Refractive optical subtyping system, refractive development prediction tool, method and apparatus

Abstract

The application discloses a refractive optical subtype classification system, a refractive development prediction tool, a method and equipment. The refractive optical subtype classification system comprises: a data acquisition module for acquiring refractive multi-modal optical characteristic parameters of an individual to be analyzed; a data comparison module for comparing the refractive multi-modal optical characteristic parameters of the individual to be analyzed with preset refractive characteristic vectors; and an optical subtype determination module for determining a refractive optical subtype of the individual to be analyzed according to a comparison result, wherein the refractive optical subtype is obtained through unsupervised machine learning and cluster analysis on a plurality of characteristic vectors. The application discards a prior refractive state label and simply classifies the individual according to objective optical characteristics of the eyeball, thereby exploring an inherent and essential biological optical phenotype. Furthermore, whether the individual has a high myopia risk can be identified earlier, such as when the individual is still a hypermetropia or emmetropia.

Description

Technical Field

[0001] This application relates to the field of medical technology, and in particular to a refractive optical subtype classification system, a refractive development prediction tool, method and device. Background Technology

[0002] The prevalence of myopia has become a major global public health problem, with its trend of onset at younger ages and an increasing proportion of high myopia being particularly worrying. True myopia, once it develops, is usually irreversible. In related technologies, myopia risk assessment heavily relies on central refractive error (such as spherical equivalent power SER) and axial length (AL) measurements. However, these indicators have limited sensitivity and specificity in the early stages of myopia and cannot effectively distinguish between "emmetropic" eyes that are about to develop myopia and "emmetropic" eyes that can maintain healthy vision.

[0003] For example, patent application CN112614593B provides a macroscopic group perspective. By obtaining peripheral retinal defocus distribution maps of myopic individuals, the group is grouped according to central refractive error and the similarity between groups is calculated to construct an "evolutionary tree" describing the progression of myopia from low to high. However, this method groups individuals based on "known myopia status," focusing on post-hoc description rather than pre-hoc prediction, and it does not involve pre-myopic individuals or multimodal optical information.

[0004] For example, patent document CN112530596A extracts the average defocus value of a specific area above the retina (+8° to +16°) as a representative indicator and correlates it with the risk of myopia progression, thus achieving risk stratification for some emmetropic children. However, because the feature parameter used is relatively singular (only using the average defocus value), it fails to fully utilize global and multidimensional optical information, thus limiting its assessment accuracy and interpretability. Furthermore, it only applies to emmetropic children and does not address the high-risk identification of hyperopic children. Summary of the Invention

[0005] Therefore, it is necessary to provide a refractive optical subtype classification system, refractive development prediction tool, method and device to address the above-mentioned technical problems, which can identify the risk of high myopia earlier, such as when the child is still farsighted or emmetropic.

[0006] Firstly, a refractive optical subtype classification system is provided, comprising: The data acquisition module is used to acquire the refractive multimodal optical characteristic parameters of the individual to be analyzed; The data comparison module is used to compare the refractive multimodal optical feature parameters of the individual to be analyzed with a preset refractive feature vector. The preset refractive feature vector is extracted after processing and analysis of the refractive multimodal optical feature parameters of multiple data-providing individuals, and the data-providing individuals include myopic individuals, emmetropic individuals, and hyperopic individuals. An optical subtype determination module is used to determine the refractive optical subtype of the individual to be analyzed based on the comparison results of the data comparison module, wherein the refractive optical subtype is obtained by performing unsupervised machine learning and cluster analysis on multiple feature vectors.

[0007] In some examples, a clinical prediction module is also included to provide a clinical prediction outcome for refractive development based on the refractive optical subtype, wherein a corresponding correlation is established between the refractive optical subtype and the clinical prediction outcome for refractive development.

[0008] In some examples, the refractive multimodal optical characteristic parameters include defocus distribution, higher-order aberrations, and visual quality indices.

[0009] In some examples, the refractive multimodal optical characteristic parameters also include axial length and corneal curvature.

[0010] In some examples, the preset refractive feature vector includes global statistical features, regional statistical features, gradient distribution features, and higher-order aberration coefficient features.

[0011] In some examples, each of the aforementioned refractive optical subtypes has a well-defined purpose.

[0012] In some examples, the associations were established through retrospective and prospective clinical studies of each of the refractive optical subtypes.

[0013] In a second aspect, a refractive development prediction tool is provided, comprising: a refractive optical subtype classification system as described in the first aspect above.

[0014] Thirdly, a method for predicting refractive development is provided, including: Obtain the refractive multimodal optical characteristic parameters of the individual to be analyzed; The refractive multimodal optical characteristic parameters are input into the refractive development prediction tool described in the second aspect above, and the refractive optical subtype is determined after data comparison and analysis. Based on the correlation between the refractive optical subtype and the clinical predicted outcome of refractive development, the predicted result of refractive development is output.

[0015] Fourthly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements the steps of the refractive development prediction method of any possible implementation of the third aspect above.

[0016] Using the embodiments of this application, the refractive optical subtype of the individual to be analyzed can be determined by comparing the refractive multimodal optical characteristic parameters of the individual with a preset refractive feature vector. Since the refractive optical subtype is obtained in advance through unsupervised machine learning and cluster analysis of the refractive multimodal optical characteristic parameters of multiple data-provided individuals, compared with the prior art's method of grouping or risk assessment based on known refractive status (e.g., after being diagnosed with myopia), it abandons the pre-defined refractive status label and performs "blind subtyping" based solely on the objective optical characteristics of the individual's eyeball, thereby uncovering the intrinsic and essential biological optical phenotype. Furthermore, it is possible to identify the risk of high myopia earlier, when the individual is still hyperopic or emmetropic, based on this optical subtype. Attached Figure Description

[0017] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 A structural block diagram of the refractive optical subtype classification system provided in the embodiments of this application; Figure 2a and Figure 2b These are baseline relative retinal refractive power (RPR) topographic maps and baseline spherical aberration (SA) topographic maps, respectively. Figure 3 This is a schematic diagram of the zoning of a two-dimensional retinal refractive topography map; Figure 4 A flowchart of the refractive development prediction method provided in the embodiments of this application; Figure 5 This is a schematic diagram of a nodal chart model; Figure 6 This is a structural block diagram of a computer device provided in an embodiment of this application. Detailed Implementation

[0018] The present application will now be described in further detail with reference to the embodiments and accompanying drawings. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the application. Furthermore, it should be noted that, for ease of description, only the parts relevant to the application are shown in the accompanying drawings.

[0019] It should be noted that, unless otherwise specified, the embodiments and features of the embodiments in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] The following describes in detail, with reference to the accompanying drawings, a refractive optical subtype classification system, a refractive development prediction tool, a method, and an apparatus according to embodiments of this application.

[0021] Figure 1 This is a structural block diagram of a refractive optical subtype classification system according to an embodiment of this application, as shown below. Figure 1 As shown, a refractive optical subtype classification system according to an embodiment of this application includes: a data acquisition module 110, a data comparison module 120, and an optical subtype determination module 130, wherein: The data acquisition module 110 is used to acquire the refractive multimodal optical characteristic parameters of the individual to be analyzed.

[0022] In this context, the individual being analyzed may be a child, and the refractive multimodal optical characteristic parameters may include defocus distribution, higher-order aberrations, and visual quality indices. Of course, in other examples, the refractive multimodal optical characteristic parameters may also include parameters such as axial length and corneal curvature. In other words, the refractive multimodal optical characteristic parameters can be expanded upon based on defocus distribution, higher-order aberrations, and visual quality indices, for example, by adding additional parameters such as axial length and corneal curvature to enrich the dimensions of the characteristic space.

[0023] In a specific example, a wide-area retinal refractive topography detector can be used to detect the fundus retina of the individual being analyzed within a preset range, collecting the individual's refractive multimodal optical characteristic parameters. These parameters include central retinal refractive power and peripheral retinal defocus (i.e., defocus distribution, referring to relative peripheral defocus, Relative Peripheral Refraction, abbreviated RPR), as well as higher-order optical aberration parameters (i.e., higher-order aberrations, such as spherical aberration, abbreviated SA; coma) and visual quality parameters (i.e., visual quality indices, such as Strehl ratio, abbreviated SR; Modulation Transfer Function, abbreviated MTF). These refractive multimodal optical characteristic parameters constitute the unique "optical fingerprint" of the individual being analyzed.

[0024] like Figure 2a As shown, a baseline relative retinal refractive power (RPR) topographic map is presented. Figure 2b The baseline spherical aberration (SA) topographic map is shown. In the refractive power topographic map, positive and negative values ​​on the X-axis represent the nasal and temporal retina, and positive and negative values ​​on the Y-axis represent the superior and inferior retina. Different depth bars represent different refractive power / spherical aberration values.

[0025] In the above example, the wide-field retinal refractive topography detector is, for example, the Earth100 device based on the Hartmann-Shack principle. The preset range is, for example, a field of view of ±40°.

[0026] The data comparison module 120 is used to compare the refractive multimodal optical feature parameters of the individual to be analyzed with a preset refractive feature vector. The preset refractive feature vector is extracted after processing and analysis of the refractive multimodal optical feature parameters of multiple data-providing individuals, and the data-providing individuals include myopic individuals, emmetropic individuals, and hyperopic individuals.

[0027] For example, multiple data providers, such as multiple children in the study, are collected in advance with their refractive multimodal optical characteristic parameters. Then, based on these parameters, a feature vector is obtained for each data provider.

[0028] In the above example, the preset refractive feature vector includes, but is not limited to, global statistical features, regional statistical features, gradient distribution features, and higher-order aberration coefficient features. The multiple children included in the study included children with myopia, emmetropia, and hyperopia.

[0029] Similarly, after the data acquisition module 110 acquires the refractive multimodal optical feature parameters of the individual to be analyzed, feature extraction is performed on the refractive multimodal optical feature parameters of the individual to be analyzed to obtain a feature vector. For example, the refractive multimodal optical feature parameters of the individual to be analyzed are preprocessed, wherein the preprocessing includes standardization and denoising; features are extracted from the preprocessed refractive multimodal optical feature parameters of the individual to be analyzed and a feature vector is formed.

[0030] For example, the refractive multimodal optical feature parameters of the individual to be analyzed are preprocessed, such as standardization and denoising, to improve data quality. Then, feature vectors are extracted from the refractive multimodal optical feature parameters of the individual. These feature vectors are of the same type as the preset refractive feature vectors, including but not limited to: global statistical features, regional features, gradient features, and higher-order aberration features, where: Global statistical characteristics refer to statistical indicators such as the average value and standard deviation of each optical parameter across the entire retinal detection field of view.

[0031] Regional features refer to dividing the retinal field of view into multiple fan-shaped or ring-shaped sub-regions (e.g., the four quadrants of superior nasal, inferior nasal, superior temporal, and inferior temporal), and calculating the average value of various optical parameters in each region.

[0032] like Figure 3The diagram shows a two-dimensional retinal refractive topographic map with the origin of the coordinate axis as the core. The map is divided into eight parts with radii of 20° and 25° and y=0. The upper part of the retina extends from the temporal side to the nasal side and consists of the UZ1, UZ2, UZ3, and UZ4 regions. The lower part of the retina extends from the temporal side to the nasal side and consists of the LZ1, LZ2, LZ3, and LZ4 regions.

[0033] Gradient features refer to comparing the differences or ratios of optical parameters between different regions (such as the nasal and temporal sides, or the upper and lower sides). For example, the difference in defocus between the nasal and temporal regions is calculated to obtain the "nasal-temporal defocus gradient," which is used to quantify the asymmetry of the horizontal refractive power distribution.

[0034] Higher-order aberration features refer to the values ​​of the aberration coefficients of each order of Zernike polynomial.

[0035] The optical subtype determination module 130 is used to determine the refractive optical subtype of the individual to be analyzed based on the comparison result of the data comparison module 120, wherein the refractive optical subtype is obtained by performing unsupervised machine learning and cluster analysis on multiple feature vectors.

[0036] In a specific example, multiple refractive optical subtypes can be determined in advance by performing unsupervised machine learning and cluster analysis on multiple feature vectors using a clustering model. For example, if there are 1000 feature vectors and 5 refractive optical subtypes, then these 1000 feature vectors correspond to these 5 refractive optical subtypes. The correspondence is as follows: one refractive optical subtype can correspond to multiple feature vectors, and one feature vector corresponds to one refractive optical subtype. Thus, the refractive optical subtype of the individual to be analyzed can be determined based on the comparison results of the data comparison module 120.

[0037] In the above description, an unsupervised machine learning algorithm is used beforehand to perform cluster analysis on multiple feature vectors input into a clustering model. Through clustering, multiple data providers can be automatically classified into several refractive optical subtypes according to the similarity of their optical features.

[0038] It should be noted that the number of refractive optical subtypes obtained by clustering can be automatically determined by the algorithm based on data characteristics, or a pre-set empirical value (e.g., dividing into 3-5 optical subtypes). Understandably, the difference from existing technologies lies in the fact that the refractive optical subtype classification in this application is entirely based on the objective optical characteristics of an individual's eye provided by multiple data sources, and is not limited by their original refractive state (e.g., hyperopia / emmetropia / myopia). Therefore, it is able to discover biologically significant refractive optical subtypes that transcend traditional refractive classifications.

[0039] In one embodiment of this application, the unsupervised machine learning algorithm can be implemented as needed, such as K-Means clustering, hierarchical clustering, or DBSCAN algorithm.

[0040] According to the refractive optical subtype classification system of this application, the refractive optical subtype of the individual to be analyzed can be determined by comparing the refractive multimodal optical characteristic parameters of the individual with a preset refractive feature vector. Since the refractive optical subtype is obtained in advance by unsupervised machine learning and cluster analysis of the refractive multimodal optical characteristic parameters of multiple data-provided individuals, compared with the prior art's method of grouping or risk assessment based on known refractive status (e.g., after being diagnosed with myopia), it abandons the pre-defined refractive status label and performs "blind classification" based solely on the objective optical characteristics of the individual's eyeball, thereby uncovering the intrinsic and essential biological optical phenotype. Furthermore, it can identify the risk of high myopia earlier, when the individual is still hyperopic or emmetropic, based on this optical subtype.

[0041] In one embodiment of this application, each refractive optical subtype has a specific definition. Furthermore, the refractive optical subtype classification system further includes: a clinical prediction module ( Figure 1 (Not shown in the image), the clinical prediction module provides clinical predictions of refractive development outcomes based on refractive optical subtypes, wherein a corresponding association is established between the refractive optical subtype and the predicted clinical outcome. In this example, the association is established through retrospective and prospective clinical studies on each of the refractive optical subtypes.

[0042] Specifically, each refractive optical subtype is defined, and a correlation is established between each subtype and its corresponding clinical outcome. For example, for each refractive optical subtype obtained from clustering, the average distribution characteristics of its various optical parameters are calculated to form a unique average "optical feature map" for that subtype, and a clear definition is given accordingly (e.g., a subtype can be defined as "high nasal myopic defocus-low coma type," indicating that its nasal retina exhibits high myopic relative defocus with low higher-order coma values; another subtype can be defined as "symmetrical hyperopic defocus-high coma type," indicating that its entire peripheral retina exhibits symmetrical hyperopic defocus accompanied by high coma). Subsequently, through retrospective and prospective clinical studies, the association between each optical subtype and specific clinical outcomes is established. The above associations include, but are not limited to: For children with hyperopia or emmetropia: the probability of developing myopia in the next 1 to 3 years is significantly higher or lower than that of other subtypes; for children who are already myopic: the rate of increase in axial length of the eye in the next 1 year is significantly faster or slower than that of other subtypes.

[0043] Furthermore, embodiments of this application also provide a refractive development prediction tool, including: a refractive optical subtype classification system according to any of the above embodiments.

[0044] According to the refractive development prediction tool of this application, the refractive optical subtype of the individual to be analyzed can be determined by comparing the refractive multimodal optical characteristic parameters of the individual with a preset refractive feature vector. Since the refractive optical subtype is obtained in advance by unsupervised machine learning and cluster analysis of the refractive multimodal optical characteristic parameters of multiple data-provided individuals, compared with the prior art's method of grouping or risk assessment based on known refractive status (e.g., after being diagnosed with myopia), it abandons the pre-defined refractive status label and simply performs "blind subtyping" based on the objective optical characteristics of the individual's eyeball, thereby discovering the intrinsic and essential biological optical phenotype. Furthermore, it can identify the risk of high myopia earlier, when the individual is still hyperopic or emmetropic, based on this optical subtype.

[0045] Based on the refractive development prediction tool provided in the embodiments of this application, a refractive development prediction method is also provided, such as... Figure 4 As shown, it includes: S401: Obtain the refractive multimodal optical characteristic parameters of the individual to be analyzed; S402: Input the refractive multimodal optical characteristic parameters into the refractive development prediction tool, and determine the refractive optical subtype after data comparison and analysis; S403: Output the refractive development prediction results based on the corresponding correlation between the refractive optical subtype and the clinical predicted outcome of refractive development.

[0046] For example, the system outputs the refractive optical subtype of the individual being analyzed (e.g., classified as "protective (P-type)") and corresponding clinical risk information (e.g., "the probability of myopia development in this type of child within three years is <5%"). This allows for reliable prediction of the risk of myopia development in the analyzed individual, such as a child, enabling early medical intervention to reduce the likelihood of myopia onset. In other words, it can accurately identify individuals with a high risk of developing myopia or those with natural myopia-protective optical characteristics before or at a very early stage of myopia development, providing them with targeted early warnings and intervention guidance.

[0047] like Figure 5 The diagram shows a nomogram model that predicts axial elongation in myopic children after a period of time, such as one year, following orthokeratology lens treatment. This model can significantly reduce the risk of developing high myopia.

[0048] In one embodiment, a computer device is provided. Figure 6 This is a structural block diagram of the computer device provided in the embodiments of this application, with reference to... Figure 6 The computer device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the aforementioned embodiment of the refractive development prediction method. For example, it executes: acquiring the refractive multimodal optical characteristic parameters of the individual to be analyzed; The refractive multimodal optical characteristic parameters are input into the refractive development prediction tool, and the refractive optical subtype is determined after data comparison and analysis. Based on the correlation between the refractive optical subtype and the clinical predicted outcome of refractive development, the predicted result of refractive development is output.

[0049] This application also provides a computer-readable storage medium storing a computer program. When the processor executes the computer program, it implements the aforementioned refractive development prediction method embodiment. For example, it executes: acquiring the refractive multimodal optical characteristic parameters of the individual to be analyzed; The refractive multimodal optical characteristic parameters are input into the refractive development prediction tool, and the refractive optical subtype is determined after data comparison and analysis. Based on the correlation between the refractive optical subtype and the clinical predicted outcome of refractive development, the predicted result of refractive development is output.

[0050] This application provides a computer program product including instructions that, when executed, cause the method described in this application embodiment to be performed. For example, it can execute... Figure 4 The steps of the refractive development prediction method shown include, for example, obtaining the refractive multimodal optical characteristic parameters of the individual to be analyzed; The refractive multimodal optical characteristic parameters are input into the refractive development prediction tool, and the refractive optical subtype is determined after data comparison and analysis. Based on the correlation between the refractive optical subtype and the clinical predicted outcome of refractive development, the predicted result of refractive development is output.

[0051] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0052] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0053] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A refractive optical subtype classification system, characterized in that, include: The data acquisition module is used to acquire the refractive multimodal optical characteristic parameters of the individual to be analyzed; The data comparison module is used to compare the refractive multimodal optical feature parameters of the individual to be analyzed with a preset refractive feature vector. The preset refractive feature vector is extracted after processing and analysis of the refractive multimodal optical feature parameters of multiple data-providing individuals, and the data-providing individuals include myopic individuals, emmetropic individuals, and hyperopic individuals. An optical subtype determination module is used to determine the refractive optical subtype of the individual to be analyzed based on the comparison results of the data comparison module, wherein the refractive optical subtype is obtained by performing unsupervised machine learning and cluster analysis on multiple feature vectors.

2. The refractive optical subtype classification system according to claim 1, characterized in that, It also includes a clinical prediction module for providing clinical prediction outcomes of refractive development based on the refractive optical subtype, wherein a corresponding correlation is established between the refractive optical subtype and the clinical prediction outcomes of refractive development.

3. The refractive optical subtype classification system according to claim 1, characterized in that, The refractive multimodal optical characteristic parameters include defocus distribution, higher-order aberrations, and visual quality indices.

4. The refractive optical subtype classification system according to claim 3, characterized in that, The refractive multimodal optical characteristic parameters also include axial length and corneal curvature.

5. The refractive optical subtype classification system according to claim 1, characterized in that, The preset refractive feature vector includes global statistical features, regional statistical features, gradient distribution features, and higher-order aberration coefficient features.

6. The refractive optical subtype classification system according to claim 1, characterized in that, Each of the aforementioned refractive optical subtypes has a clear definition.

7. The refractive optical subtype classification system according to claim 2, characterized in that, The associations were established through retrospective and prospective clinical studies of each of the aforementioned refractive optical subtypes.

8. A tool for predicting refractive development, characterized in that, Including the refractive optical subtype classification system according to any one of claims 1-7.

9. A method for predicting refractive development, characterized in that, include: Obtain the refractive multimodal optical characteristic parameters of the individual to be analyzed; The refractive multimodal optical characteristic parameters are input into the refractive development prediction tool as described in claim 8, and the refractive optical subtype is determined after data comparison and analysis. Based on the correlation between the refractive optical subtype and the clinical predicted outcome of refractive development, the predicted result of refractive development is output.

10. A computer device 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 program, it implements the refractive development prediction method according to claim 9.

Images