Methods, terminals, and storage media for extracting intraocular pressure information from anterior segment images

Abstract

This invention provides a method, terminal, and storage medium for extracting intraocular pressure (IOP) information from anterior segment images. First, an anterior segment image of the subject is acquired using a three-dimensional corneal tomography system. Then, anterior segment information is extracted from the image. Finally, the IOP of the subject is determined based on the anterior segment information and a pre-trained IOP prediction model. By using image recognition technology to extract features from the captured anterior segment image and then analyzing them using the trained IOP prediction model, IOP detection can be achieved without contact with the patient or damage to the eyeball, effectively ensuring the safety of the detection process.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology, and particularly relates to a method, terminal and storage medium for extracting intraocular pressure information from anterior segment images. Background Technology

[0002] Intraocular pressure (IOP) is the pressure exerted by the contents of the eye on the eyeball wall and the pressure of the contents themselves. Normal IOP is stable within a certain range, 10 mmHg-21 mmHg, which maintains the normal shape of the eyeball and ensures proper refractive function at the refractive interfaces. The role of IOP is to maintain the normal structure and function of the eyeball. Excessively high or low IOP can lead to pathological changes in the eye's structure, such as glaucoma. Common methods for measuring IOP include contact tonometers (e.g., the Goldmann tonometer) and non-contact tonometers (e.g., the jet tonometer).

[0003] Conventional applanation tonometers inevitably involve contact, requiring local anesthesia and fluorescein staining, making the procedure complex and time-consuming. This is particularly problematic for patients with corneal disease or trauma, and contact measurement increases the risk of viral infection. Jet tonometers also lead to increased aerosol production in the local space. Furthermore, recent studies have found that non-contact tonometer measurement still carries the risk of damage from direct pressure on the corneal endothelium, iris, and anterior lens capsule. Summary of the Invention

[0004] In view of this, the present invention provides a method, terminal and storage medium for extracting intraocular pressure information from anterior segment images, aiming to solve the problem of low safety in conventional intraocular pressure detection methods in the prior art.

[0005] A first aspect of this invention provides a method for extracting intraocular pressure information from anterior segment images, comprising:

[0006] Acquire anterior segment images of the subject from a three-dimensional corneal tomography system;

[0007] Extracting anterior segment information from anterior segment images;

[0008] The intraocular pressure of the subject is determined based on anterior segment information and a pre-trained intraocular pressure prediction model.

[0009] A second aspect of the present invention provides an intraocular pressure information extraction device for anterior segment images, comprising:

[0010] The acquisition module is used to acquire anterior segment images of the subject captured by the three-dimensional corneal tomography system;

[0011] The extraction module is used to extract anterior segment information from anterior segment images;

[0012] The determination module is used to determine the intraocular pressure of the subject based on anterior segment information and an intraocular pressure prediction model established through training.

[0013] A third aspect of the present invention provides a terminal including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the method for extracting intraocular pressure information from anterior segment images as described in the first aspect above.

[0014] A fourth aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method for extracting intraocular pressure information from anterior segment images as described in the first aspect above.

[0015] This invention provides a method, terminal, and storage medium for extracting intraocular pressure (IOP) information from anterior segment images. First, an anterior segment image of the subject is acquired using a three-dimensional corneal tomography system. Next, anterior segment information is extracted from the image. Finally, the IOP of the subject is determined based on the anterior segment information and a pre-trained IOP prediction model. By using image recognition technology to extract features from the captured anterior segment image and then analyzing them using the trained IOP prediction model, IOP detection can be achieved without contact with the patient or damage to the eyeball, effectively ensuring the safety of the detection process. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is an application scenario diagram of the intraocular pressure information extraction method for anterior segment images provided in this embodiment of the invention;

[0018] Figure 2 This is a flowchart illustrating the implementation of the method for extracting intraocular pressure information from anterior segment images provided in this embodiment of the invention.

[0019] Figure 3 This is a schematic diagram of the structure of the intraocular pressure information extraction device for anterior segment images provided in an embodiment of the present invention;

[0020] Figure 4 This is a schematic diagram of the terminal provided in an embodiment of the present invention. Detailed Implementation

[0021] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0022] Figure 1 This is an application scenario diagram of the intraocular pressure information extraction method for anterior segment images provided in this embodiment of the invention. For example... Figure 1 As shown, in some embodiments, the intraocular pressure information extraction method for anterior segment images provided by the present invention can be applied to, but is not limited to, this application scenario. In this embodiment of the invention, the system includes: a three-dimensional corneal tomography scanning device 11 and a terminal 12.

[0023] The three-dimensional corneal tomography scanning device 11 includes a base, a scanning camera, and a support. During the examination, the subject places their face against the support and looks at the scanning camera. The scanning camera captures an image of the subject's anterior segment and reports it to the terminal 12. The terminal 12 can be a mobile phone, computer, etc., and is not limited here.

[0024] Figure 2 This is a flowchart illustrating the implementation of the method for extracting intraocular pressure information from anterior segment images provided in this embodiment of the invention. Figure 2 As shown, in some embodiments, the method for extracting intraocular pressure information from anterior segment images is applied to... Figure 1 In terminal 12, the method includes:

[0025] S210: Acquire an anterior segment image of the subject taken by the three-dimensional corneal tomography system.

[0026] In this embodiment of the invention, the anterior segment image can be in bmp or jpeg format.

[0027] S220, Extract anterior segment information from anterior segment image.

[0028] In this embodiment of the invention, the anterior segment is a part of the eyeball, specifically including: the entire cornea, iris, ciliary body, anterior chamber, posterior chamber, lens suspensory ligaments, anterior chamber angle, part of the lens, peripheral vitreous body, etc. There can be one or more anterior segment images, without limitation. When there are multiple anterior segment images, anterior segment information can be extracted from each image. If the same anterior segment information is extracted from more than a preset number of anterior segment images, it indicates that the anterior segment information is valid; otherwise, it is considered image noise.

[0029] In embodiments of the present invention, the anterior segment information includes at least one of the following: corneal thickness, corneal aberration, corneal curvature, corneal volume, corneal eccentricity, corneal astigmatism, pupil size, pupil position, anterior chamber depth, anterior chamber volume, and anterior chamber angle.

[0030] S230 determines the intraocular pressure of the subject based on anterior segment information and a pre-trained intraocular pressure prediction model.

[0031] In this embodiment of the invention, the intraocular pressure prediction model is a machine learning model or an ensemble model; the machine learning model may include, but is not limited to, at least one of the following: decision tree, linear regression model, neural network model, optical gradient booster, extreme gradient booster model, classification booster model, extra tree regressor model, random forest model; the ensemble model is obtained by combining the random forest model, neural network model, classification booster model and extreme gradient booster model in sequence.

[0032] In this embodiment of the invention, the above models can be trained using, but not limited to, the mean absolute error (MAE) loss function, and cross-validation of, but not limited to, 5x can be applied to compare the results of each iteration.

[0033] In this embodiment of the invention, the coefficient of determination, mean absolute error, mean squared error, and root mean squared error can be used to evaluate the performance of the regression model in predicting the target IOP. The specific calculation formulas are as follows:

[0034]

[0035] in, R 2 These are the coefficients of determination, MAE is the mean absolute error, MSE is the root mean square error, and RMSE is the root mean square error. y i This is the actual value of intraocular pressure. ŷ i It is a predicted value for intraocular pressure. ȳ It is the average of the sample. n It represents the number of samples.

[0036] In this embodiment of the invention, information from 143 anterior segment segments in 7624 images measured by a high-resolution scanning camera was used as feature vectors to construct eight different machine learning models. Five-fold cross-validation was used to train and validate each model. The model with the best performance was selected and its performance was evaluated using mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE), and coefficient of determination (R²) on an external test set consisting of 1907 eyes.

[0037] Therefore, it can be confirmed that the intraocular pressure information extraction method of the anterior segment image of the present invention can accurately predict intraocular pressure. The non-contact prediction method can reduce the spread of aerosols and minimize the spread or infection of viruses in ophthalmology outpatients.

[0038] In this embodiment of the invention, image recognition technology is used to extract features from anterior segment images and then analyze them based on a trained intraocular pressure prediction model. This enables intraocular pressure detection without contact with the patient or damage to the eyeball, effectively ensuring the safety of the detection process. This provides a safe, non-invasive, non-contact, and aerosol-free measurement method.

[0039] In some embodiments, prior to S230, the method further includes: acquiring anterior eye sample images of multiple target individuals, demographic information of each target individual, and sample data of refractive error and intraocular pressure, wherein the ratio of healthy individuals to unhealthy individuals among the target individuals is a first preset ratio; using the anterior segment information of the target individuals extracted from the anterior eye sample images as input, and the demographic information, refractive error, and intraocular pressure sample data of each target individual as output, training the initial prediction model using the mean absolute error loss function and a 5x cross-validation method to obtain an intraocular pressure prediction model.

[0040] In this embodiment of the invention, to ensure the accuracy of the intraocular pressure prediction model, the target population needs to be screened to avoid interference from irrelevant factors in model training. The first preset ratio can be [0.8, 1.2].

[0041] In this embodiment of the invention, segmented images of the anterior eye need to be obtained from one or both eyes of the target person in the same interocular visual environment. Before performing the corneal morphology examination, the target person is asked to blink several times to ensure the integrity and smoothness of the tear film on the corneal surface, thereby reducing the impact on the examination results. To avoid detection bias, all measurements are performed in automatic mode, and a corneal diameter greater than 8 mm is considered acceptable.

[0042] In some embodiments, after acquiring anterior eye sample images of multiple target individuals, demographic information of each target individual, and refractive error and intraocular pressure sample data, the method further includes: forming an original feature set from the anterior eye sample images of multiple target individuals, demographic information of each target individual, and refractive error and intraocular pressure sample data; establishing a first decision tree model based on the original feature set and determining the hyperparameters of the first decision tree model; constructing a first feature based on the hyperparameters; wherein the first feature is uniformly distributed between [0, 1]; constructing a second decision tree model based on the first feature and the original feature set; calculating the importance of each original feature and the first feature in the original feature set based on the second decision tree model; and deleting original features with less importance than the first feature.

[0043] Using the anterior segment information of the target person extracted from the previous eye sample image as input, and the demographic information, refractive error, and intraocular pressure sample data of each target person as output, the initial prediction model is trained using the mean absolute error loss function and the 5-fold cross-validation method to obtain the intraocular pressure prediction model.

[0044] In this embodiment of the invention, these anterior segment features are classified into eight categories according to the measurement type, including G1: corneal thickness; G2: corneal aberration; G3: corneal curvature; G4: corneal volume; G5: anterior chamber depth, volume, and angle; G6: corneal eccentricity; G7: corneal astigmatism; and G8: pupil size and position.

[0045] In some embodiments, after deleting original features whose importance is less than that of the first feature, the method further includes: grouping the original features into pairs to obtain multiple feature pairs; calculating the importance of each feature pair according to the second decision tree model; and using feature pairs whose importance is greater than a preset threshold as combined features.

[0046] In this embodiment of the invention, although the original features can achieve accurate prediction, the number of samples required to train the initial prediction model and obtain the intraocular pressure prediction model is enormous, making the training process complex. To simplify the training of the intraocular pressure prediction model, the m original features can be combined in pairs to obtain m*(m-1) / 2 feature pairs. Then, combined features can be constructed by subtraction or division. The combined features can be: m1-m2, m1 / m2, m2-m1, m2 / m1. m1 and m2 are two features in a feature pair.

[0047] In some embodiments, the anterior segment information of the target person extracted from the anterior eye sample image is used as input, and the demographic information, refractive error, and intraocular pressure sample data of each target person are used as output. The initial prediction model is trained using the mean absolute error loss function and a 5x cross-validation method to obtain the intraocular pressure prediction model. This includes: using the anterior segment information of the target person as input and the demographic information, refractive error, and intraocular pressure sample data of each target person as output, the initial prediction model is trained using the mean absolute error loss function and a 5x cross-validation method to obtain the intraocular pressure prediction model; or, selecting the original features and combined features at a second preset ratio to obtain a training feature set, using the anterior segment information of the target person in the training feature set as input and the demographic information, refractive error, and intraocular pressure sample data of each target person as output, the initial prediction model is trained using the mean absolute error loss function and a 5x cross-validation method to obtain the intraocular pressure prediction model.

[0048] In this embodiment of the invention, an intraocular pressure (IOP) prediction model can be obtained by training an initial prediction model using combined features. While combining features effectively reduces the training process, it also leads to a decrease in prediction accuracy. To ensure accuracy, the initial prediction model can be trained using a combination of original and combined features. For example, 155 important features (143 original features + 12 combined features) can be used to construct the model. This ensures both training speed and accuracy.

[0049] In this embodiment of the invention, by parsing 1060-dimensional raw data, a second decision tree model is used to filter out the raw features from this 1060-dimensional raw data. 917 raw features with lower importance are excluded, leaving 143 raw features to form the raw feature dataset. These 143 raw features are then combined to obtain 12 highly important combined features. Finally, the above 155-dimensional (143+12) features are modeled to obtain the intraocular pressure prediction model.

[0050] In some embodiments, extracting anterior segment information from an anterior segment image includes: performing binarization and denoising preprocessing on the anterior segment image to obtain a binarized image; performing blob analysis on the binarized image to obtain multiple eye feature regions; performing boundary localization on the multiple eye feature regions using the K-means clustering algorithm to obtain boundary information of each eye feature region; and determining anterior segment information based on the boundary information of each eye feature region.

[0051] In this embodiment of the invention, blob analysis involves extracting and labeling connected components in the binary image after foreground / background separation. Each labeled blob represents a foreground object, and then some relevant features of the blob can be calculated. Through blob analysis, the approximate regions where each eye feature is located can be extracted. Then, threshold segmentation is performed based on K-means clustering to identify the boundaries of each region, thereby realizing the extraction of anterior segment information.

[0052] In summary, the beneficial effects of this invention are as follows: by using image recognition technology, after capturing anterior segment images, feature extraction is performed and then analyzed based on a trained intraocular pressure prediction model, intraocular pressure can be detected without contact with the patient or damage to the eyeball, effectively ensuring the safety of the detection process.

[0053] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0054] Figure 3 This is a schematic diagram of the structure of the intraocular pressure information extraction device for anterior segment images provided in an embodiment of the present invention. Figure 3 As shown, in some embodiments, the intraocular pressure information extraction device 3 for anterior segment images includes:

[0055] The acquisition module 310 is used to acquire anterior segment images of the subject captured by the three-dimensional corneal tomography system;

[0056] Extraction module 320 is used to extract anterior segment information from anterior segment images;

[0057] The determination module 330 is used to determine the intraocular pressure of the subject based on the anterior segment information and a pre-trained intraocular pressure prediction model.

[0058] Optionally, the anterior segment image intraocular pressure information extraction device 3 further includes: a training module, used to acquire anterior segment sample images of multiple target individuals, demographic information of each target individual, refractive error and intraocular pressure sample data, wherein the ratio of healthy individuals to unhealthy individuals among the target individuals is a first preset ratio; the anterior segment information of the target individuals extracted from the anterior segment sample images is used as input, and the demographic information, refractive error and intraocular pressure sample data of each target individual are used as output, and the initial prediction model is trained using the mean absolute error loss function and the 5x cross-validation method to obtain the intraocular pressure prediction model.

[0059] Optionally, the training module is also used to assemble an original feature set from anterior eye sample images of multiple target individuals, demographic information of each target individual, and refractive error and intraocular pressure sample data; to build a first decision tree model based on the original feature set and determine the hyperparameters of the first decision tree model; to construct a first feature based on the hyperparameters; wherein the first feature is uniformly distributed between [0, 1]; to construct a second decision tree model based on the first feature and the original feature set; to calculate the importance of each original feature and the first feature in the original feature set based on the second decision tree model; and to delete original features with less importance than the first feature.

[0060] The training module specifically includes: taking the anterior segment information of the target person from the original features as input, and the demographic information, refractive error and intraocular pressure sample data of each target person as output, and using the mean absolute error loss function and 5x cross-validation to train the initial prediction model to obtain the intraocular pressure prediction model.

[0061] Optionally, the training module is also used to group the original features in pairs to obtain multiple feature pairs; calculate the importance of each feature pair according to the second decision tree model; and use feature pairs whose importance is greater than a preset threshold as combined features.

[0062] Optionally, the training module is specifically used to train the initial prediction model using the anterior segment information of the target person as input and the demographic information, refractive error, and intraocular pressure sample data of each target person as output, employing the mean absolute error loss function and a 5x cross-validation method to obtain an intraocular pressure prediction model; or, by selecting the original features and combined features at a second preset ratio to obtain a training feature set, using the anterior segment information of the target person in the training feature set as input and the demographic information, refractive error, and intraocular pressure sample data of each target person as output, employing the mean absolute error loss function and a 5x cross-validation method to train the initial prediction model to obtain an intraocular pressure prediction model.

[0063] Optionally, the intraocular pressure prediction model is a machine learning model or an ensemble model; the machine learning model includes at least one of the following: decision tree, linear regression model, neural network model, optical gradient booster, extreme gradient booster model, classification booster model, extra tree regressor model, random forest model; the ensemble model is obtained by combining the random forest model, neural network model, classification booster model and extreme gradient booster model in sequence.

[0064] Optionally, anterior segment information is extracted from the anterior segment image. The anterior segment information includes: performing binarization and denoising preprocessing on the anterior segment image to obtain a binarized image; performing blob analysis on the binarized image to obtain multiple eye feature regions; performing boundary localization on the multiple eye feature regions according to the K-means clustering algorithm to obtain the boundary information of each eye feature region; and determining the anterior segment information based on the boundary information of each eye feature region.

[0065] Optionally, the anterior segment information includes at least one of the following: corneal thickness, corneal aberration, corneal curvature, corneal volume, corneal eccentricity, corneal astigmatism, pupil size, pupil position, anterior chamber depth, anterior chamber volume, and anterior chamber angle.

[0066] The intraocular pressure information extraction device for anterior segment images provided in this embodiment can be used to execute the above method embodiment. Its implementation principle and technical effect are similar, and will not be described again here.

[0067] Figure 4 This is a schematic diagram of the terminal structure provided in an embodiment of the present invention. Figure 4 As shown, an embodiment of the present invention provides a terminal 4, which includes a processor 40, a memory 41, and a computer program 42 stored in the memory 41 and executable on the processor 40. When the processor 40 executes the computer program 42, it implements the steps in the above embodiments of the methods for extracting intraocular pressure information from anterior segment images, for example... Figure 2 Steps 210 to 240 are shown. Alternatively, when processor 40 executes computer program 42, it implements the functions of each module / unit in the above system embodiments, for example... Figure 3 The functions of modules 310 to 340 are shown.

[0068] For example, computer program 42 can be divided into one or more modules / units, one or more of which are stored in memory 41 and executed by processor 40 to complete the present invention. One or more modules / units can be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 42 in terminal 4.

[0069] Terminal 4 can be a terminal or a server. The terminal can be a mobile phone, MCU, ECU, industrial control computer, etc., without limitation. The server can be a physical server, cloud server, etc., without limitation. Terminal 4 may include, but is not limited to, processor 40 and memory 41. Those skilled in the art will understand that... Figure 4 This is merely an example of terminal 4 and does not constitute a limitation on terminal 4. It may include more or fewer components than shown, or combine certain components, or different components. For example, the terminal may also include input / output devices, network access devices, buses, etc.

[0070] The processor 40 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0071] The memory 41 can be an internal storage unit of the terminal 4, such as a hard disk or RAM of the terminal 4. The memory 41 can also be an external storage device of the terminal 4, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the terminal 4. Furthermore, the memory 41 can include both internal and external storage units of the terminal 4. The memory 41 is used to store computer programs and other programs and data required by the terminal. The memory 41 can also be used to temporarily store data that has been output or will be output.

[0072] This invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps described in the above-described method for extracting intraocular pressure information from anterior segment images.

[0073] A computer-readable storage medium stores a computer program 42. The computer program 42 includes program instructions. When executed by the processor 40, the program instructions implement all or part of the processes in the methods described in the above embodiments. The computer program 42 can also instruct related hardware to complete the process. The computer program 42 can be stored in a computer-readable storage medium. When executed by the processor 40, the computer program 42 can implement the steps of the various method embodiments described above. The computer program 42 includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0074] The computer-readable storage medium can be an internal storage unit of the terminal in any of the foregoing embodiments, such as the terminal's hard disk or memory. The computer-readable storage medium can also be an external storage device of the terminal, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the terminal. Furthermore, the computer-readable storage medium can include both internal storage units and external storage devices of the terminal. The computer-readable storage medium is used to store computer programs and other programs and data required by the terminal. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0075] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0076] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0077] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0078] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0079] In the embodiments provided by this invention, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0080] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0081] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0082] If an integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0083] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for extracting intraocular pressure information from anterior segment images, characterized in that, include: Acquire anterior segment images of the subject from a three-dimensional corneal tomography system; Extract anterior segment information from the anterior segment image; Based on the anterior segment information and the pre-trained intraocular pressure prediction model, the intraocular pressure of the subject is determined; The anterior segment information includes at least one of the following: corneal thickness, corneal aberration, corneal curvature, corneal volume, corneal eccentricity, corneal astigmatism, pupil size, pupil position, anterior chamber depth, anterior chamber volume, and anterior chamber angle; Before determining the intraocular pressure of the subject based on the anterior segment information and a pre-trained intraocular pressure prediction model, the method further includes: Acquire anterior eye sample images of multiple target individuals, demographic information of each target individual, and sample data of refractive error and intraocular pressure; wherein, the ratio of healthy individuals to unhealthy individuals among the target individuals is a first preset ratio; The anterior segment information of the target person extracted from the previous eye sample image is used as input, and the demographic information, refractive error and intraocular pressure sample data of each target person are used as output. The initial prediction model is trained by the mean absolute error loss function and the 5-fold cross-validation method to obtain the intraocular pressure prediction model. The method further includes, after acquiring anterior ocular sample images of multiple target individuals, demographic information of each target individual, and refractive error and intraocular pressure sample data: The original feature set is composed of the anterior eye sample images of the multiple target individuals, the demographic information of each target individual, and the refractive error and intraocular pressure sample data; A first decision tree model is established based on the original feature set, and the hyperparameters of the first decision tree model are determined. A first feature is constructed based on the hyperparameters; wherein the first feature is uniformly distributed between [0, 1]. Based on the first feature and the original feature set, a second decision tree model is constructed; The importance of each original feature and the first feature in the original feature set is calculated based on the second decision tree model. Delete original features that are less important than the first feature; The anterior segment information of the target person extracted from the previous eye sample images is used as input, and the demographic information, refractive error, and intraocular pressure sample data of each target person are used as output. The initial prediction model is trained using the mean absolute error loss function and a 5-fold cross-validation method to obtain the intraocular pressure prediction model, which includes: Using the anterior segment information of the target person in the original features as input, and the demographic information, refractive error and intraocular pressure sample data of each target person as output, the initial prediction model is trained using the mean absolute error loss function and the 5-fold cross-validation method to obtain the intraocular pressure prediction model. The method further includes, after deleting original features whose importance is less than the first feature: The original features are grouped into pairs to obtain multiple feature pairs; The importance of each feature pair is calculated based on the second decision tree model; Feature pairs whose importance is greater than a preset threshold are used as combined features.

2. The method for extracting intraocular pressure information from anterior segment images according to claim 1, characterized in that, The anterior segment information of the target person extracted from the previous eye sample images is used as input, and the demographic information, refractive error, and intraocular pressure sample data of each target person are used as output. The initial model is trained using the mean absolute error loss function and a 5x cross-validation method to obtain the intraocular pressure prediction model, including: Using the anterior segment information of the target person as input and the demographic information, refractive error and intraocular pressure sample data of each target person as output, the initial prediction model is trained using the mean absolute error loss function and the 5x cross-validation method to obtain the intraocular pressure prediction model. Alternatively, the original features and combined features are selected at a second preset ratio to obtain a training feature set. The anterior segment information of the target person in the training feature set is used as input, and the demographic information, refractive error and intraocular pressure sample data of each target person are used as output. The initial prediction model is trained using the mean absolute error loss function and the 5x cross-validation method to obtain the intraocular pressure prediction model.

3. The method for extracting intraocular pressure information from anterior segment images according to claim 1, characterized in that, The intraocular pressure prediction model is a machine learning model or an ensemble model; The machine learning model includes at least one of the following: decision tree, linear regression model, neural network model, optical gradient boosting machine, extreme gradient boosting model, classification boosting model, extra tree regressor model, and random forest model; The ensemble model is obtained by combining the random forest model, the neural network model, the classification boosting model, and the extreme gradient boosting model in sequence.

4. The method for extracting intraocular pressure information from anterior segment images according to claim 1, characterized in that, Extracting anterior segment information from the anterior segment image, the anterior segment information includes: The anterior segment image is preprocessed by binarization and denoising to obtain a binarized image; Blob analysis was performed on the binarized image to obtain multiple eye feature regions; The boundaries of the multiple eye feature regions are located using the K-means clustering algorithm to obtain the boundary information of each eye feature region; The anterior segment information is determined based on the boundary information of each eye feature region.

5. A terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for extracting intraocular pressure information from anterior segment images as described in any one of claims 1 to 4.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method for extracting intraocular pressure information from anterior segment images as described in any one of claims 1 to 4.

Images