A method, system and device for constructing a digital twin eyeball model and a storage medium

Abstract

This invention relates to the field of ophthalmology, and more particularly to a method, software system, device, and storage medium for constructing a digital twin eye model. The construction method includes: acquiring multi-source input data of the target eye; performing three-dimensional reconstruction to obtain the anterior and posterior segment three-dimensional structures; constructing a virtual three-dimensional structure of the intermediate segment based on axial length data and the anterior and posterior segment three-dimensional structures as constraints; fusing these data to form a full-eye three-dimensional geometric model; and mapping relevant characteristic measurements to corresponding regions in the full-eye three-dimensional geometric model to construct the digital twin eye model. This invention systematically solves three core problems: "full-eye modeling," "gold standard replacement," and "functional simulation," ultimately producing a complete, accurate, usable, and reliable digital twin eye model, providing a revolutionary technical means for precise research, in vitro simulation, disease prediction, and prevention of ophthalmic diseases.

Description

Technical Field

[0001] This invention relates to the field of ophthalmology, and more particularly to a method, system, device, and storage medium for constructing a digital twin eye model. Background Technology

[0002] In recent years, the application of digital twin (DT) technology in the medical field has received increasing attention. In ophthalmology, constructing a digital twin eye model that can dynamically reflect the morphology and function of the eyeball is expected to open up new avenues for disease mechanism research, progression prediction, and personalized intervention.

[0003] Many ophthalmic diseases, such as pathological myopia and glaucoma, are accompanied by significant changes in eyeball morphology, making precise and convenient tools for assessing eyeball morphology urgently needed in clinical practice. For example, in myopia research, morphological parameters such as axial length and scleral curvature are closely related to the occurrence and development of myopia. If a digital twin eyeball model integrating biomechanical properties and morphological parameters can be established, the dynamic trajectory of eyeball changes during myopia progression can be simulated, thereby enabling quantitative analysis and risk stratification of myopia development. This will not only help to deepen the understanding of disease mechanisms but also provide crucial evidence for early warning and personalized prevention and control.

[0004] Currently, three-dimensional models of the eyeball constructed based on magnetic resonance imaging (MRI) data are considered the "gold standard" for morphological assessment, and theoretically can provide an ideal foundation for achieving high-precision digital twins of the eye. However, MRI examinations are costly, time-consuming, and inconvenient, making them unsuitable for large-scale population screening and long-term follow-up, thus significantly limiting their widespread clinical application.

[0005] In recent years, ultra-widefield swept-source optical coherence tomography (UWF SS-OCT) technology has developed rapidly. Its wide scanning range, covering a 120° to 150° area in the posterior pole, has made it possible to achieve high-resolution three-dimensional imaging of the posterior segment of the eye. Based on this technology, existing studies have been able to calculate the curvature of the posterior retinal surface to assess complications related to high myopia, and it has been preliminarily applied to morphological studies of childhood myopia. However, these studies still have significant shortcomings: firstly, they only focus on the local morphology of the posterior segment, lacking simultaneous integration of the anterior segment (such as the cornea and lens), and cannot construct a complete three-dimensional model of the whole eye; secondly, and more importantly, the constructed models have not been systematically quantitatively compared and validated with 3D-MRI, the gold standard, and their geometric accuracy and reliability have not been fully confirmed.

[0006] Therefore, existing technologies need to be improved. Summary of the Invention

[0007] In view of the shortcomings of the prior art, the purpose of this invention is to provide a method, system, device and storage medium for constructing a digital twin eye model, in order to solve the problem of the lack of a way to construct a complete, high-precision and rigorously validated digital twin eye model under clinically available conditions.

[0008] The technical solution of the present invention is as follows: In a first aspect, the present invention provides a method for constructing a digital twin eyeball model, comprising the following steps: S1. Obtain multi-source input data of the target eyeball, the multi-source input data including: axial length data, at least one measurement value of the biomechanical or biophysical properties of the eyeball, and three-dimensional scan datasets of the anterior segment and posterior segment; S2. Perform three-dimensional reconstruction on the anterior segment three-dimensional scan dataset and the posterior segment three-dimensional scan dataset respectively to obtain the three-dimensional structure of the anterior segment and the three-dimensional structure of the posterior segment. S3. Based on the axial length data, and constrained by the rear geometric contour of the anterior segment three-dimensional structure and the front geometric contour of the posterior segment three-dimensional structure, construct a virtual three-dimensional structure of the middle segment of the target eyeball; S4. Integrate the three-dimensional structure of the anterior segment, the virtual three-dimensional structure of the middle segment, and the three-dimensional structure of the posterior segment to form a full-eye three-dimensional geometric model of the target eyeball; S5. Associate and map the at least one biomechanical or biophysical characteristic measurement value to the corresponding region in the three-dimensional geometric model of the whole eye, thereby constructing the digital twin eyeball model.

[0009] Optionally, in step S1, the anterior segment three-dimensional scan dataset and the posterior segment three-dimensional scan dataset are acquired by optical coherence tomography (SS-OCT) with an ultra-wide field scanning source and a scanning range of 120-150°.

[0010] Optionally, the axial length ranges from 22 to 32 mm.

[0011] The three-dimensional geometric model of the entire eye (CET-1 model) was quantitatively compared with a three-dimensional reference model of the same target eye reconstructed based on magnetic resonance imaging data to verify the geometric accuracy of the entire eye three-dimensional geometric model. The comparison revealed that within the 22-32mm axial length range, the entire eye three-dimensional geometric model obtained by this invention has a high degree of matching with the three-dimensional reference model reconstructed based on magnetic resonance imaging data. This indicates that within the 22-32mm axial length range, the geometric model of this invention has accuracy comparable to 3D-MRI, with higher resolution (1000 times), displaying more details. Furthermore, the CET-1 model requires only about 10 seconds for anterior and posterior segment examination, significantly improving screening speed. This provides a simple, easy-to-use, and non-invasive new method for assessing eye morphology, offering an important tool for establishing digital twin eyes.

[0012] Optionally, in step S2, the three-dimensional reconstruction of the anterior segment three-dimensional scan dataset includes: Based on the aforementioned anterior segment 3D scan dataset, anterior segment images are obtained; The anterior segment image is refractively corrected to obtain the corrected image; Based on the corrected image, the corneal and lens surfaces are segmented and extracted; The corneal and / or lens surfaces are fitted with a surface to reconstruct their optical surfaces.

[0013] Optionally, in step S3, a virtual three-dimensional structure of the middle segment of the target eyeball is constructed by interpolation fitting. Specifically, the virtual three-dimensional structure of the middle segment of the target eyeball can be constructed based on a surface interpolation algorithm using thin-plate spline functions or a deformation algorithm based on an elasticity model.

[0014] Optionally, in step S1, the at least one biomechanical or biophysical characteristic measurement of the eyeball includes any one or more of the following: intraocular pressure, corneal mechanical parameters, scleral mechanical parameters, and refractive mechanical parameters.

[0015] Secondly, the present invention provides a digital twin eyeball model system, comprising: The data acquisition and input module is used to acquire multi-source input data of the target eyeball. The multi-source input data includes: axial length data, at least one biomechanical or biophysical characteristic measurement value, and three-dimensional scan datasets of the anterior segment and posterior segment. The 3D modeling engine module, which is communicatively connected to the data acquisition and input module, performs 3D reconstruction on the anterior segment 3D scan dataset and the posterior segment 3D scan dataset to obtain the anterior segment 3D structure and the posterior segment 3D structure, respectively. Based on the axial length data and constrained by the rear geometric contour of the anterior segment 3D structure and the front geometric contour of the posterior segment 3D structure, a virtual 3D structure of the middle segment of the eyeball is constructed. The anterior segment 3D structure, the virtual 3D structure of the middle segment, and the posterior segment 3D structure are fused to generate a full-eye 3D geometric model of the target eyeball. The digital twin construction module is communicatively connected to the 3D modeling engine module and the data acquisition and input module, and associates and maps the at least one biomechanical or biophysical characteristic measurement value to the corresponding anatomical structure region in the complete full-eye 3D geometric model, thereby constructing the digital twin eyeball model that can be used for simulation analysis.

[0016] Optionally, the digital twin eye model system further includes: The simulation analysis module is communicatively connected to the digital twin construction module and is used to perform simulation analysis of the physiological or pathological state of the eyeball based on the digital twin eyeball model.

[0017] Thirdly, the present invention provides a terminal comprising: a memory, a processor, and a computing program stored in the memory and executable on the processor, wherein the processor implements the steps of the method when executing the computer program.

[0018] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that can be executed to implement the steps of the method for constructing the digital twin eyeball model.

[0019] Beneficial effects: This invention provides a method, system, device and storage medium for constructing a digital twin eyeball model.

[0020] This invention overcomes the limitation of existing UWF-based SS-OCT research, which can only acquire local information of the posterior segment and cannot construct a whole-eye model. Through a unique technical path of "acquiring multi-source data → reconstructing the anterior and posterior segments separately → constructing a virtual intermediate segment → fusing them into a whole-eye model," it achieves, for the first time, the digital reconstruction of the complete three-dimensional geometry of the eyeball, including the anterior segment (cornea and lens), posterior segment (retina and choroid), and intermediate sclera, based on clinically convenient SS-OCT technology. This provides an unprecedented and complete data foundation for comprehensive ocular morphological analysis and biomechanical research. This invention combines data from the anterior and posterior segments to obtain their structures. More importantly, by using axial length data to geometrically constrain and fit the anterior and posterior segment structures, the constructed whole-eye model possesses anatomical rationality similar to that based on individual 3D-MRI data in key dimensions. Therefore, without relying on 3D-MRI, it provides a high-precision morphological assessment tool comparable to it, greatly expanding the application scenarios of precision ophthalmic examinations. This invention actively correlates and maps at least one biomechanical or biophysical property measurement (intraocular pressure, biomechanical properties, etc.) to the corresponding region of a three-dimensional geometric model. This results in a final output that is no longer an "empty shell" model, but a "digital twin" integrating individualized anatomical structures and physiological / mechanical parameters. This twin can serve as an in vitro experimental platform to simulate the morphological and functional responses of the eyeball under conditions such as myopia progression, intraocular pressure changes, and surgical interventions, providing a powerful new tool for disease mechanism research, treatment plan simulation, and personalized risk assessment. Through quantitative comparison of the three-dimensional morphology with a three-dimensional reference model of the same target eyeball reconstructed based on magnetic resonance imaging data, the whole-eye three-dimensional geometric model obtained by this invention shows a high degree of matching with the three-dimensional reference model reconstructed based on magnetic resonance imaging data within the 22-32mm axial length range. In summary, this invention systematically solves three core problems—"whole-eye modeling," "gold standard replacement," and "functional simulation"—through an innovative and clinically feasible process, ultimately producing a complete, accurate, usable, and reliable digital twin eye model, providing a revolutionary technical means for the precise research, prediction, and prevention of ophthalmic diseases. Attached Figure Description

[0021] Figure 1 This is a flowchart of a preferred embodiment of the method for constructing a digital twin eyeball model in this invention.

[0022] Figure 2 This is a flowchart illustrating the three-dimensional reconstruction of three-dimensional scan datasets of the anterior and posterior segments of the eye, respectively, in one implementation method.

[0023] Figure 3 This is a full-eye three-dimensional geometric model of the target eyeball.

[0024] Figure 4 This is a flowchart illustrating a method for 3D reconstruction of an anterior segment 3D scan dataset.

[0025] Figure 5 This is a schematic diagram of the structure of an eye digital twin system in one embodiment.

[0026] Figure 6 This is a schematic diagram of the structure of a digital twin system for the eyeball in a preferred embodiment.

[0027] Figure 7 This is a functional principle block diagram of a preferred embodiment of the terminal in this invention. Detailed Implementation

[0028] This invention provides a method, system, device, and storage medium for constructing a digital twin eyeball model. To make the objectives, technical solutions, and effects of this invention clearer and more explicit, the invention is further described in detail below. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0029] While existing UWF SS-OCT technology can acquire high-resolution 3D data of the anterior and posterior segments separately, these two segments are independent in anatomical coordinate systems. Furthermore, the ciliary body and peripheral sclera in the mid-segment of the eye (i.e., the area between the posterior boundary of the anterior segment scan and the anterior boundary of the posterior segment scan) cannot be directly imaged, creating a "data vacuum." Therefore, the primary challenge in achieving whole-eye digital twins is how to spatially unify the discrete anterior and posterior segment data and reliably reconstruct the invisible intermediate transition structures to generate a complete, closed, and anatomically sound 3D eyeball shell. This involves not only data stitching but also creative modeling problems related to anatomy and biomechanics. Current technologies only utilize localized data.

[0030] Based on this, this embodiment provides a method for constructing a digital twin eyeball model, such as... Figure 1 As shown, it includes the following steps: S1. Obtain multi-source input data of the target eyeball, the multi-source input data including: axial length data, at least one measurement value of the biomechanical or biophysical properties of the eyeball, and three-dimensional scan datasets of the anterior segment and posterior segment; S2. Perform 3D reconstruction on the anterior segment 3D scan dataset and the posterior segment 3D scan dataset respectively to obtain the 3D structure of the anterior segment and the 3D structure of the posterior segment, such as... Figure 2 As shown; S3. Based on the axial length data, and constrained by the rear geometric contour of the anterior segment three-dimensional structure and the front geometric contour of the posterior segment three-dimensional structure, construct a virtual three-dimensional structure of the middle segment of the target eyeball; S4. Integrate the anterior segment three-dimensional structure, the intermediate segment virtual three-dimensional structure, and the posterior segment three-dimensional structure to form a full-eye three-dimensional geometric model of the target eyeball. The result is as follows: Figure 3 As shown; S5. Associate and map the at least one biomechanical or biophysical characteristic measurement value to the corresponding region in the three-dimensional geometric model of the whole eye, thereby constructing the digital twin eyeball model.

[0031] It should be noted that this embodiment uses the individual axial length of the eye as a key constraint and the separately acquired three-dimensional structures of the anterior and posterior segments as fixed boundaries. By reconstructing the intermediate virtual portion, a full-eye model is formed through fusion. Subsequently, independently measured biomechanical parameters are mapped onto this geometric model, completing the functionalization of the digital twin. Finally, its geometric accuracy is fully verified through high-precision spatial registration and comparison with a 3D-MRI model. Therefore, this invention, through its unique technical path of "acquiring multi-source data → separately reconstructing the anterior and posterior segments → constructing a virtual intermediate segment → fusing into a full-eye model," is the first to achieve, based on clinically convenient technologies (such as SS-OCT), the digital reconstruction of the complete three-dimensional geometry of the eyeball, including the anterior segment (including the cornea and lens), posterior segment (including the retina and choroid), and intermediate scleral region. This provides an unprecedented and complete data foundation for comprehensive ocular morphological analysis and biomechanical research. In S2 of this invention, the anterior and posterior segment structures are obtained by combining data from the anterior and posterior segments. More importantly, S3 utilizes axial length data to geometrically constrain and fit the anterior and posterior segments of the eye. The resulting whole-eye model possesses anatomical plausibility in key dimensions similar to that based on individual MRI data. This provides a high-precision morphological assessment tool comparable to MRI without relying on it, significantly expanding the application scenarios for precise ophthalmic examinations. This invention actively maps at least one biomechanical or biophysical property measurement (intraocular pressure, biomechanical properties, etc.) to the corresponding region of the three-dimensional geometric model. This results in a digital twin that is no longer an "empty shell" model, but rather a "digital twin" integrating individualized anatomical structures and physiological / mechanical parameters. This twin can serve as an in vitro experimental platform to simulate the morphological and functional responses of the eyeball under conditions such as myopia progression, intraocular pressure changes, and surgical interventions, providing a powerful new tool for disease mechanism research, treatment plan simulation, and personalized risk assessment. By quantitatively comparing the three-dimensional morphology with a three-dimensional reference model of the same target eye reconstructed based on magnetic resonance imaging (MRI) data, the whole-eye three-dimensional geometric model obtained by this invention shows a high degree of matching with the three-dimensional reference model reconstructed based on MRI data within a certain axial length range. In summary, this invention systematically solves the three core problems of "whole-eye modeling," "gold standard replacement," and "functional simulation" through an innovative and clinically feasible process, ultimately producing a complete, accurate, usable, and reliable digital twin eye model, providing a revolutionary technical means for the precise research, prediction, and prevention of ophthalmic diseases.

[0032] In some specific embodiments, in step S1, the anterior segment three-dimensional scan dataset and the posterior segment three-dimensional scan dataset are acquired by optical coherence tomography (SS-OCT) with an ultra-wide field scanning source and a scanning range of 120-150°.

[0033] It should be noted that this embodiment uses ultra-wide field scanning optical coherence tomography (SS-OCT) to acquire the anterior and posterior segment 3D scan datasets. Its rapid scanning speed allows the entire data acquisition process to be completed within tens of seconds, perfectly meeting the clinical need for rapid, non-invasive, and accurate examination. Furthermore, the wide scanning range allows for the acquisition of a large area of ​​coherent anterior and posterior segment data necessary for constructing a whole-eye model in a single scan, fundamentally avoiding data stitching and ensuring the quality and efficiency of the source data. The wide field of view enables clear imaging and precise extraction of the anterior segment boundaries (such as the retina), providing accurate input conditions for subsequent high-confidence, low-reconstruction reconstruction of the intermediate transition zone using constrained fitting algorithms.

[0034] The axial length data in this embodiment can be obtained through an optical biometer, or it can be acquired through ultra-wide field scanning optical coherence tomography (SS-OCT).

[0035] At least one biomechanical or biophysical property measurement value can be obtained through corresponding dedicated clinical measurement equipment or computational analysis.

[0036] In some specific embodiments, in step S1, the at least one biomechanical or biophysical characteristic measurement value of the eyeball includes any one or more of the following: intraocular pressure value, corneal mechanical parameters, scleral mechanical parameters, and refractive mechanical parameters. Intraocular pressure value can be obtained by measuring with a Goldmann applanation tonometer or a non-contact tonometer; corneal mechanical parameters (such as corneal hysteresis, corneal resistance factor, etc.) can be obtained by measuring with an Oral Response Analyzer (ORA) or a corneal biomechanical analyzer such as the Corvis ST; refractive mechanical parameters (such as equivalent spherical power) can be obtained by examining with an automated refractometer or a comprehensive refractometer; and other biophysical characteristic measurements (such as central corneal thickness, anterior chamber depth, etc.) can also be automatically extracted and calculated from the anterior segment SS-OCT three-dimensional data using image analysis software.

[0037] In one embodiment, the axial length ranges from 22 to 32 mm, and the axial length can be 22 mm, 23 mm, 24 mm, 25 mm, 26 mm, 27 mm, 28 mm, 29 mm, 30 mm, 31 mm, or 32 mm.

[0038] The three-dimensional geometric model of the entire eye (CET-1 model) was quantitatively compared with a three-dimensional reference model of the same target eye reconstructed based on magnetic resonance imaging data to verify the geometric accuracy of the entire eye three-dimensional geometric model. The comparison revealed that within the 22-32mm axial length range, the entire eye three-dimensional geometric model obtained by this invention has a high degree of matching with the three-dimensional reference model reconstructed based on magnetic resonance imaging data. This indicates that within the 22-32mm axial length range, the geometric model of this invention has accuracy comparable to 3D-MRI, with higher resolution (1000 times), displaying more details. Furthermore, the CET-1 model requires only about 10 seconds for anterior and posterior segment examination, significantly improving screening speed. This provides a simple, easy-to-use, and non-invasive new method for assessing eye morphology, offering an important tool for establishing digital twin eyes.

[0039] In one implementation, in step S2, the three-dimensional reconstruction of the anterior segment three-dimensional scan dataset is performed, such as... Figure 2 , Figure 4 As shown, it includes: S21. Based on the aforementioned anterior segment 3D scan dataset, obtain anterior segment images, such as... Figure 2 As shown in A; S22. Perform refractive correction on the anterior segment image to obtain the corrected image; S23. Based on the corrected image, segment and extract the corneal and lens surfaces, such as... Figure 2 As shown in B; S24. Perform surface fitting on the cornea and / or lens surface to reconstruct its optical surface, such as... Figure 2 As shown in C.

[0040] It should be noted that, in this embodiment, based on the aforementioned anterior segment 3D scan dataset (e.g., a volume data file in .img format), the device's accompanying software or a custom reading program is used to parse it into a series of continuous 2D tomographic images (B-scans) according to the scanning sequence and spatial encoding. These images form the basis for all subsequent processing. The purpose is to convert the 3D data into a 2D sequence that can be manipulated layer by layer by image processing algorithms, facilitating operations such as correction and segmentation. When the OCT beam passes through media with different refractive indices, such as the cornea, aqueous humor, and lens, refraction occurs, resulting in a systematic deviation between the depth and contour position of tissue structures in the images acquired by the device and their actual physical positions. For example, the posterior surface of the cornea may appear more forward in the image than it actually is. A reverse ray tracing algorithm based on Snell's law can be used. This algorithm, knowing the incident angle of the OCT beam and the assumed or measured refractive indices of each medium (air, cornea, aqueous humor, etc.), calculates the refraction path of the light at each interface to inversely deduce the true coordinates of the tissue interface in physical space. The output is a set of anterior segment tomographic images with corrected geometric distortions. The positions and morphologies of surfaces such as the cornea and lens are closer to their anatomical counterparts, thus requiring refractive correction. On the refractively corrected images, a suitable image segmentation algorithm (e.g., a deep learning-based segmentation model, such as U-Net or its variants) is used for pixel-level classification. This model, trained on a large amount of labeled data, can automatically and accurately identify the contours of key interfaces such as the corneal epithelium, corneal endothelium, anterior lens capsule, and posterior lens capsule, thus outputting a two-dimensional contour point set for each key interface on all B-scan images. Combining these point sets according to their three-dimensional spatial coordinates forms a three-dimensional point cloud of the anterior and posterior surfaces of the cornea and lens. Finally, surface fitting is performed on the discrete three-dimensional point cloud data. For surfaces with regular optical properties, such as the cornea and lens, Zernike polynomials are an extremely effective fitting tool. Zernike polynomials are a set of polynomials orthogonal on a unit circle, well-suited for describing wavefront aberrations or morphology of circular optical surfaces. The method involves fitting the coordinates of extracted corneal or lens surface point clouds to a Zernike polynomial surface equation of a finite order (such as 7th or 10th order). This process uses mathematical optimization methods, such as least squares, to find a smooth, continuous mathematical surface that best matches all data points, resulting in a smooth corneal and lens surface model expressed by mathematical equations. These models can be directly used for ray tracing to calculate refractive power or converted into triangular mesh models for biomechanical analysis.

[0041] The refractive correction step in this embodiment fundamentally eliminates the inherent optical distortion of OCT imaging, enabling the reconstructed key biometric parameters such as corneal curvature, anterior chamber depth, and lens thickness to achieve sub-micron anatomical accuracy. This is the cornerstone for digital twins to be used in reliable optical simulations (such as vision prediction and intraocular lens calculation). Uncorrected models will produce optical simulation results that deviate from actual physiological states. Employing segmentation extraction of the corneal and lens surfaces using methods such as deep learning can stably and rapidly extract individualized tissue boundaries from images of varying quality, unaffected by operator subjectivity, ensuring the repeatability and high efficiency of the reconstruction results, and meeting the needs of large-scale clinical applications. Furthermore, the use of Zernike polynomial fitting not only yields a smooth model but, more importantly, a parameterized expression with clear optical meaning. This expression can directly read optical parameters such as curvature and aberrations; it has excellent compatibility with ray tracing software; compared to mesh models composed of millions of polygons, the parameterized surface data is extremely small, resulting in high computational efficiency; and it provides an ideal mathematical foundation for subsequent advanced applications such as "model-based inverse optimization of refractive states."

[0042] Step S2, the step of performing 3D reconstruction on the posterior three-dimensional scan dataset is similar to the step of performing 3D reconstruction on the anterior three-dimensional scan dataset, and will not be described again. Figure 2 The middle D is based on the acquisition of anterior segment images from the aforementioned anterior segment 3D scan dataset. Figure 2 The middle E is based on the corrected image, segmenting and extracting the surface. Figure 2 F is the optical surface reconstructed by surface fitting.

[0043] In some implementations, in step S3, a virtual three-dimensional structure of the middle segment of the target eyeball is constructed by interpolation fitting. Specifically, the virtual three-dimensional structure of the middle segment of the target eyeball can be constructed based on a surface interpolation algorithm using thin-plate spline functions or a deformation algorithm based on an elasticity model.

[0044] It should be noted that constructing the virtual three-dimensional structure of the intermediate segment is the core innovative step of this invention, aiming to solve the "data vacuum" problem that the mid-segment of the eyeball (ciliary body and surrounding scleral region) cannot be directly imaged by OCT. Since the posterior geometric contour of the anterior segment three-dimensional structure (e.g., the curved surface of the posterior capsule of the lens) and the anterior geometric contour of the posterior segment three-dimensional structure (e.g., the curved surface of the ora serrata region of the retina) have been obtained in the aforementioned steps, they are known and fixed three-dimensional spatial curves or point clouds.

[0045] For example, a surface interpolation algorithm based on thin-plate spline functions is a mathematical interpolation technique based on geometric constraints. Its core idea is to generate the "smoothest" transition surface between known boundaries. Thin-plate spline interpolation finds a smooth surface that passes through or approximates these given boundary points by solving a variational problem that minimizes bending energy. This bending energy function corresponds to the square integral of the surface curvature, ensuring that the generated intermediate surface is geometrically the "most natural" or "least strain-energy" transitional form. Combined with axial length data, this provides crucial axial scale constraints for the interpolation process, ensuring that the generated virtual structure conforms to the actual individual anatomy in terms of length proportions. The specific steps can be summarized as follows: 1. Boundary point set extraction and correspondence: The posterior boundary point set P-front is obtained by upsampling from the reconstructed anterior segment model, and the anterior boundary point set -back is obtained by upsampling from the posterior segment model. Based on anatomical continuity, a rough point-to-point or region correspondence is established for these two sets of points. 2. Construct the interpolation function: Define the radial basis function φ(r) = r for thin plate splines. 2 log(r) (where r is the distance between points). By solving a system of linear equations, a set of coefficients is determined such that the three-dimensional displacement field, formed by the linear combination of these basis functions, can smoothly map P_front to P_back, or generate a continuous family of surfaces between them. 3. Surface generation: In the three-dimensional parameter space of the entire intermediate region, a series of continuous isoparametric surfaces located between the front and rear boundaries are calculated and generated using the determined interpolation function. These surfaces together constitute the virtual three-dimensional shell of the intermediate segment.

[0046] For example, the deformation algorithm based on the elasticity model is a deformation technique based on physical simulation. Its core idea is to treat the intermediate segment as an elastic shell with specific material properties, deforming under boundary constraints to reach equilibrium. The specific steps can be summarized as follows: 1. Initial mesh generation and material assignment: Based on the front and rear boundaries, generate an initial finite element mesh (usually a shell mesh) connecting the two. Assign approximate mechanical properties of scleral tissue to the mesh elements. 2. Apply boundary conditions: "Bind" or force displacement matching the nodes corresponding to P-front and P-back on the initial mesh with the corresponding nodes on the reconstructed anterior segment posterior surface and posterior segment anterior surface, respectively. 3. Solve and iterate: Run the finite element solver to calculate the nodal displacements when the mesh reaches mechanical equilibrium under given boundary conditions. Update the mesh node coordinates to obtain the deformed structure. 4. Smoothing and post-processing: Perform necessary smoothing on the solved mesh to eliminate local irregularities caused by possible numerical instability, outputting the final virtual 3D structure.

[0047] This embodiment also provides a digital twin eyeball model system, such as Figure 5As shown, it includes: The data acquisition and input module 1 is used to acquire multi-source input data of the target eyeball. The multi-source input data includes: axial length data, at least one biomechanical or biophysical characteristic measurement value, and three-dimensional scan datasets of the anterior segment and posterior segment. The 3D modeling engine module 2, which is communicatively connected to the data acquisition and input module 1, performs 3D reconstruction on the anterior segment 3D scan dataset and the posterior segment 3D scan dataset to obtain the anterior segment 3D structure and the posterior segment 3D structure, respectively. Based on the axial length data and constrained by the rear geometric contour of the anterior segment 3D structure and the front geometric contour of the posterior segment 3D structure, a virtual 3D structure of the middle segment of the eyeball is constructed. The anterior segment 3D structure, the virtual 3D structure of the middle segment, and the posterior segment 3D structure are fused to generate a full-eye 3D geometric model of the target eyeball. The digital twin construction module 3 is communicatively connected to the 3D modeling engine module 2 and the data acquisition and input module 1, and associates and maps the at least one biomechanical or biophysical characteristic measurement value to the corresponding anatomical structure region in the complete full-eye 3D geometric model, thereby constructing the digital twin eyeball model that can be used for simulation analysis.

[0048] In a preferred embodiment, such as Figure 6 As shown, the digital twin eye model system also includes: The simulation analysis module 4 is communicatively connected to the digital twin construction module and is used to perform simulation analysis of the physiological or pathological state of the eyeball based on the digital twin eyeball model.

[0049] For example, it can be used for simulating the risk of pathological myopia progression, simulating intraocular pressure management in glaucoma, or simulating cataract surgery planning and prognosis.

[0050] This embodiment also provides a terminal, such as Figure 7 As shown, it includes: a memory 501, a processor 502, and a computing program stored in the memory 501 and executable on the processor 502, wherein the processor executes the computer program to implement the steps of the method.

[0051] In one implementation, the terminal further includes: A communication interface for communication between memory 501 and processor 502.

[0052] The memory 501 is used to store computer programs that can run on the processor 502.

[0053] Memory 501 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0054] If the memory 501, processor 502, and communication interface are implemented independently, they can be interconnected via a bus to communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, only one line is used in the diagram, but this does not imply that there is only one bus or one type of bus.

[0055] In practical implementation, if the memory 501, processor 502, and communication interface are integrated on a single chip, then the memory 501, processor 502, and communication interface can communicate with each other through an internal interface. The processor 502 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0056] This embodiment also provides a computer-readable storage medium storing a computer program that can be executed to implement the steps of the method for constructing the digital twin eyeball model.

[0057] In summary, this invention provides a method, system, device, and storage medium for constructing a digital twin eye model. Through a unique technical path of "acquiring multi-source data → reconstructing the anterior and posterior segments separately → constructing a virtual intermediate segment → fusing them into a whole-eye model," this invention, for the first time, utilizes clinically convenient SS-OCT technology to digitally reconstruct the complete three-dimensional geometric morphology of the eyeball, including the anterior segment (cornea and lens), posterior segment (retina and choroid), and intermediate sclera. This provides an unprecedented and complete data foundation for comprehensive ocular morphological analysis and biomechanical research. This invention combines data from the anterior and posterior segments to obtain their structures. More importantly, by using axial length data to geometrically constrain and fit the anterior and posterior segment structures, the constructed whole-eye model possesses anatomical rationality in key dimensions similar to that based on individual 3D-MRI data. Therefore, without relying on 3D-MRI, it provides a high-precision morphological assessment tool comparable to 3D-MRI, greatly expanding the application scenarios of precise ophthalmic examinations. This invention actively correlates and maps at least one biomechanical or biophysical property measurement (intraocular pressure, biomechanical properties, etc.) to the corresponding region of a three-dimensional geometric model. This results in a final output that is no longer an "empty shell" model, but a "digital twin" integrating individualized anatomical structures and physiological / mechanical parameters. This twin can serve as an in vitro experimental platform to simulate the morphological and functional responses of the eyeball under conditions such as myopia progression, intraocular pressure changes, and surgical interventions, providing a powerful new tool for disease mechanism research, treatment plan simulation, and personalized risk assessment. Through quantitative comparison of the three-dimensional morphology with a three-dimensional reference model of the same target eyeball reconstructed based on magnetic resonance imaging data, the whole-eye three-dimensional geometric model obtained by this invention shows a high degree of matching with the three-dimensional reference model reconstructed based on 3D-MRI imaging data within the 22-32mm axial length range.

[0058] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the claims.

[0059] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0060] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus or device (such as a computer-based system, a processor-included system or other system that can read and execute instructions from and from an instruction execution system, apparatus or device).

[0061] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

Claims

1. A method for constructing a digital twin eyeball model, characterized in that, Includes the following steps: S1. Obtain multi-source input data of the target eyeball, the multi-source input data including: axial length data, at least one measurement value of the biomechanical or biophysical properties of the eyeball, and three-dimensional scan datasets of the anterior segment and posterior segment; S2. Perform three-dimensional reconstruction on the anterior segment three-dimensional scan dataset and the posterior segment three-dimensional scan dataset respectively to obtain the three-dimensional structure of the anterior segment and the three-dimensional structure of the posterior segment. S3. Based on the axial length data, and constrained by the rear geometric contour of the anterior segment three-dimensional structure and the front geometric contour of the posterior segment three-dimensional structure, construct a virtual three-dimensional structure of the middle segment of the target eyeball; S4. Integrate the three-dimensional structure of the anterior segment, the virtual three-dimensional structure of the middle segment, and the three-dimensional structure of the posterior segment to form a full-eye three-dimensional geometric model of the target eyeball; S5. Associate and map the at least one biomechanical or biophysical characteristic measurement value to the corresponding region in the three-dimensional geometric model of the whole eye, thereby constructing the digital twin eyeball model.

2. The method for constructing a digital twin eyeball model according to claim 1, characterized in that, In step S1, the anterior segment three-dimensional scan dataset and the posterior segment three-dimensional scan dataset are acquired by optical coherence tomography (OCT) with an ultra-wide field scanning source and a scanning range of 120-150°.

3. The method for constructing a digital twin eyeball model according to claim 1, characterized in that, The axial length ranges from 22 to 32 mm.

4. The method for constructing a digital twin eyeball model according to claim 1, characterized in that, In step S2, the three-dimensional reconstruction of the anterior segment three-dimensional scan dataset includes: Based on the aforementioned anterior segment 3D scan dataset, anterior segment images are obtained; The anterior segment image is refractively corrected to obtain the corrected image; Based on the corrected image, the corneal and lens surfaces are segmented and extracted; The corneal and / or lens surfaces are fitted with a surface to reconstruct their optical surfaces.

5. The method for constructing a digital twin eyeball model according to claim 1, characterized in that, In step S3, a virtual three-dimensional structure of the middle segment of the target eyeball is constructed by interpolation fitting.

6. The method for constructing a digital twin eyeball model according to claim 1, characterized in that, In step S1, the at least one biomechanical or biophysical characteristic measurement of the eyeball includes any one or more of the following: intraocular pressure, corneal mechanical parameters, scleral mechanical parameters, and refractive parameters.

7. A digital twin eyeball model system, characterized in that, include: The data acquisition and input module is used to acquire multi-source input data of the target eyeball. The multi-source input data includes: axial length data, at least one biomechanical or biophysical characteristic measurement value, and three-dimensional scan datasets of the anterior segment and posterior segment. The 3D modeling engine module, which is communicatively connected to the data acquisition and input module, performs 3D reconstruction on the anterior segment 3D scan dataset and the posterior segment 3D scan dataset to obtain the anterior segment 3D structure and the posterior segment 3D structure, respectively. Based on the axial length data and constrained by the rear geometric contour of the anterior segment 3D structure and the front geometric contour of the posterior segment 3D structure, a virtual 3D structure of the middle segment of the eyeball is constructed. The anterior segment 3D structure, the virtual 3D structure of the middle segment, and the posterior segment 3D structure are fused to generate a full-eye 3D geometric model of the target eyeball. The digital twin construction module is communicatively connected to the 3D modeling engine module and the data acquisition and input module, and associates and maps the at least one biomechanical or biophysical characteristic measurement value to the corresponding anatomical structure region in the whole eye 3D geometric model, thereby constructing the digital twin eyeball model that can be used for simulation analysis.

8. A digital twin eyeball model system according to claim 7, characterized in that, The digital twin eye model system also includes: The simulation analysis module is communicatively connected to the digital twin construction module and is used to perform simulation analysis of the physiological or pathological state of the eyeball based on the digital twin eyeball model.

9. A terminal, characterized in that, include: A memory, a processor, and a computing program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps of the method according to any one of claims 1-6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that can be executed to implement the steps of the method for constructing a digital twin eye model according to any one of claims 1-6.

Images