KR102303522B1 - Method for determining lens and apparatus using the method - Google Patents

Abstract

A method of determining a lens and an apparatus using the same are disclosed. According to an embodiment, in a method for determining a lens to be inserted into the eye during lens insertion using machine learning, obtaining a plurality of examination data of a prospective surgery and determining a lens by the obtained plurality of examination data of a prospective surgery and determining the size of a lens to be inserted into the eye of the person to be operated on from among a plurality of lens sizes by inputting it into a model, wherein the lens determination model is different from a formula for determining a lens to be inserted into the eye during the lens insertion operation It is different, and it is possible to provide a lens determination method that is learned based on examination data of an operator who has undergone lens insertion in the past and size information of a lens to which the operator's intraocular insertion is applied.

Description

Lens determination method and apparatus using the same

The following embodiments relate to a method for determining a lens used for lens implantation and an apparatus using the same. More specifically, it relates to a method and apparatus for determining a lens used for lens implantation using artificial intelligence.

Intraocular lens implantation, which is one of the surgical methods for correcting the poor visual acuity due to refractive error, is to insert a special lens designed to correct the refractive error into the normal eye with the lens.

In the prior art, when inserting a lens, a lens is selected using a program provided by a lens manufacturer, which uses only the basic eye-related numerical values of the surgical person as input data, and the lens size and power without considering the characteristics of the patient's eyes. is decided Programs provided by lens manufacturers are generally created based on simple formulas, in which only basic eye-related figures of the prospective surgery are used as input data. As a result, there is a problem that the prospective surgery has to undergo reoperation, such as removing the lens, because many side effects such as cataract and glaucoma occur due to the insertion of an inappropriate size and frequency.

Recently, various studies are being conducted to prevent side effects and improve the quality of vision in relation to lens implantation.

delete

1. US Patent Application Publication No. 2018-0161098 (2018.06.14)

One object of the present invention relates to providing information about an implantable lens suitable for the ocular characteristics of a prospective lens implantation surgery using machine learning.

One task is to provide a lens determination assistance system and process using artificial intelligence.

One task is to determine a more suitable lens for the prospective surgery in consideration of the ocular characteristics of each prospective lens implantation surgery, reduce the possibility of side effects of lens insertion, and provide improved visual acuity.

The problem to be solved by the present invention is not limited to the above-mentioned problems, and the problems not mentioned will be clearly understood by those of ordinary skill in the art to which the present invention belongs from the present specification and the accompanying drawings. .

According to an exemplary embodiment, a method of determining a lens to be inserted into the eye of a surgical patient during lens implantation may be provided through a lens determination model learned using machine learning.

Solutions of the present invention are not limited to the above-described solutions, and solutions not mentioned will be clearly understood by those of ordinary skill in the art to which the present invention belongs from the present specification and the accompanying drawings. will be able

According to one embodiment, by determining a lens size and power suitable for the eye of a person scheduled for lens implantation surgery, it is possible to minimize the occurrence of side effects after surgery.

According to one embodiment, by determining a lens size and power suitable for the eye of a person scheduled to undergo lens implantation surgery, reoperation of lens implantation surgery can be prevented.

The effects of the present invention are not limited to the above-mentioned effects, and the effects not mentioned will be clearly understood by those of ordinary skill in the art to which the present invention belongs from the present specification and the accompanying drawings.

1 is a view for explaining a position at which a lens is inserted during lens insertion.
2 is a diagram illustrating a lens determination assisting system according to an exemplary embodiment.
3 is a diagram illustrating a lens determination assisting system according to another exemplary embodiment.
4 is a diagram illustrating a lens determination assisting system according to another exemplary embodiment.
5 is a diagram schematically illustrating a lens determination assistance system using a server.
6 is a diagram illustrating a relationship between a server device and a client device.
7 is a diagram illustrating a lens determination model according to an exemplary embodiment.
8 is a diagram of a lens determination assistance process according to an embodiment.
9 is a diagram of a lens size determination module according to an embodiment.
10 is a view for explaining the side effects of the lens implantation surgery according to an embodiment.
11 is a diagram of a lens size determination process according to an embodiment.
12 is a diagram illustrating an embodiment of determining a lens size.
13 is a diagram illustrating an embodiment of determining a lens size.
14 is a diagram illustrating another embodiment of determining a lens size.
15 is a diagram illustrating another embodiment of determining a lens size.
16 is a diagram illustrating another embodiment of determining a lens size.
17 is a diagram schematically illustrating a plurality of sub-models of a lens size determination model.
18 is a diagram for defining a bolting value.
19 is a diagram of a bolting value prediction module according to an embodiment.
20 is a diagram of a bolting value prediction process according to an embodiment.
21 is a diagram illustrating an embodiment of predicting a bolting value.
22 is a diagram illustrating another embodiment of predicting a bolting value.
23 is a diagram of a lens power determination module according to an exemplary embodiment.
24 is a diagram of a lens power determination process according to an embodiment.
25 is a view for explaining an example that occurs during a corneal incision process of a prospective surgery.
26 is a diagram illustrating an embodiment of determining the lens power.
27 is a diagram illustrating another embodiment of determining the lens power.

A method of determining a lens to be inserted into the eye during lens implantation using machine learning according to an embodiment includes: acquiring a plurality of examination data of a prospective surgery; and inputting the obtained plurality of examination data of the prospective surgery to a lens decision model, and determining a size of a lens to be inserted into the eye of the surgery prospect from among a plurality of lens sizes. It is different from a formula for determining a lens to be inserted into the eye during lens insertion, and may be learned based on examination data of an operator who has undergone lens insertion in the past and size information of a lens to which the operator's intraocular insertion is applied.

The plurality of examination data of the surgical candidate includes any one of first data and second data, and the priority of input data that is input to the lens determination model to increase the accuracy of the determined lens size is the first data may be higher than the second data.

The determining of the lens size may include inputting the second data into the lens determination model even when the examination data of the surgery prospect does not include the first data and includes the second data, and The size of the lens to be inserted into the eye is determined, but the accuracy of the lens size determined when the examination data of the surgery prospect includes the first data is the second data, not the first data. It may be higher than the accuracy of the lens size determined when including .

In the step of determining the lens size, a reliability for the accuracy of the lens size derived according to the examination data of the surgical person may be calculated, and the calculated reliability may be presented to the user to determine the lens size.

In the method for determining a lens according to an embodiment, when the examination data of the surgical patient excludes the first data and includes the second data or other data, the first data from the second data or other data Further comprising the step of estimating; the step of estimating, the accuracy of the lens size derived when the first data is input to the lens decision model is that the second data is input to the lens decision model In this case, it may be higher than the accuracy for the derived lens size.

The first data among the plurality of examination data of the surgical candidate is ATA (Anterior Chamber Angle), ACD-epi (Anterior Chamber Depth), ACD-endo, CCT (Central Corneal Thickness), A crystalline lens rise (CLR), WTW, Axial Length, BUT, a distance measurement between the iris, and a space size value in which the lens is to be inserted are obtained, and the second data may be obtained using a general ophthalmic examination.

In the step of determining the lens size, the size of the lens to be inserted into the eye of the surgical patient may be determined as any one of a plurality of predetermined lens sizes.

In the determining of the lens size, the lens size may be determined as any one of non-standardized lens sizes derived by inputting the inspection data to the lens determination model, rather than a plurality of predetermined lens sizes.

The method for determining a lens according to an embodiment further includes: inputting the obtained examination data of the prospective surgery to a lens decision model to determine the number of lenses to be inserted into the eye of the surgeon from among a plurality of lens powers; In the step of determining the lens power, when the lens to be determined by the lens determination model is inserted into the eye of the surgery prospect, the lens power is determined so that the target visual acuity of the surgery candidate is derived, and the lens determination model is , can be learned based on the examination data of the operator who has undergone lens implantation in the past and the incision information of the operator.

The acquired examination data of the prospective surgery includes diopter, astigmatism axis, and astigmatism direction parameters measured from the eyes of the prospective surgery, and the test data of the prospective surgery and the corneal incision of the lens implantation in the lens determination model By inputting incision information expected in the process, it is possible to determine a lens power suitable for the target visual acuity of the surgeon.

When all the examination data of the prospective surgery is input to the lens determination model, the lens power for deriving the target visual acuity of the prospective surgery and incision information expected in the corneal incision process of the prospective surgery may be determined.

The incision information includes at least one of a corneal incision method, a corneal incision position, a corneal incision direction and/or a corneal incision degree, a coma position, corneal astigmatism and lens astigmatism, myopia, and astigmatism in the corneal incision process of the lens implantation surgery. It may be information including one.

An apparatus for determining a lens to be inserted into the eye during lens insertion using machine learning according to an embodiment comprises: a memory for storing a plurality of examination data of a prospective surgery; and a processor, wherein the processor obtains a plurality of examination data of the surgical candidate stored from the memory, and inputs the obtained plurality of examination data of the surgical candidate to the lens determination model, among a plurality of lens sizes The size of the lens to be inserted into the eye of the surgical candidate is determined, and the lens determination model is different from the formula for determining the lens to be inserted into the eye during the lens insertion operation, and the examination data of the operator who has undergone lens insertion in the past and It may be learned based on the size information of the lens to which the operator's intraocular insertion is applied.

A method of predicting a bolting value indicating a distance between a rear surface of a lens to be inserted into the eye of a prospective operation of lens implantation according to an embodiment and the front of the lens is a method of predicting a bolting value representing a distance between a plurality of examination data and at least one lens size of the surgical candidate input to the predictive model; and predicting the bolting value corresponding to each of the at least one input lens size from the bolting value prediction model, wherein the bolting value prediction model includes a plurality of examination data of an operator who has undergone lens implantation in the past. , can be learned based on the size of the lens inserted into the operator's eye and the bolting value measured after the operator's surgery.

The bolting value may be defined as the shortest distance among a plurality of distances between a rear surface of a lens to be inserted into the eye of a prospective person for lens implantation and an front surface of the lens.

The step of predicting the bolting value; the step of providing information on the suitability of the lens to be inserted into the eye of the patient of the inputted lens size according to whether the predicted bolting value satisfies a condition of a predetermined range ; may be included.

When the predicted bolting value satisfies the condition of a predetermined range, the input lens size provides information about suitability as a lens to be inserted into the eye of the surgical person, and the predicted bolting value is within the predetermined range When the condition of , the input lens size is not satisfied as a lens to be inserted into the eye of the surgical candidate, information about inappropriateness may be provided.

The condition of the predetermined range may satisfy that the predicted bolting value is included within 250 to 750 μm.

A method of predicting a bolting value indicating a distance between a rear face of a lens to be inserted into the eye of a prospective operation of lens implantation according to an embodiment and the front of the lens includes: inputting a plurality of test data of the surgical candidate into a bolting value prediction model ; and predicting an expected lens size suitable for the eye of the surgery prospect and the bolting value corresponding to each of the expected lens size from the bolting value prediction model by inputting a plurality of examination data of the surgery prospect. The bolting value prediction model may be learned based on a plurality of examination data of a patient who has undergone a previous operation of the lens implantation surgery, a lens size inserted into the patient's eyeball, and a bolting value measured after the operation of the patient.

The predicted lens size suitable for the eye of the surgical patient may include selecting any one of a plurality of predetermined lens sizes.

The predicted lens size suitable for the eye of the surgical patient may include selecting any one of non-standardized lens sizes, not a plurality of predetermined lens sizes.

An apparatus for predicting a bolting value indicating a distance between a rear surface of a lens to be inserted into an eye of a prospective operation of lens implantation according to an embodiment and the front of a lens includes a memory for storing a plurality of examination data of the surgical candidate; and a processor, wherein the processor inputs a plurality of examination data and at least one lens size of the surgery prospect to a bolting value prediction model, and the at least one input lens size from the bolting value prediction model. The corresponding bolting value is predicted, and the bolting value prediction model is based on a plurality of examination data of an operator who has undergone lens implantation in the past, a lens size inserted into the operator's eye, and a bolting value measured after the operator's surgery can be learned with

Hereinafter, specific embodiments of the present invention will be described in detail with reference to the drawings. However, the spirit of the present invention is not limited to the presented embodiments, and those skilled in the art who understand the spirit of the present invention may add, change, delete, etc. other elements within the scope of the same spirit, through addition, change, deletion, etc. Other embodiments included within the scope of the invention may be easily proposed, but this will also be included within the scope of the invention.

In addition, components having the same function within the scope of the same idea shown in the drawings of each embodiment will be described using the same reference numerals.

1 Definition of terms

(1) Lens implantation

Lens implantation is one of the surgical methods for correcting the lowered naked eye vision due to refractive error, and it is an operation to insert a special lens designed to correct refractive error in a normal eye with a crystalline lens.

1 is a view for explaining a position at which a lens is inserted during lens insertion. There are two types of lens implantation surgery: anterior lens insertion, which inserts a lens between Co (cornea) and I (iris), and posterior lens insertion, which inserts a lens into the space between the back of the iris and the lens. Referring to FIG. 1 , the lens may be inserted at the IN1 (first lens insertion) position in the rear lens insertion procedure, and the lens may be inserted at the IN2 (second lens insertion) position in the front lens insertion procedure. Hereinafter, for convenience of description, the present invention will be described with a focus on posterior lens insertion, also called an implantable collamer lens (ICL). However, the present invention is not limited thereto, and it goes without saying that the present invention can be applied to anterior lens implantation.

(2) lens

As used herein, a lens refers to an intraocular lens used for lens insertion, and may be distinguished from a hard lens and a soft lens worn on the ocular surface.

The information about the lens may include information about a lens size and a lens dioptric power. The lens size may include a plurality of lens sizes. The lens power may have a plurality of lens powers. In addition, the expression "determining a lens" may mean determining any one of a combination of a plurality of lens sizes and a plurality of lens powers.

(3) lens crystal model

The lens determination model refers to an algorithm and/or model that determines an intraocular lens to be inserted into the eye using artificial intelligence. The lens determination model to be described below is a model for deriving information about a lens to be inserted into the eye of a prospective surgery as output data corresponding to the input data when the examination data of a prospective surgery is input through the model as input data. . Hereinafter, the configuration, generation process, and judgment process of the lens decision model will be described in detail.

(4) learning

Learning refers to the process of learning the lens decision model based on the training data and the labeling data or the unlabeled data, so that the lens decision model can determine the output data with respect to the input data. That is, the lens decision model determines by forming a rule for the data.

The lens determination model may be trained through training data. Learning the lens decision model means adjusting the weight of the model.

As a learning method, various methods such as supervised learning, unsupervised learning, reinforcement learning, and imitation learning exist.

2 Lens Determination Assist System

2.1 Composition of Lens Determination Assist System

2 shows a lens determination assistance system 1 according to an embodiment. Referring to FIG. 2 , the lens determination assistance system 1 includes a lens size determination module 1000 for deriving a lens size from information about a lens to be inserted into the eye of a prospective surgery, and a bolting value prediction module to assist in determining the lens size (2000) and a lens power determination module 3000 for deriving the lens power.

The lens determination assistance system 1 may perform a function of determining a lens size, predicting a bolting value, and determining a lens dioptric power. Specifically, the lens determination assisting system 1 may determine a lens size, predict a bolting value, and determine a lens dioptric power by using a lens determination model learned through machine learning.

Of course, in FIG. 2, the lens determination assistance system 1 is illustrated as including all of the lens size determination module 1000, the bolting value prediction module 2000, and the lens power determination module 3000, but is not limited thereto. Accordingly, the lens determination assistance system may include at least one of a lens size determination module, a bolting value prediction module, and a lens dioptric power determination module.

Also, the lens size determination module 1000 , the bolting value prediction module 2000 , and the lens power determination module 3000 may be implemented in one device or may be implemented in different devices. For example, when the lens size determination module for deriving the lens size from the lens determination assistance system 1 is implemented in any one device, only size information of a lens to be inserted into the eye of the surgical patient may be obtained.

Alternatively, in the lens determination assistance system 1 , at least two modules among a lens size determination module, a bolting value prediction module, and a lens dioptric power determination module may be implemented by interworking with each other. For example, in order to obtain information on the size of the lens to be inserted into the eye of the surgical candidate, the lens size determination module and the bolting value prediction module may be combined and implemented in conjunction to derive a bolting value predicted together with the size of the lens to be inserted. have. Hereinafter, each module of the lens determination assisting system will be described one by one.

3 is a diagram showing the configuration of the learning apparatus 11 and the determination assistance apparatus 21 of the lens determination assistance system 1 . In one embodiment, the lens decision assisting system 1 may include a learning device 11 and a decision assisting device 21 .

The learning apparatus 11 may train a lens determination model. Specifically, the learning apparatus 11 may learn the lens determination model based on the learning data. The learning apparatus 11 may train the lens determination model using various learning methods. For example, the learning apparatus 11 may train the lens determination model by a method such as supervised learning, unsupervised learning, reinforcement learning, or imitation learning. The learning apparatus 11 may train the lens determination model by providing labeled data with respect to the training data. However, the labeled data is not necessarily used, and unlabeled data may be used.

The decision assisting device 21 may receive and use the lens decision model learned from the learning device 11 . In detail, the decision assisting device 21 may output auxiliary information for determining a lens to be inserted into the eye of the surgical patient by using the learned lens decision model. Specifically, when receiving input data, such as examination data of a prospective surgery, the decision assisting device 21 may output information about a lens suitable for the eye of the prospective surgery. The decision assisting device 21 may allow a user to determine a lens to be inserted into the eye of a person who is expected to undergo lens implantation surgery through the output information on the lens.

The information about the lens may be information about a lens size, a predicted value of a bolting value, a lens power, and the like.

The lens determination model learned using the training data in the learning apparatus 11 may be transmitted to the determination assistance apparatus 21 . Of course, in FIG. 3 , the learning device 11 and the decision assisting device 21 are illustrated as separate, but the present invention is not limited thereto, and in some cases, the learning device 11 and the decision assisting device 21 are implemented separately. It may be, or it may be implemented as one without being separated. For example, the decision assistance device may be the same device as the learning device or a separate device.

4 is a diagram showing the configuration of a learning apparatus and/or a decision assisting apparatus. Referring to FIG. 4 , the learning apparatus and/or the decision assistance apparatus may include a memory unit 31 , a control unit 33 , and a communication unit 35 .

The learning device and/or the decision assisting device may include a control unit 33 . The controller 33 may control the operation of the learning apparatus and/or the decision assistance apparatus. The control unit 33 may read a system program and various processing programs stored in the memory unit 31 .

The control unit 33 includes at least one of a central processing unit (CPU), a random access memory (RAM), a graphic processing unit (GPU), one or more microprocessors, and other electronic components capable of processing input data according to a predetermined logic. may include.

The learning apparatus and/or the decision assistance apparatus may include a memory unit 31 . The memory unit 31 may store data and a learning model required for learning. The memory unit 31 may store examination data of a prospective surgery.

The memory unit 31 may store learning data, labeling data, unlabeled data, input data, output data, and the like.

The memory unit 31 includes a nonvolatile semiconductor memory, a hard disk, a flash memory, a RAM, a read only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), or other tangible nonvolatile recording media, etc. can be implemented as

The memory unit 31 may store various processing programs, parameters for performing processing of the programs, or data as a result of such processing.

The learning device and/or the decision assisting device may further include a communication unit 35 . The communication unit 35 may communicate with an external device. The communication unit 35 may perform wired or wireless communication. The communication unit 35 may perform two-way or one-way communication.

The learning device and/or decision aid device may include a processor, volatile memory, non-volatile memory, mass storage device, and a communication interface. The processor may learn the lens decision model through the learning device and/or the decision assisting device.

5 is a diagram schematically illustrating a lens determination assistance system using a server. Referring to FIG. 5 , the lens determination assistance system may include a plurality of client devices and server devices. Hereinafter, the first client device among the plurality of client devices will be exemplarily described, but the same operation may be performed in the second client device.

The first client device 51 may request information from the server device 40 and may obtain information on lens determination assistance transmitted in response thereto, and may request lens determination assistance information from the server device 40 .

The first client device 51 may obtain data necessary for lens determination, and transmit the obtained data from the determination assistance device.

The first client device 51 may be a portable device such as a smart phone or a tablet PC.

The server device 40 may store and/or drive the lens crystal model. The server device 40 may store weights constituting the learned lens determination model. The server device 40 may collect and/or store data used to assist in lens determination.

The server device 40 may output a result of the lens determination assistance process using the lens determination model to the first client device 51 . The server device 40 may obtain feedback from the first client device 51 .

In an embodiment, the first client device 51 may obtain the lens determination model from the server device 40 and drive the obtained lens determination model. In this case, the first client device 51 may obtain information on the lens determination assistance by driving the lens determination model without providing input data to the server apparatus 40 .

The server device 40 may communicate with the first client device 51 to obtain the first lens determination assistance information and/or the second client device 52 to obtain the second lens determination assistance information.

6 is a diagram schematically illustrating the relationship between the server device 40 and the client device 50 . Referring to FIG. 6 , the server device 40 may communicate with the client device 50 through a communication unit. The communication unit may perform wired or wireless communication. The communication unit may perform two-way or one-way communication. The client device 50 may also communicate with the server device 40 through a communication unit.

In one embodiment, when the client device 50 transmits the input data of the surgery prospect to the server device, information on the lens to be inserted into the eye of the surgery prospect is transmitted using the lens determination model learned in the server device 40 . can receive

In one embodiment, the control unit 53 of the client device 50 obtains input data from the memory unit 51 , and the obtained input data is transmitted to the communication unit 45 of the server device 40 through the communication unit 55 . can be transmitted. In addition, the control unit 43 of the server device 40 obtains a result value by inputting input data into the lens determination model stored in the memory unit 41 , and the obtained result value is transmitted to the client device ( 50) and may be transmitted to the communication unit 55 of FIG.

2.2 Lens Determination Model

7 is a diagram showing the configuration of a lens crystal model. Referring to FIG. 7 , the lens determination model 100 may include a lens size determination model 110 , a bolting value prediction model 120 , and a lens power determination model 130 .

Of course, the lens determination model 100 in FIG. 7 is illustrated as including all of the lens size determination model 110, the bolting value prediction model 120, and the lens frequency determination model 130, but is not limited thereto, and in some cases Accordingly, the lens determination model 100 may include at least one of a lens size determination model, a bolting value prediction model, and a lens power determination model.

In an embodiment, the lens determination model 100 may include a lens size determination model and a lens power determination model, and may include a lens size determination model and a bolting value prediction model.

Also, the lens determination model 100 may be implemented in one device or may be implemented in different devices. For example, when the lens determination model includes a lens size determination model and a bolting value prediction model, each model may be implemented in one device by interworking with each other. Alternatively, when the lens determination model 100 includes a lens size determination model and a lens power determination model, the lens size determination model may be implemented in different devices independently of the lens power determination model.

8 is a diagram of a lens determination assistance process. Referring to FIG. 8 , the lens determination assistance process may be divided into a learning step ( S100 ) of learning the lens determination model and a determination step ( S200 ) of performing the lens determination model using the learned lens determination model.

Referring to FIG. 8 , the learning step ( S100 ) may be a process of learning a lens determination model using training data. Also, the learning step S100 may be performed by a learning apparatus.

According to an embodiment, in the learning step ( S100 ), learning data may be obtained, and a lens determination model may be learned from the obtained data. That is, the learning step ( S100 ) is a process of generating a lens determination model, and according to the generation of the lens determination model, model parameters constituting the lens determination model may be obtained. As an example, the model parameters may include weights adjusted while learning the lens determination model.

In one embodiment, the learning data is a plurality of examination data of an operator who has undergone lens implantation in the past, information on lenses to which the operator's intraocular insertion has been applied (lens size and lens power), surgical parameters, and balls measured after the operator's surgery It may include value data.

In addition, the examination data among the learning data may include examination data obtained by an operator who has undergone lens implantation in the past from a plurality of examination equipment related to eye measurement.

In an embodiment, the lens information among the learning data may include a lens size and/or lens power inserted into the eye of an operator who has undergone lens implantation in the past. In this case, the information on the lens may include the lens size and/or the lens power inserted into the operator's eye when the operator who has received the previous lens implantation has no side effects after the lens implantation operation. Of course, according to an embodiment, the information on the lens may include the lens size and/or the lens power inserted into the eye of the operator when a side effect occurs after the lens implantation surgery by the operator who has received the lens implantation in the past.

In addition, the surgical parameter among the learning data may relate to corneal incision information in a corneal incision process of an operator who has undergone lens implantation in the past. As an example, the surgical parameter may include a corneal incision method, a corneal incision location, a corneal incision degree, and the like.

In one embodiment, the bolting value data among the learning data refers to a value indicating the distance between the rear surface of the lens to be inserted into the eye of the prospective person for lens implantation and the front of the lens, and is measured for an operator who has undergone lens implantation in the past. It may mean bolting value data.

The lens determination model may be a model that outputs lens information based on the training data. As the lens determination model, at least one of a plurality of learning algorithms for calculating lens information may be selected. For example, the algorithm may be a logistic regression, a K-nearest neighbor algorithm, a support vector machine, a decision tree, or the like.

The lens determination model may use a plurality of learning algorithms among a plurality of learning algorithms for calculating predicted values. For example, an ensemble method may be used for the lens decision model, and better prediction performance may be obtained compared to a case where learning algorithms are used separately.

The lens determination model may be implemented in the form of a classifier that generates lens information. The classifier can perform double classification or multiple classification.

The lens determination model derives lens information and may be implemented in the form of regression. The regression method may be a linear regression method, a logistic regression method, or the like.

In one embodiment, the learning step S100 obtains a result value (output data) using a model to which arbitrary weights are given, compares the obtained result value (output data) with the labeling data of the training data, and the This can be performed by performing backpropagation according to the error and optimizing the weights.

Although not shown, the learning step S100 may include an evaluation step of evaluating the performance of the learned lens determination model. In the evaluation step, the lens decision model may be evaluated using the evaluation data set. The evaluation of the lens determination model may be a step of evaluating the lens determination model learned by the learning step and making predictions on new data using the lens determination model. Specifically, the evaluation step may be a step of measuring whether the learned lens determination model can be generalized to new data.

In addition, in the learning step (S100) of the lens determination model, the training data set and the evaluation data set may be distinguished. Here, the learning data set may mean a set of learning data used in the learning process of the learning step, and the evaluation data set may mean a set of evaluation data used in the evaluation process of the evaluation step. In this case, the training data set used to train the lens determination model may not be used in the evaluation step of the lens determination model.

Also, referring to FIG. 8 , in the determining step S200 , the lens determination model may be used by acquiring model parameters in the learning step. Specifically, in the determining step ( S200 ), after acquiring input data such as examination data of a prospective surgery, information (results) of a lens to be inserted into the eye of a prospective surgery may be acquired using the learned lens decision model. Also, the determining step S200 may be performed by a decision assisting device.

The input data may include a plurality of examination data of a prospective person for lens implantation surgery.

The result value may include information on a lens to be inserted into the eye of a prospective surgery. The lens information may include a lens size, a lens power, a predicted voltage value, and the like. Hereinafter, lens size determination, bolting value prediction, and lens dioptric power determination will be described in more detail.

3 Determine the lens size

3.1 Configuration of lens size determination module

9 is a diagram illustrating the configuration of the lens size determination module 1000 . In an embodiment, the lens size determination module 1000 may output the size of the lens to be inserted into the eye of the surgical patient from the input data.

Referring to FIG. 9 , the lens size determination module 1000 may include an input unit 1100 , a lens size determination unit 1300 , and an output unit 1500 .

The input unit 1100 may obtain input data from a database. Here, the input data may be a plurality of examination data of a scheduled surgery.

Specifically, the input unit 1100 may be directly connected to a database to obtain input data. In addition, the input unit 1100 may receive and acquire input data from a server or other external device.

The input data may be examination data of a prospective surgery. The input data may include a plurality of parameters. Specifically, the input data may include inspection data indicating different parameters obtained from the same or different inspection equipment, or may include inspection data indicating the same parameter obtained from different measurement equipment.

In addition, the input data may include test data measured at the same time point or may include test data measured at different time points.

The input data may be the same and/or different parameters obtained from the same inspection equipment, and may be the same and/or different parameters obtained from different inspection equipment.

Also, there may be one input data or a plurality of input data. Each input data may have a different degree of influence on the result value (the size of the lens to be inserted into the eye of the surgical patient). In an embodiment, each input data may include a parameter, and the degree to which the input data affects the result value may vary according to the type of the parameter included in the input data. Here, the parameter is defined to represent the characteristics of the eyeball of the surgery prospect, and may be expressed numerically. For example, the parameter is ATA (angle to angle distance) indicating the distance between the anterior angles, ACD (Anterior Chamber Depth)-epi, ACD-endo, CCT (Central Corneal Thickness), CLR (Crystalline Lens Rise), WTW (White) to white), AL(Axial Length), corneal curvature, refractive error (myopia, astigmatism, hyperopia), pupil size, intraocular pressure, visual acuity, corneal shape, corneal thickness, eye length, lens insertion space, etc. can

The person scheduled for surgery may include a person who is scheduled to undergo surgery by selecting lens implantation during vision correction surgery. The person scheduled for surgery may be a person who is going to perform lens implantation using information about the lens output from the lens determination assisting system. In addition, the user may be a person who obtains information about a lens to be inserted into the eye of the surgical patient by using the lens determination assistance system. For example, it may include a doctor who performs lens implantation, a lens manufacturer, and the like.

The plurality of examination data may be data obtained from a plurality of examination equipment for measuring the eyeball, and may include a plurality of eye-related parameters.

In addition, the examination data may include questionnaire data. Specifically, it may include the desired target visual acuity after lens implantation of a prospective surgery.

Also, the test data may be measurement data for the cornea. For example, it may include corneal shape, corneal symmetry, corneal thickness measurement data, corneal structure tomography data, corneal shape analysis data, corneal curvature, corneal endothelial cell examination data, and the like.

In addition, the test data may be measurement data for visual acuity and/or refraction. For example, it may include data on the power of glasses worn in the past, an eye exam, and refractive errors (degree of nearsightedness, astigmatism, and farsightedness).

In addition, the examination data may be measurement data for a distance within the eyeball. Specifically, it may include the pupil size, the length of the eyeball, the distance of the space in which the lens is to be inserted, and the like.

In addition, the test data may be data for an ocular disease and/or a possessed disease. For example, data on the presence or absence of retinal diseases, glaucoma, retinal degeneration, etc., cataracts, diseases of the back of the iris, and the like may be included.

In addition, the examination data may be measurement data for the retina. For example, it may include taking pictures of the fundus retina.

And, the inspection data may be measured from one or a plurality of equipment.

Of course, the plurality of examination data may not be limited to only data obtained from a plurality of equipment for measuring the eyeball. In addition, various data may be included. For example, it may include test data such as eye-related genes and blood.

The lens size determining unit 1300 may determine a lens size suitable for the eye of the surgical patient when input data such as a plurality of examination data of the surgical patient is input. Here, the suitable lens size may mean a lens size that minimizes the possibility of side effects when a lens implantation procedure is performed for a prospective surgery. A detailed operation of the lens size determiner 1300 will be described in more detail with reference to FIG. 11 .

In one embodiment, the lens size determiner 1300 may calculate the reliability of the accuracy of the lens size derived according to the examination data of the surgical patient. The information on the reliability may be stored in advance or may be provided from the outside. For example, information including that the reliability of the result value of the first equipment among the plurality of inspection equipment is 90% and the reliability of the result value of the second equipment is 80% may be transmitted through a pre-stored external device . In an embodiment, the lens size determination model 110 may calculate the reliability of the accuracy of the lens size derived according to the examination data of the surgical patient based on the reliability stored in advance in the learning step. For example, when the inspection data measured by the first equipment is used, it may be calculated that the reliability of the output lens size accuracy is 90%, and may be presented to the user through the output unit 1500 .

The output unit 1500 may output the lens size obtained through the lens size determiner 1300 to the user. In an embodiment, the output unit 1500 may provide a display that visually outputs the output data to the screen. In addition, the output unit 1500 may output in various formats, such as images and text.

The output unit 1500 may output information (output data) on the size of the lens to be inserted into the eye of the surgical patient through the lens size determiner 1300 .

The output unit 1500 may output a standardized lens size according to the learning method of the lens size determination model.

According to an embodiment, when the lens size determination model is implemented in the form of a classifier, the output unit may output a standardized lens size. The standardized lens size may be a ready-made lens size. The ready-made lens size may be a size made according to a predetermined standard in advance. For example, the standardized lens size may be 12.1mm, 12.6mm, 13.2mm, and 13.7mm. It will be described in more detail in Table of Contents 3.3 below. However, the present invention is not limited thereto, and when the lens size determination model is implemented in the form of regression, the output unit may output a non-standardized lens size. The non-standardized lens size means a numerical value of the lens size itself rather than selecting any one of predetermined categories such as a standardized lens size, and the output unit may output a numerical value of the lens size. Details on this will be described in Table of Contents 3.3.

10 is a view for explaining the side effects of lens implantation. Referring to FIG. 10, FIGS. 10 (a) and (c) show that an inappropriate lens size is inserted after lens insertion, and FIG. 10 (b) shows that a suitable lens size is inserted during lens insertion. will be. In addition, I shows the iris, La, Lb, and Lc show an intraocular lens, and C shows the lens.

The side effects of lens implantation may occur because the surgery is performed with an unsuitable lens size without considering the characteristics of the eye of the prospective surgery. For example, referring to (a) of FIG. 10 , a lens having a lens size of La may be inserted into the eye of a person scheduled for surgery. This is because a small lens size is selected and inserted without considering the characteristics of the eye of the prospective surgery, and friction between the lens and the lens may cause damage to the lens and cause cataracts. Also, referring to (c) of FIG. 10 , a lens having a lens size of Lc may be inserted into the eye of a prospective surgery. This is because a large lens size is selected and inserted without considering the characteristics of the eye of the prospective surgery, and the distal end of Lc is sandwiched between the lens and the iris, blocking the flow of aqueous humor, which may cause glaucoma.

Therefore, for lens implantation, the lens size that minimizes the possibility of side effects should be determined in consideration of the ocular characteristics of the prospective surgery. In one embodiment, referring to (b) of FIG. 10 , Lb may be a lens size suitable for the eye of the surgery expected in consideration of the characteristics of the eye of the surgery expected. Lb is a size in which friction does not occur between C and I, and is inserted at a position that maintains an appropriate distance between I and I, and may be a lens size that minimizes the possibility of side effects.

In this way, the possibility of side effects can be minimized when the lens size inserted into the eyeball, the lens, the lens, the iris, and the size of the lens insertion space are inserted in consideration of the characteristics of the eye of the person to be operated on.

3.2 Lens Sizing Process

11 is a diagram showing a flowchart of the lens size determination process S1000. Referring to FIG. 11 , the lens size determination process ( S1000 ) includes obtaining input data such as a plurality of examination data of a prospective surgery ( S1100 ), and deriving a lens size using a lens size determination model ( S1300 ). can do. The lens size determination process S1000 may be performed by the lens size determination module 1000 described above in FIG. 2 .

Specifically, in the step of obtaining the input data ( S1100 ), the input data may be a plurality of examination data obtained from a plurality of examination equipment related to the eye measurement of a prospective surgery. In an embodiment, the plurality of eye measurement-related inspection equipment may be inspection equipment that measures using a laser and/or high-frequency ultrasound. For example, it may include ultrasound biomicroscopy (UBM), optical coherence tomography (OCT), and the like.

In an embodiment, the plurality of inspection data may include the parameters described above with reference to FIG. 9 . Specifically, the plurality of test data includes: corneal curvature, refractive error (nearsightedness, astigmatism, farsightedness), pupil size, intraocular pressure, visual acuity, corneal shape, corneal thickness, eye length, lens insertion space, ATA (Anterior Chamber Angle) , ACD (Anterior Chamber Depth)-epi, ACD-endo, CCT (Central Corneal Thickness), CLR (Crystalline Lens Rise), WTW (White to white), AL (Axial Length), BUT, distance measurement between the iris, etc. It may include parameters.

The lens size may be derived from the obtained input data using a lens size determination model.

Also, in an embodiment, since the degree of influence on the result value is different depending on the type of parameter included in the input data, the result value may be different for each input data.

Also, in another embodiment, each input data may include at least some of the same parameters, but may be derived by different inspection equipment. Even if the same parameter is derived from different inspection equipment, the indicated values may be different. This may be due to different methods and principles of measuring parameters for each inspection equipment. For example, the A parameter obtained from the UBM examination equipment (here, the A parameter indicates an arbitrary parameter) is a parameter that performs the same function as the A parameter obtained from the OCT examination equipment, but the numerical value indicating this may be different . As such, since the accuracy of the parameter measured is different depending on the test equipment, the degree of influence of the input data on the result value may be different.

In addition, each parameter or the same parameter may have a different degree of influence on a result value depending on whether the parameter is output from any test equipment, and accordingly, the priority between input data may be changed.

In order to increase the accuracy of the result value, the weight may be increased for a high priority parameter or a parameter derived from a high priority inspection equipment.

In one embodiment, the input data may include priority data. Priority data may be prioritized according to the type of parameter included in the input data. For example, when a first parameter having a large degree of influence on derivation of a lens size is included, the input data may be first priority data. When the degree of influence on derivation of the lens size includes the second parameter smaller than the first parameter, the input data may be second priority data. For convenience of explanation, only the first priority data and the second priority data have been described, but the present invention is not limited thereto, and a plurality of priority data may be included.

In an embodiment, when the lens size is determined using the input data including the first priority data, the first lens size may be derived, and the lens size is determined using the input data including the second priority data. If determined, the second lens size may be derived. For example, the probability for the accuracy of the result value may be higher in the first lens size than in the second lens size. At this time, when the scheduled operation is performed with the first lens size, side effects may occur less than when the operation is performed with the second lens size.

In one embodiment, in the lens size deriving step ( S1300 ), when the first priority data is not included in the input data and the second priority data is included, the lens size determining unit uses the second priority data to obtain a lens size can be derived.

In an embodiment, the input data may essentially include first priority data. This is to ensure the accuracy of the results.

In the learning step ( S100 ), in the lens size determination model, the first priority data and the second priority data may be learned together, or they may be learned separately. According to an embodiment, when the first priority data and the second priority data are learned together, any one of the first priority data and the second priority data may be used in the lens size derivation step ( S1300 ). According to another embodiment, when only the first priority data is learned, in the lens size deriving step S1300, only the first priority data may be used, and when only the second priority data is learned, in the lens size deriving step S1300 , only the second priority data may be used. Of course, the present invention is not limited thereto, and according to an embodiment, even if only the first priority data is learned, the second priority data may be used in the lens size derivation step (S1300), and even if only the second priority data is learned, the lens size derivation step In (S1300), the first priority data may be used.

3.3 Examples

12 and 13 are diagrams illustrating an embodiment of determining a lens size. That is, the step S1300 described above with reference to FIG. 11 will be described in more detail with reference to FIGS. 12 and 13 . 12 and 13 , in an embodiment, the lens size determination model may be implemented including a classifier.

In one embodiment, the classifier may use an algorithm of any kind, such as a decision tree, a support vector machine, a random forest, or the like. This is only an example, and is not limited thereto.

In an embodiment, the lens size determination module 1000 may input input data to the lens size determination model 110 and derive a lens size from the lens size determination model 110 . The lens size determination model 110 may include a classifier, and the classifier may determine any one of lens sizes having a predetermined value.

In addition, a standardized lens size may be obtained from the input data of the surgeon using a lens size determination model implemented including a classifier. The classifier may determine the size of one lens to be inserted into the eye of the surgeon from the input data of the surgeon.

In addition, the standardized lens size may be a ready-made lens size. The ready-made lens size may be a size made according to a predetermined standard in advance. In one embodiment, a lens size standardized using a lens size determination model implemented as a classifier from input data of a prospective surgery, for example, a lens size of 12.6 mm, which is one of 12.1 mm, 12.6 mm, 13.2 mm, and 13.7 mm can be determined.

14 and 15 are diagrams illustrating another embodiment of determining a lens size. That is, the step S1300 described above with reference to FIG. 11 will be described in more detail in FIGS. 14 and 15 as well. 14 and 15 , in an embodiment, the lens size determination model may be implemented including a regression model.

In an embodiment, the regression model may use types of algorithms such as linear regression, regression tree, support vector regression, and kernel regression. This is only an example, and is not limited thereto.

In an embodiment, the lens size determination module 1000 may input input data to the lens size determination model 110 and derive a lens size from the lens size determination model 110 . The lens size determination model 110 may include a regression model, and the regression model may determine any one of a lens size that may or may not be a predetermined value.

In addition, the input data of the surgeon may obtain a normalized and/or non-standardized lens size using a lens size determination model implemented including a regression model. The regression model may derive a probability for the size of the lens to be inserted into the eye of the surgeon from the input data of the surgical patient. The lens size derived with the highest probability may be the most suitable lens size among the lens sizes to be inserted into the eye of the surgeon.

According to an embodiment, the output unit 1500 may output a non-standardized lens size according to the learning method of the lens size determination model. The non-standardized lens size may be expressed as the entire lens size including the standard lens size. Specifically, the non-standardized lens size is the size of the lens to be inserted into the eyeball in consideration of the characteristics of the eye of the person to be operated on, and may be larger or smaller than the existing lens size, or may be a size between the existing lens sizes. The non-standardized lens size may be a lens size to be custom-made in consideration of the characteristics of the eye of the surgical person. The non-standardized lens size may be a lens size optimized for the eye of the surgeon rather than the standardized lens size. This may be a customized lens size for the eye of the surgical person.

In another embodiment, the lens size determination model 110 may be implemented including a classifier and a regression model. That is, the lens size determination model 110 may be implemented by serially or parallelly combining a classifier and a regression model. As an example, the lens size determination module 1000 inputs input data to the lens size determination model 110 in which the classifier and the regression model are combined, and derives the lens size and numerical value from the combined lens size determination model 110 . can For example, from a lens size of 12.5 mm output through the regression model, a standardized lens size of 12.6 mm may be derived through a classifier. Alternatively, a lens size of 13.2 mm output through the classifier may derive a non-standardized lens size of 13.3 mm through a regression model. This is only an example, and the lens size output through the classifier and the lens size output through the regression model may be simultaneously derived.

16 is a diagram illustrating another embodiment of determining a lens size. Referring to FIG. 16 , the lens size determination module 1000 may further include a data complementing unit 1200 for supplementing input data.

The data complementing unit 1200 determines the size of a lens to be inserted into the eye of a prospective surgery during lens implantation in the case where priority data cannot be obtained from the examination equipment or in an environment where examination equipment capable of obtaining priority data is not provided. A more accurate lens size can be derived.

According to an embodiment, the input data may include the first priority data and the second priority data, may include only the first priority data, and may include only the second priority data. When the user can use only the second priority data as input data and the first priority data cannot be used, the accuracy of the result value may be lower than when only the first priority data is used. As in this case, even when the first priority data is missing, the first priority data may be estimated using the data complementing unit 1200 in order to improve the accuracy of the result value.

The data complementing unit 1200 may estimate the first priority data from the input data.

The priority data may include parameters having a large degree of influence on the result value. In order to increase the accuracy of the result value, high-priority data, that is, data including a parameter with a large degree of influence on the result value may be input. For example, when the lens size determiner determines the lens size by using the first input data including ATA among parameters, the accuracy of the result value may be high, and the second input data not including ATA and only including CCT may be used. If the lens size is determined using the In this case, the first input data includes first priority data having a high priority, and the accuracy of the result value may be high.

In order to increase accuracy even when the first priority data cannot be obtained and only the second priority data can be obtained depending on circumstances, the data complementing unit may estimate the first priority data by using the second input data. The estimated first priority data may be used as input data in the lens size determiner 1300 . For example, an ATA corresponding to a missing value may be estimated from parameters such as CCT and vision test data included in the second input data.

In an embodiment, when the inspection equipment capable of measuring the first priority data is not provided, the first priority data may be estimated using the data complementing unit.

In an embodiment, the data complementing unit 1200 may estimate input data other than the first priority data as the first priority data using a separate formula.

In an embodiment, the data complementation unit 1200 may estimate input data other than the first priority data as the first priority data using a separate formula using the data complementation model. Although not shown, the data complementation model may be learned based on the first priority data and the second priority data as training data. As an example, the data complementation model may be trained to receive the second priority data and derive the first priority data. The learned data supplementation model may derive the estimated first priority data when the second priority data is input as input data. This is only an example, and is not limited thereto.

Also, in an embodiment, the lens size determination model may use a plurality of machine learning algorithms among a plurality of machine learning algorithms for calculating a predicted value. For example, the lens size determination model may be learned using an ensemble method and the lens size may be estimated. As the ensemble technique is used for the lens size determination model, the accuracy of the lens size determination model may be improved.

17 is a diagram schematically illustrating a plurality of sub-models of a lens size determination model. Referring to FIG. 17 , the lens size determination model may include a plurality of sub-models. Each of the plurality of sub-models may independently determine a lens size. For example, the first submodel may be a model that is learned by a random forest technique to determine a lens size, and the second submodel may be a model that is learned by a decision tree technique to determine a lens size. In addition, although only the first sub-model and the second sub-model are illustrated in FIG. 17 , this is only an example, and the sub-model may include a plurality of sub-models.

A plurality of sub-models may be connected to each other in parallel. Here, input data input to the plurality of sub-models and output values output by the plurality of sub-models may be the same or different.

The lens size determination model may output a prediction result based on an output value of the sub-model.

The lens size determination model may include an output sub-model that outputs a prediction result based on output values of a plurality of sub-models connected in parallel. In an embodiment, when the first output value and the second output value that are output values of the first sub-model and the second sub-model are the same, the output sub-model may output the same value. In another embodiment, when the first output value and the second output value, which are output values of the first sub-model and the second sub-model, are different, the output sub-model considers the output values of the plurality of sub-models at a certain ratio and outputs a prediction result Alternatively, a specific value among the plurality of output values may be output. In other words, the output submodel may give a weight to the first output value from the first submodel and the second output value from the second submodel, and may output the lens size by reflecting the weight assigned to each output value. For example, a weight of 0.8 is given to the first output value and a weight of 0.2 is given to the second output value, the first output value represents 12.6 mm among the standardized lens sizes, and the second output value is 13.2 mm among the standardized lens sizes , the output sub-model may derive a lens size of 12.6 mm, which is a first output value having a high weight, as an output value.

In addition, for example, a weight of 0.8 is given to the first output value, a weight of 0.2 is given to the second output value, the first output value represents 12.6 mm among non-standardized lens sizes, and the second output value is non-standardized. In the case of representing 13.2 mm among the lens sizes, the output sub-model may derive the lens size of 12.7 mm reflecting the weight as an output value.

Also, the weight may be determined during a learning process. That is, the above-described learning step S100 in FIG. 8 may be performed for the lens size determination model including a plurality of sub-models, and weights may be determined in the learning step S100.

In another embodiment, the output sub-model may output another value generated based on the plurality of output values as a prediction result. Here, the output value of the output sub-model may be of the same type as the plurality of output values or may be of a different type.

In an embodiment, the first submodel and the second submodel may be the same, and input data input to each submodel may be different. The input data may be first priority data or second priority data. A sub-model to which high-priority data is input may be given a high weight. For example, the first sub-model to which the first priority data is input may be given a higher weight than the second sub-model to which the second priority data is input, so that the first output value may be derived as the lens size output value.

4 Predicting the voltage value

4.1 Definition of Bolt Value

18 is a diagram for defining a bolting value. The vaulting value (or also referred to as a vault value) is a value indicating the distance between the rear surface of the lens to be inserted into the eye of the prospective person for lens implantation and the front of the lens, specifically, the value of the lens to be inserted into the eye. It is defined as the shortest distance among a plurality of distances between the posterior surface and the anterior surface of the lens. Referring to FIG. 18 , L denotes a lens inserted into the eye of a prospective surgery, I denotes the iris, C denotes the lens, and V denotes a bolting value. L can be inserted in the space between I and C. A plurality of distances may exist between the lens inserted into the eye and the front of the lens. Among them, the shortest distance between the rear surface of the lens and the front surface of the lens, that is, the distance between the lens and the lens in the vertical direction from the center of the cornea may be V.

In general, in order to check whether a lens of a suitable size is inserted into an operator's eye after lens insertion, a bolting value may be measured. If the bolting value measured after surgery is included within a certain range, it may be determined that the size of the lens inserted into the eye is a lens size suitable for the operator's eye. For example, in the predetermined range, the bolting value may be included in the range of 250 to 750 μm. In one embodiment, when the bolting value is 250 μm or less, the size of the lens inserted into the operator's eye can be considered to be smaller than the size suitable for the operator's eye. When a size smaller than the size of the lens suitable for the operator's eye is inserted, as described in FIG. 10A , cataracts may be induced. In another embodiment, when the bolting value is 750 μm or more, the size of the lens inserted into the operator's eye can be considered to be larger than the size suitable for the operator's eye. When a size larger than the lens size suitable for the operator's eye is inserted, glaucoma may be induced as described in FIG. 10( c ). Therefore, in order to prevent the side effects of lens implantation, the bolting value after surgery may have to be included within an appropriate range. In other words, it is necessary to accurately design the lens before inserting the lens and then insert it. Hereinafter, a bolting value prediction module for predicting a bolting value and a process therefor will be described.

4.2 Volt Value Prediction Module Configuration

19 is a diagram showing the configuration of the bolting value prediction module 2000. As shown in FIG. In an embodiment, the bolting value prediction module 2000 may output the predicted intraocular bolting value of the surgical patient from the input data.

Referring to FIG. 19 , the bolting value prediction module 2000 may include an input unit 2100 , a bolting value prediction unit 2300 , and an output unit 2500 .

The input unit 2100 may obtain input data from a database. The input data may be a plurality of examination data of a prospective surgery.

Specifically, the input unit 2100 may be directly connected to a database to obtain input data. In addition, the input unit 2100 may receive and acquire input data from a server or other external device.

The input data may be examination data of a prospective surgery. The test data may be the same as described in Table of Contents 3.1 above. Hereinafter, only the different contents will be described.

According to an embodiment, the input data may include any expected lens size to be inserted into the eye of the surgical patient. Any expected lens size may be a normalized or non-standardized lens size.

In an embodiment, the bolting value predicting unit 2300 may predict the bolting value of the surgical patient from the input data.

In addition, when the predicted bolting value is within a certain range, the bolting value prediction unit 2300 may provide the predicted bolting value, and the user can determine that surgery has been performed with an appropriate lens size based on the predicted bolting value. can

In an embodiment, the bolting value prediction unit 2300 may provide information on whether the inputted lens size enables lens implantation by a prospective surgery according to the predicted bolting value. For example, when 200 μm, which is not included in the range of 250 to 750 μm, is derived from the predicted bolting value, it may be determined that the lens implantation surgery is impossible. This is merely an example and is not limited thereto, and conversely, when the predicted bolting value is within a certain range, it may provide a determination that lens insertion is possible.

In addition, the bolting value prediction unit 2300 may provide information on whether the input lens size is suitable for a lens implantation operation of a prospective surgery according to the predicted bolting value. For example, when the predicted bolting value is 500 μm, it is possible to provide information that a lens size of 13.2 mm input as input data is suitable for a lens implantation operation of a prospective surgery. In addition, when the predicted bolting value is 800 μm, it is possible to provide information that the lens size of 13.2 mm input as input data is not suitable for lens implantation of a prospective surgery. This is only an example, and is not limited thereto.

A detailed operation of the bolting value prediction unit 2300 will be described in more detail with reference to FIG. 20 .

The output unit 2500 may output the bolting value obtained through the bolting value prediction unit 2300 to the user. In an embodiment, the output unit 2500 may provide a display that visually outputs the output data to the screen. In addition, the output unit 2500 may output in various formats, such as images and text.

In an embodiment, the predicted bolting value may be a criterion for determining the result of lens implantation. For example, if the predicted bolting value is included in the range of 250 to 750 μm, the user may determine that the operation has been performed with a lens size suitable for the eye of the surgical patient.

In an embodiment, according to the result of the bolting value prediction unit 2300 , information on the suitability and/or failure of the lens implantation surgery of the prospective surgery with respect to the lens size input as input data may be output.

4.3 Vault Value Prediction Process

20 is a diagram illustrating a flowchart of a bolting value prediction process ( S2000 ). Referring to FIG. 20 , the bolting value prediction process (S2000) includes a step of obtaining input data such as examination data of a scheduled surgery (S2100), and a step of deriving a predicted bolting value using a bolting value prediction model (S2300). can The bolting value prediction process ( S2000 ) may be performed by the bolting value prediction module 2000 described above in FIG. 2 .

Specifically, in the step of obtaining the input data ( S2100 ), the input data may be a plurality of examination data obtained from a plurality of examination equipment related to eye measurement of a prospective surgery. This is omitted because it is the same as the lens size determination process in Table of Contents 3.2.

In an embodiment, the input data may include the plurality of inspection data and an arbitrary lens size.

In the step of deriving the predicted bolting value ( S2300 ), a bolting value corresponding to an arbitrary lens size may be predicted by using a bolting value prediction model based on a plurality of examination data of the scheduled surgery.

In an embodiment, as described with reference to FIG. 3 , the bolting value prediction model 120 may be trained by the learning apparatus 11 and may derive a bolting value predicted by the decision assistance apparatus 21 . In addition, the bolting value prediction model 120 may be learned through the learning step ( S100 ), and a predicted bolting value may be derived through the determining step ( S200 ). The matters described with reference to FIGS. 3 and 8 may be directly applied to the bolting value prediction model 110 .

4.4 Examples

In an embodiment, the bolting value prediction module may predict the bolting value by using a plurality of examination data and/or an expected lens size of the surgical patient as input data.

21 is a diagram illustrating an embodiment of a bolting value prediction. Referring to FIG. 21 , the bolting value predicting unit 2300 may predict a bolting value and a lens size by using a plurality of examination data of a prospective surgery as input data.

The bolting value may be derived differently depending on the size of the lens to be inserted into the eyeball.

In an embodiment, the bolting value prediction module 2000 may output the lens size and the predicted bolting value together when the inspection data is input to the bolting value prediction model 120 . By outputting the lens size and the predicted bolting value as the output data together, the user can accurately determine the lens size suitable for the characteristics of the eye of the surgical person based on the predicted bolting value. For example, when a lens size of 12.6 mm and 500 μm are output, the predicted bolting value is included within an appropriate range, so the user can determine that the output lens size is a lens size suitable for the characteristics of the eye of the surgical person. . Alternatively, when the lens size of 13.2 mm and 900 μm are output, the predicted bolting value is not included in the appropriate range, so the user can determine that the output lens size is a lens size that does not fit the characteristics of the eye of the person scheduled for surgery. . This is only an example, and is not limited thereto.

In an embodiment, the bolting value prediction model 120 may be trained using training data that inputs inspection data and outputs a bolting value and a lens size. In the learning step of the bolting value prediction model, the learning step S100 described with reference to FIG. 8 may be applied as it is.

In an embodiment, the predicted bolting value may be a bolting value that satisfies a predetermined range. That is, the output unit 2500 may output a bolting value that satisfies a predetermined range and a lens size corresponding thereto. For example, the output unit 2500 may output a lens size of 12.6 mm when the bolting value is 500 μm, which satisfies the range of 250 to 750 μm.

22 is a diagram illustrating another embodiment of a bolting value prediction. Referring to FIG. 22 , the bolting value predicting unit 2300 may predict a bolting value by using a plurality of examination data and an arbitrary lens size of a prospective surgery as input data.

In an embodiment, the bolting value prediction module 2000 may output a predicted bolting value when inspection data and an arbitrary lens size are input to the bolting value prediction model 120 . The user may determine whether the inputted arbitrary lens size is a lens size suitable for the characteristics of the eye of the surgeon by looking at the predicted bolting value for the inputted arbitrary lens size. For example, after inputting examination data and a lens size of 13.2 mm as input data, when a predicted bolting value of 450 μm is output, the lens size of 13.2 mm, which is the arbitrary lens size, is the characteristic of the eye of the surgeon. It can be judged that the lens size is suitable for the This is only an example, and is not limited thereto.

In an embodiment, the bolting value prediction model 120 may be trained using training data that inputs inspection data and a lens size, and outputs a bolting value. In the learning step of the bolting value prediction model, the learning step S100 described with reference to FIG. 8 may be applied as it is.

Although not shown, in an embodiment, the bolting value predictor 2300 may be implemented in conjunction with the lens size determiner 1300 . This allows the user to verify the accuracy of the result value derived through the lens size determiner 1300 . For example, when 13.2 mm, which is the lens size derived through the lens size determination unit 1300, is input as the input data of the bolting value prediction unit 2300 together with the inspection data, if the predicted bolting value is 500 μm, this As the bolting value is included within a certain range, it can be verified that the lens size of 13.2 mm, which is the result value of the lens size determining unit, is a suitable result value in the eye of the surgical person, and the accuracy of the result value is also high.

In an embodiment, the lens size determiner 1300 and the volt value predictor 2300 may be serially connected to each other. Specifically, the lens size (output data) derived using the lens size determiner 1300 may be obtained as input data of the bolting value predictor 2300 . That is, the bolting value predicting unit 2300 may be the lens size, which is the result value of the lens size determining unit, as input data, and a plurality of examination data of the scheduled surgery. Through this, the bolting value prediction unit 2300 may output a predicted bolting value corresponding to the input lens size.

5 Determination of lens power

5.1 Composition of the lens power determination module

Even if the maximum corrected visual acuity of the eye after lens implantation exceeds the target visual acuity, the quality of visual acuity may deteriorate due to residual astigmatism after surgery. For example, after lens implantation, even if the corrected visual acuity of the eye reaches the target visual acuity of 1.2 and astigmatism is partially corrected, residual astigmatism may remain. In this case, the operator cannot obtain the expected surgical result due to the residual astigmatism. Therefore, factors of astigmatism, such as corneal astigmatism, may need to be considered when determining the power of a lens used for lens implantation.

In one embodiment, when determining the dioptric power of the lens, it is necessary to determine the dioptric power of the lens in consideration of not only the maximum corrected visual acuity but also corneal astigmatism induced by surgery. In this case, the residual astigmatism may be predicted in advance, and a factor for correcting the residual astigmatism may be reflected in advance in the lens. Accordingly, the user can obtain not only the target visual acuity but also the desired visual quality.

23 is a diagram illustrating the configuration of the lens power determination module 3000 . In an embodiment, the lens power determination module 3000 may output the lens power to be inserted into the eye of the surgical patient from the input data.

Referring to FIG. 23 , the lens power determination module 3000 may include an input unit 3100 , a lens size determination unit 3300 , and an output unit 3500 .

The input unit 3100 may obtain input data from a database. The input data may be a plurality of examination data of a prospective surgery.

Specifically, the input unit 3100 may be directly connected to a database to obtain input data. In addition, the input unit 3100 may receive and acquire input data from a server or other external device.

The input data may be examination data of a prospective surgery. The test data may be the same as described in Table of Contents 3.1 above. Hereinafter, only the different contents will be described.

According to an embodiment, the examination data may include the measured unaided visual acuity of the prospective surgery, diopters measured from the eyeball, astigmatism axis, astigmatism direction parameters, corneal astigmatism and lens astigmatism, ratio data of myopia and astigmatism, etc. .

According to an embodiment, the input data may include corneal incision information. The corneal incision information may refer to information about a predicted or planned corneal incision in the course of incision of the cornea of a prospective surgery prior to inserting a lens during lens implantation. The corneal incision information may include a corneal incision method, a corneal incision position, a corneal incision direction, and/or a corneal incision degree, during a corneal incision process for a lens implantation procedure of a prospective surgery. The amount of change in astigmatism may be different according to the location of the corneal incision among the corneal incision information, and the surgically induced astigmatism value (SIA) may be different according to the size of the corneal incision. Therefore, when determining the lens power, considering the corneal incision information, predicting astigmatism adjusted for factors such as changes in astigmatism and/or surgically induced astigmatism, and then determining the lens power, the effect of improving visual acuity can bring

In an embodiment, the lens power determination unit 3300 may apply input data, such as a plurality of examination data of the surgery prospect, to the lens power determination model 130 to determine the lens power suitable for the eyes of the surgery prospect. Here, the appropriate lens power may mean a lens power of which the possibility of side effects is minimized and the quality of vision is high when a lens implantation procedure is performed for a prospective surgery. Side effects in determining the lens power may include decreased visual acuity and headaches due to decreased visual acuity.

In one embodiment, in order to determine a suitable lens power, the lens power determining unit 3300 inputs input data such as a plurality of examination data of a prospective surgery and/or expected corneal incision information of a prospective surgery, and provides an eye of the prospective surgery. Suitable lens powers can be output.

A detailed operation of the lens power determining unit 3300 will be described in more detail with reference to FIG. 24 .

The output unit 3500 may output information (output data) on the power of a lens to be inserted into the eye of the surgical patient through the lens power determination unit 3300 .

The output unit may output a lens power suitable for the eye of the surgical patient according to the learning method of the lens power determination model.

According to an embodiment, when the lens power determination model is implemented in the form of a regression, the output unit may output a lens power suitable for the target visual acuity of the eye of the surgeon. Among the plurality of lens powers, it is possible to output a lens power having the highest probability to be suitable for the eye of a person scheduled for surgery. This is exemplary and not limited thereto, and the lens power determination model may be implemented using a classifier. In this case, it is possible to output a lens power suitable for the eye of a person scheduled for surgery from among the plurality of standardized lens powers.

5.2 Lens power determination process

Fig. 24 is a diagram showing a flow chart of the lens power determination process (S3000). Referring to FIG. 24 , the lens power determination process ( S3000 ) includes a step of obtaining input data such as a plurality of examination data of a prospective surgery ( S3100 ), and a step of deriving the lens power using a lens power determination model ( S3300 ). can do. The lens power determination process S3000 may be performed by the lens power determination module 3000 described above in FIG. 2 .

Specifically, in the step of obtaining the input data ( S3100 ), the input data may be a plurality of examination data obtained from a plurality of examinations related to eye measurements of a prospective surgery. In an embodiment, the plurality of examinations related to eye measurement may include slit lamp microscopy, fundus examination, automatic refraction and corneal curvature examination, corneal topography examination, and the like.

In an embodiment, the plurality of examination data may include naked eye acuity, a position of a coma, corneal astigmatism and lens astigmatism, and a ratio between myopia and astigmatism.

In an embodiment, the input data may include information on expected corneal incision of a prospective surgery. For example, the expected corneal incision information may include a degree of corneal incision, a corneal incision position, a corneal incision direction, etc.

In the step of determining the lens power, corneal incision information may be considered. 25 is a diagram for explaining corneal incision information. Specifically, FIG. 25 (a) is a schematic diagram of a corneal incision without astigmatism correction, and FIG. 25 (b) is an exemplary schematic view of a corneal incision with astigmatism correction.

Referring to (a) of FIG. 25 , in an embodiment, when there is no astigmatism correction during vision correction of a prospective surgery, the direction of corneal incision may be the x-axis direction with the pupil as the center. Also, the degree of corneal incision may be 1/4 the length of the pupil.

Referring to (b) of FIG. 25 , in an embodiment, when astigmatism correction is performed during vision correction of a prospective surgery, the direction of corneal incision may be a y-axis direction with respect to the pupil. Also, the degree of corneal incision may be 1/4 the length of the pupil.

25 illustrates an example generated in the course of corneal incision of a prospective surgery, but is not limited thereto, and may be different depending on the degree of astigmatism and astigmatism ratio of the prospective surgery.

In one embodiment, the input data may be used by inputting expected corneal incision information as described with reference to FIG. 25 in addition to the examination data. When the corneal incision information is input, the lens power determination unit may determine the lens power in consideration of this.

5.3 Examples

26 is a diagram illustrating an embodiment of determining the lens power. Referring to FIG. 26 , the lens power determination unit 3300 may output a lens power by using a plurality of examination data of a prospective surgery as input data.

In an embodiment, the lens power determination module 3000 may output the lens power when the inspection data is input to the lens power determination model 130 .

In an embodiment, the lens power determination model 130 may be trained using learning data that inputs inspection data and outputs lens power. In the learning step of the lens power determination model, the learning step S100 described with reference to FIG. 8 may be applied as it is.

Although not shown, in an exemplary embodiment, the lens power determination unit 3300 may output the lens power and corneal incision information by using the examination data of the surgical patient as input data. In the corneal incision process, by outputting the expected lens power at the same time as the corneal incision information such as the degree of corneal incision, the location of the corneal incision, and the degree of corneal incision, the lens power in consideration of factors such as astigmatism corrected according to the corneal incision information can be output have.

In an embodiment, the lens power determination model 130 may be trained using learning data that inputs examination data and outputs lens power and corneal incision information. In the learning step of the lens power determination model, the learning step S100 described with reference to FIG. 8 may be applied as it is.

Although not shown, in an embodiment, the lens power determination unit 3300 may output astigmatism parameters such as lens power, corneal incision information, and surgically induced astigmatism (SIA) by using the examination data of the patient to be operated on as input data. have. In the corneal incision process, factors such as astigmatism corrected according to the corneal incision information and surgery-induced astigmatism are taken into account by outputting the expected lens power simultaneously with the astigmatism parameters such as corneal incision information and surgery-induced astigmatism (SIA). You can print the lens power.

In an embodiment, the lens power determination model 130 may be trained using learning data that inputs examination data and outputs astigmatism parameters such as lens power, corneal incision information, and surgically induced astigmatism (SIA). . In the learning step of the lens power determination model, the learning step S100 described with reference to FIG. 8 may be applied as it is.

27 is a diagram illustrating another embodiment of determining the lens power. Referring to FIG. 27 , the lens power determination unit 3300 may output a lens power by using a plurality of examination data and corneal incision information of a prospective surgery as input data.

In an embodiment, the lens power determination model 130 may be learned using learning data that inputs examination data and corneal incision information and outputs the lens power. In the learning step of the lens power determination model, the learning step S100 described with reference to FIG. 8 may be applied as it is.

Although not shown, in an embodiment, the lens power determining unit 3300 uses a plurality of examination data and corneal incision information of a prospective surgery as input data, and outputs astigmatism parameters such as an astigmatism value (SIA) induced by a predicted operation. can

In an embodiment, the lens power determination model 130 may be trained using learning data that inputs examination data and corneal incision information and outputs astigmatism parameters. In the learning step of the lens power determination model, the learning step S100 described with reference to FIG. 8 may be applied as it is.

Although not shown, in an embodiment, the lens power determining unit 3300 uses a plurality of examination data and corneal incision information of a prospective surgery as input data, and astigmatism parameters such as predicted surgery-induced astigmatism (SIA) and corneal incision. information can be printed.

In an embodiment, the lens power determination model 130 may be trained using learning data that inputs examination data and corneal incision information, and outputs astigmatism parameters and corneal incision information. In the learning step of the lens power determination model, the learning step S100 described with reference to FIG. 8 may be applied as it is.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computing devices and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment, or may be known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic such as floppy disks. - Includes magneto-optical media and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computing device using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

In the above, the configuration and features of the present invention have been described based on the embodiments, but the present invention is not limited thereto, and it will be appreciated by those skilled in the art that various changes or modifications can be made within the spirit and scope of the present invention. It is evident that such changes or modifications are intended to be within the scope of the appended claims.

Claims (14)

In the method of determining a lens to be inserted into the eye during lens insertion using machine learning,
acquiring a plurality of examination data of the prospective surgery; and
By inputting the obtained plurality of examination data of the prospective surgery to a lens decision model, a size of a lens to be inserted into the eye of the surgery prospect among a plurality of lens sizes and a lens power to be inserted into the eye of the surgery prospect out of a plurality of lens powers Including;
The step of determining the lens power,
When a lens of a size to be determined by the lens determination model is inserted into the eye of the surgery prospect, the lens power is determined so that the target vision of the surgery candidate is derived,
The lens crystal model is
A method of determining a lens that is learned based on examination data of an operator who has undergone lens implantation in the past and size information of a lens to which the operator's intraocular insertion is applied. According to claim 1,
The plurality of examination data of the scheduled surgery includes any one of the first data and the second data,
The priority of the input data input to the lens decision model so that the accuracy of the lens size determined by the first data is higher than that of the second data. 3. The method of claim 2,
The step of determining the lens size,
Even when the examination data of the surgery prospect does not include the first data and includes the second data, the second data is input to the lens determination model to determine the size of the lens to be inserted into the eye of the surgery prospect. ,
The accuracy of the lens size determined when the examination data of the surgery prospect includes the first data is the lens size determined when the examination data of the surgery prospect includes the second data other than the first data. A method of determining a lens with a higher than accuracy. According to claim 1,
In the step of determining the lens size,
A method of determining the lens size by calculating the reliability of the accuracy of the lens size derived according to the examination data of the surgery prospect, and presenting the calculated reliability to the user. 3. The method of claim 2,
When the examination data of the surgical patient excludes the first data and includes the second data or other data, estimating the first data from the second data or other data; further comprising ,
The estimating step is
Lens, characterized in that the accuracy for the lens size derived when the first data is input to the lens decision model is higher than the accuracy for the lens size derived when the second data is input to the lens decision model How to decide. 3. The method of claim 2,
The first data of the plurality of examination data of the surgical candidate is ATA (Anterior Chamber Angle), ACD-epi (Anterior Chamber Depth), ACD-endo, CCT (Central Corneal Thickness), CLR (crystalline lens rise), WTW, Axial Length, BUT, the distance between the iris is measured, and the value of the space size for the lens is obtained,
wherein the second data is obtained using a general ophthalmic examination. According to claim 1,
In the step of determining the lens size,
The size of the lens to be inserted into the eye of the surgical person is determined by the lens size of any one of a plurality of predetermined lens sizes. According to claim 1,
In the step of determining the lens size,
A method for determining a lens in which any one of the non-standardized lens sizes derived by inputting the inspection data into the lens determination model, rather than a plurality of predetermined lens sizes, is determined. According to claim 1,
The lens crystal model is
A method of determining a lens that is learned based on the operator's incision information. 10. The method of claim 9,
The obtained plurality of examination data of the prospective surgery includes diopters, astigmatism axis, and astigmatism direction parameters measured from the eyes of the surgery prospect,
A lens determination method for determining the lens power suitable for the target vision of the surgery prospect by inputting the examination data of the surgery prospect and the incision information expected in the corneal incision process of the lens implantation surgery of the prospective surgery to the lens decision model. 10. The method of claim 9,
When a plurality of examination data of the prospective surgery is input to the lens determination model, the lens power for deriving the target visual acuity of the prospective surgery and incision information expected in the corneal incision process of the prospective surgery are determined. How to decide 10. The method of claim 9,
The incision information includes at least one of a corneal incision method, a corneal incision position, a corneal incision direction and/or a corneal incision degree, a coma position, corneal astigmatism and lens astigmatism, myopia, and astigmatism in the corneal incision process of the lens implantation surgery. A method of determining a lens, which is information comprising one. A computer-readable recording medium in which a program for performing the method of any one of claims 1 to 12 is recorded. In the device for determining a lens to be inserted into the eye during lens insertion using machine learning,
a memory for storing a plurality of examination data of a prospective surgery; and
processor; including;
The processor is
Obtaining a plurality of examination data of the surgical candidate stored from the memory,
By inputting the obtained plurality of examination data of the prospective surgery to a lens decision model, the size of the lens to be inserted into the eye of the surgery prospect among the plurality of lens sizes and the lens power to be inserted into the eye of the surgery prospect out of the plurality of lens powers to decide,
When a lens of a size to be determined by the lens determination model is inserted into the eye of the surgery prospect, the lens power is determined so that the target vision of the surgery candidate is derived,
The lens crystal model is
A lens determining device that is learned based on examination data of an operator who has undergone lens implantation in the past, and size information of a lens to which the operator's intraocular insertion is applied.

Images