneal refractive lens. Specifically, the electronic device (130) can compare data output based on a model for predicting parameters of a corneal refractive lens with external verification data classified among previously constructed test data to calculate an error. The electronic device (130) can update the model for predicting parameters of a corneal refractive lens by reflecting the calculated error.

[0124] According to an embodiment of the present invention, the electronic device (130) can determine the performance of a model for predicting parameters of a corneal refractive lens by utilizing both external verification data and internal verification data.

[0125] According to various embodiments of the present invention, the electronic device (130) can compare data output based on a model for predicting parameters of a corneal refractive lens with training data to calculate an error. The electronic device (130) can periodically calculate the error to evaluate the performance of the learning model.

[0126] Although not illustrated in the drawing, the electronic device (130) can update a model for predicting parameters of a corneal refractive lens in a direction that reduces the calculated error. For example, the electronic device (130) can obtain a result value (output data) using a model to which arbitrary weights are assigned, compare the obtained result value (output data) with the labeling data of the training data, and perform backpropagation based on the error, thereby optimizing the weights.

[0127] As a result of performing operations S610 to S640, the electronic device (130) can build, learn, and improve a model for predicting parameters of a corneal refractive correction lens.

[0128] FIG. 7 illustrates an architecture (700) for predicting parameters of a corneal refractive lens according to one embodiment.

[0129] According to one embodiment, the architecture (700) may be implemented by a processor (270) of an electronic device (130).

[0130] According to one embodiment, the parameter prediction module (710) of the corneal refractive lens may include a model for predicting the parameter of the corneal refractive lens. The model for predicting the parameter of the corneal refractive lens may be a learning model configured to predict the parameter of the corneal refractive lens based on input data. The model for predicting the parameter of the corneal refractive lens may be learned based on examination data of the subject.

[0131] According to various embodiments of the present invention, a model for predicting parameters of a corneal refractive lens may be pre-trained on a different electronic device or pre-trained on the same electronic device based on input data before being placed on the electronic device (130).

[0132] The input data may include examination data of the subject. The examination data may include examination data obtained from an ocular measurement device for the subject. Furthermore, the examination data may include various types of parameters as multiple examination data. For example, the examination data may include at least one of the following data: SE, Km, pupil size, IOP, WHW, CCT, ATA, ACD, ACW, CLR, lens bolting value, AOD, TISA, corneal refraction, etc.

[0133] The examination data may include questionnaire data. Specifically, it may include data such as the subject's age and gender.

[0134] The examination data may include the subject's medical records. For example, the examination data may include data on the prescription of glasses worn in the past, the results of vision tests, and the degree of refractive error (myopia, astigmatism, hyperopia).

[0135] The examination data may include measurement data about the cornea. For example, it may include corneal shape, corneal symmetry, corneal thickness measurement data, corneal structural tomography data, corneal topography data, corneal refraction, and corneal endothelial cell examination data.

[0136] The examination data may include measurements of visual acuity and / or refraction. For example, this may include data on the prescription of previously worn glasses, vision test results, and refractive errors (myopia, astigmatism, hyperopia).

[0137] The examination data may include measurement data regarding distances within the eye, etc. Specifically, it may include pupil size, eye length, and the distance of the space where the lens will be inserted.

[0138] The examination data may include data on ocular diseases and / or underlying conditions. For example, it may include data on the presence or absence of retinal diseases, such as glaucoma and retinal degeneration, cataracts, and diseases of the posterior surface of the iris.

[0139] In addition, input data can include various types of data. For example, it can include test data such as eye-related genes and blood tests.

[0140] The output data may include parameter data of a corneal refractive lens for a subject. According to an embodiment of the present invention, the output data may include any one of a plurality of predetermined categories related to parameters of the corneal refractive lens. For example, the output data may indicate at least one of an overall diameter (OAD) of the cornea, a base curve (BC), a return zone depth (RZD), and a landing zone angle (LZA).

[0141] According to one embodiment of the present invention, the output data may further include a probability value.

[0142] According to one embodiment of the present invention, the output data may include all of the plurality of categories, but may further include probability values ​​for each category.

[0143] The configuration of FIG. 7 is exemplary and is not intended to limit the scope of the present invention. A device according to the present invention may include only some of the configurations of FIG. 7, or may include additional or different configurations compared to the configurations illustrated in FIG. 7.

[0144] FIG. 8 illustrates an architecture (800) for predicting parameters of a corneal refractive lens according to various embodiments of the present invention. According to an embodiment of the present invention, the architecture (800) may be implemented by a processor (270) of an electronic device (130).

[0145] The first corneal refractive lens parameter prediction module (810_1) may include a parameter prediction model of the first corneal refractive lens (hereinafter, referred to as the first prediction model). The first prediction model may be a learning model configured to predict the parameter of the corneal refractive lens based on first input data. The first prediction model may be learned through statistical analysis of the correlation between any one of the subject's pre-treatment data and the parameter of the final prescription lens.

[0146] By the same logic, the second corneal refractive lens parameter prediction module (810_2) may include a parameter prediction model of the second corneal refractive lens (hereinafter, referred to as the second prediction model). The second prediction model may be a learning model configured to predict the parameter of the corneal refractive lens based on second input data. The second prediction model may be learned by statistical analysis of the correlation between any one of the subject's pre-treatment data (data different from that utilized in the first prediction model) and the parameter of the final prescription lens.

[0147] According to one embodiment of the present invention, an architecture (800) for predicting parameters of a corneal refractive lens may include n corneal refractive lens parameter prediction modules, from a first corneal refractive lens parameter prediction module (810_1) to an n-th corneal refractive lens parameter prediction module (810_n, n is a natural number).

[0148] Each corneal refractive lens parameter output from the prediction module can be input into the common module (820).

[0149] According to an embodiment of the present invention, the common module (820) can set different weights for each corneal refractive lens parameter output from the prediction module to produce final parameter data of the corneal refractive lens. For example, the common module (820) can set different weights for each corneal refractive lens parameter output from the prediction module by utilizing at least one of the test data not utilized in the prediction module for each corneal refractive lens parameter, such as the age and gender of the subject.

[0150] The final output data may include parameter data of the corneal refractive lens for the subject. According to an embodiment of the present invention, the output data may include any one of a plurality of predetermined categories related to the parameters of the corneal refractive lens. For example, the output data may indicate at least one of OAD, BC, RZD, and LZA.

[0151] According to one embodiment of the present invention, the output data may further include a probability value.

[0152] According to one embodiment of the present invention, the output data may include all of the plurality of categories, but may further include probability values ​​for each category.

[0153] Although all components constituting the embodiments of the present invention have been described as being combined or operating in combination, the present invention is not necessarily limited to such embodiments. That is, within the scope of the present invention, all components may be selectively combined and operated one or more times.

[0154] Meanwhile, the various embodiments described herein may be implemented by hardware, middleware, microcode, software, and / or a combination thereof. For example, the various embodiments may be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, other electronic units designed to perform the functions presented herein, or a combination thereof.

[0155] Additionally, for example, various embodiments may be embodied or encoded in a computer-readable medium containing instructions. Instructions embodied or encoded in the computer-readable medium may cause a programmable processor or other processor to perform a method when the instructions are executed, for example. The computer-readable medium includes a computer storage medium, which may be any available medium that can be accessed by a computer. For example, such a computer-readable medium may include a RAM, a ROM, an EEPROM, a CD-ROM or other optical disk storage medium, a magnetic disk storage medium, or other magnetic storage devices.

[0156] Such hardware, software, firmware, etc. may be implemented within the same device or within separate devices to support the various operations and functions described herein. Additionally, components, units, modules, components, etc. described as “units” in the present invention may be implemented together or individually as separate but interoperable logic devices. The depiction of different features for modules, units, etc. is intended to highlight different functional embodiments and does not necessarily imply that they must be realized by separate hardware or software components. Rather, the functionality associated with one or more modules or units may be performed by separate hardware or software components, or integrated into common or separate hardware or software components.

[0157] Although operations are depicted in the drawings in a particular order, this should not be construed as requiring that these operations be performed in the particular order depicted, or in any sequential order, or that all depicted operations be performed to achieve the desired results. In certain circumstances, multitasking and parallel processing may be advantageous. Furthermore, the distinction between various components in the embodiments described above should not be construed as requiring such distinction in all embodiments, and it should be understood that the components depicted may generally be integrated together into a single software product or packaged into multiple software products.

[0158] The electronic device, server, or external device according to the various embodiments of the present document described above may include, for example, at least one of a smartphone, a tablet PC, a mobile phone, a video phone, a desktop PC, a laptop PC, a personal digital assistant (PDA), a portable multimedia player (PMP), an MP3 player, a mobile medical device, a camera, or a wearable device.

[0159] According to various embodiments, the wearable device may include at least one of an accessory type (e.g., a watch, a ring, a bracelet, an anklet, a necklace, glasses, contact lenses, or a head-mounted device (HMD)), a fabric or clothing-integrated type (e.g., an electronic garment), a body-attached type (e.g., a skin pad or tattoo), or a bio-implant type (e.g., an implantable circuit).

[0160] In some embodiments, the electronic device or external device may be a home appliance. The home appliance may include, for example, at least one of a television, a digital video disk player (DVD player), an audio device, a refrigerator, an air conditioner, a vacuum cleaner, an oven, a microwave oven, a washing machine, an air purifier, a set-top box, a home automation control panel, a security control panel, a TV box, a game console, an electronic dictionary, an electronic key, a camcorder, or an electronic picture frame.

[0161] In another embodiment, the electronic device, external device, or wearable device may include at least one of various medical devices (e.g., various portable medical measuring devices (such as a blood glucose meter, a heart rate meter, a blood pressure meter, or a body temperature meter), magnetic resonance angiography (MRA), magnetic resonance imaging (MRI), computed tomography (CT), a camera, or an ultrasound machine), a navigation device, a satellite navigation system (Global Navigation Satellite System (GNSS)), an event data recorder (EDR), a flight data recorder (FDR), an automobile infotainment device, a home robot, or an internet of things device (e.g., a light bulb, various sensors, an electric or gas meter, a sprinkler device, a fire alarm, a thermostat, a streetlight, an exercise machine, a hot water tank, a heater, a boiler, or the like).

[0162]

[0163] As described above, the best practice embodiments have been disclosed in the drawings and specifications. While specific terminology has been used herein, it is solely for the purpose of describing the present invention and is not intended to limit the scope of the invention as defined in the claims. Therefore, those skilled in the art will understand that various modifications and equivalent embodiments are possible. Therefore, the true technical protection scope of the present invention should be determined by the technical spirit of the appended claims.

Claims

1. A method for predicting parameters of a corneal refractive correction lens, A step of collecting the subject's examination data; A step of creating a parameter prediction model of a corneal refractive correction lens using the above collected inspection data; Step of receiving input data; A step of applying the above input data to the above prediction model to predict at least one parameter of the corneal refractive correction lens. Including, The parameters of the above corneal refractive lens include the overall corneal diameter (OAD), curvature radius (BC), transition zone depth (RZD), and landing zone angle (LZA). A method for predicting parameters of corneal refractive lenses.

2. In claim 1, The above examination data includes examination data before and after the procedure for the subject. A method for predicting parameters of corneal refractive lenses.

3. In claim 2, A step for selecting the inspection data stored in the database from the above inspection data before the preset time as training data. Including more, The steps for creating the above prediction model are Learning the prediction model by utilizing the above training data, A method for predicting parameters of corneal refractive lenses.

4. In claim 3, A step of randomly selecting data for internal validation from the above training data; Step of judging the performance of the prediction model by using the above internal verification data A method for predicting parameters of a corneal refractive lens including:

5. In claim 3, A step of selecting data among the above inspection data that is not selected as training data as external verification data; A step for judging the performance of the prediction model by utilizing the external verification data above. A method for predicting parameters of a corneal refractive lens including:

6. In claim 1, The above input data is Age, sex, uncorrected visual acuity, refractive power by objective examination, refractive power by autorefraction, corneal diameter, corneal vertical / horizontal / average eccentricity, corneal topography image, corneal and corneal epithelial thickness by region based on ocular optical coherence tomography image, axial length based on optical laser ophthalmometry, central corneal thickness, anterior chamber depth, corneal diameter, and pupil size, including at least one of the following. A method for predicting parameters of corneal refractive lenses.

7. In claim 2, The steps for creating the above prediction model are The prediction model is created based on statistical analysis of the correlation between the test data before and after the above procedure. A method for predicting parameters of corneal refractive lenses.

8. In an electronic device for predicting parameters of a corneal refractive correction lens, A memory that stores commands to perform various actions; and comprising at least one processor electrically connected to said memory; The above processor, Collecting examination data of a subject, creating a parameter prediction model of a corneal refractive lens using the collected examination data, receiving input data, and applying the input data to the prediction model to predict at least one parameter of the corneal refractive lens, The parameters of the above corneal refractive lens include the overall corneal diameter (OAD), curvature radius (BC), transition zone depth (RZD), and landing zone angle (LZA). An electronic device for predicting parameters of a corneal refractive lens.

Images