Eye fundus data determination method

Abstract

The invention relates to the technical field of eye detection, in particular to a fundus data determination method. The method comprises the following steps: performing layering processing on first scanning data containing an optic nerve head, determining an identifiable layering and an inner boundary membrane, determining an optic disc edge of the optic nerve head based on the identifiable layering, and determining the optic disc edge of the optic nerve head based on a surface area formed by a closed curve extending from the optic disc edge to the inner boundary membrane. And determining a closed curve when the surface area meets a preset condition as a target curve, determining an optic cup edge of the optic nerve head based on the target curve, and determining fundus data according to the optic disk edge and the optic cup edge. By adopting the method, the accuracy can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of eye detection, and in particular to a method for determining fundus data. Background Art

[0002] The optic nerve head (ONH), also known as the optic disc, is the area on the retina where the optic nerve fibers and retinal blood vessels converge and exit the eyeball. A physiological depression, called the optic cup, typically forms in the center of the optic disc.

[0003] Obtaining ONH-related fundus data (such as optic disc area, optic cup area, cup-to-disc ratio, and optic cup volume) is crucial for the diagnosis and treatment of optic nerve diseases. However, obtaining this fundus data relies on the accurate positioning of the optic disc margin and optic cup margin.

[0004] However, the accuracy of the fundus data currently obtained is not high. Summary of the Invention

[0005] Based on this, it is necessary to provide a method for determining fundus data that can improve accuracy in response to the above technical problems.

[0006] In a first aspect, the present application provides a method for determining fundus data, comprising:

[0007] determining an internal limiting membrane and an optic disc margin based on first scan data including the optic nerve head;

[0008] Based on the surface area formed by the closed curve extending from the edge of the optic disc to the internal limiting membrane, determining the closed curve when the surface area meets a preset condition as the target curve;

[0009] Determine the cup edge of the optic nerve head based on the target curve;

[0010] Fundus data were determined based on the optic disc margin and optic cup margin.

[0011] In a second aspect, the present application further provides a fundus data determination device, comprising:

[0012] The first determining module is configured to determine an internal limiting membrane and an optic disc edge based on first scanning data containing the optic nerve head.

[0013] The second determining module is configured to determine, based on a surface area formed by a closed curve extending from the optic disc edge to the internal limiting membrane, a closed curve whose surface area satisfies a preset condition as a target curve.

[0014] The third determination module is configured to determine the optic cup edge of the optic nerve head based on the target curve.

[0015] The fourth determining module is used to determine fundus data according to the optic disc edge and the optic cup edge.

[0016] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned fundus data determination method when executing the computer program.

[0017] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above-mentioned fundus data determination method when executed by a processor.

[0018] In a fifth aspect, the present application also provides a computer program product, comprising a computer program, which implements the steps of the above-mentioned fundus data determination method when executed by a processor.

[0019] The above-described fundus data determination method can determine the internal limiting membrane and optic disc margin based on the first scan data including the optic nerve head. Based on the surface area formed by a closed curve extending from the optic disc margin to the internal limiting membrane, the closed curve whose surface area satisfies a preset condition is determined as a target curve. Therefore, based on the surface area, the surface formed by the closed curve extending from the optic disc margin to the internal limiting membrane can be calculated as a whole. This not only reflects the total volume of all optic nerves, but also results in a highly continuous target curve. Furthermore, based on the target curve, the optic cup margin of the optic nerve head can be determined relatively accurately, thereby determining fundus data with high accuracy based on the optic disc margin and optic cup margin. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments of the present application or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying any creative work.

[0021] Figure 1 Schematic diagram of the optic nerve head;

[0022] Figure 2 Schematic diagram of OCT imaging technology;

[0023] Figure 3 This is a schematic diagram of the structure of a healthy human eye;

[0024] Figure 4This is an OCT image of a myopic human eye;

[0025] Figure 5 A schematic diagram of a fundus image captured by a fundus camera;

[0026] Figure 6 A diagram illustrating an application environment of a method for determining fundus data in one embodiment;

[0027] Figure 7 FIG1 is a flow chart of a method for determining fundus data in one embodiment;

[0028] Figure 8 is a schematic diagram of a stratification result in one embodiment;

[0029] Figure 9 is a schematic diagram of the optic disc edge and the optic cup edge in one embodiment;

[0030] Figure 10 is a schematic diagram of a triangle segmentation process in one embodiment;

[0031] Figure 11 A schematic diagram of a process for determining a target curve in one embodiment;

[0032] Figure 12 is a schematic diagram of a polar coordinate system in one embodiment;

[0033] Figure 13 FIG1 is a schematic diagram of another process for determining a target curve in one embodiment;

[0034] Figure 14 FIG1 is a schematic diagram of a process for determining the edge of the optic disc in one embodiment;

[0035] Figure 15 is a schematic diagram of a two-dimensional projection image and a first curve in one embodiment;

[0036] Figure 16 is a schematic diagram showing fundus data in one embodiment;

[0037] Figure 17 A schematic diagram of a process of a method for determining fundus data in one embodiment;

[0038] Figure 18 A comparison chart of the effects of the related technology and the present application;

[0039] Figure 19 FIG. 4 is a structural block diagram of a fundus data determination device in one embodiment. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0041] Figure 1 Schematic diagram of the optic nerve head. Figure 1 A fundus image collected by a fundus camera is shown. The position of the optic nerve head is shown in FIG. Figure 1 As shown by the white circle in .

[0042] Optical coherence tomography (OCT) is a technology that uses low-coherence light sources and the principle of interference to perform high-resolution cross-sectional imaging of biological tissue. It is particularly widely used in ophthalmology, enabling non-invasive and rapid scanning and measurement of various physiological structures in the anterior and posterior segments of the eye. In posterior segment OCT applications, OCT images clearly reveal the details of each retinal layer.

[0043] Figure 2 This is a schematic diagram of OCT imaging technology. Figure 2 (a) shows the OCT volume data collected near the ONH. Figure 2 Figure (b) shows a frame of tomographic image (B-scan) passing through the center of ONH. Figure 2 As shown in Figure 2, the optic cup, a concave center of the ONH, can be clearly seen using OCT technology. Compared to measuring ONH using fundus camera images, the high-resolution data obtained by OCT technology provides a more accurate and detailed data foundation.

[0044] The following describes the process of determining the position of the video disc in the related art.

[0045] Figure 3 This is a schematic diagram of the structure of a healthy human eye. Figure 3 The diagram shows the relative relationship between the optic cup margin and the retinal nerve fiber layer in a healthy human eye. Figure 3 As shown in Figure 1, the optic disc edge can be referenced by the opening of Bruch's membrane (BM) near the ONH (referred to as the Bruch's membrane opening (BMO)). In other words, in a healthy eye, the intersection of the BM and the optic nerve fiber layer is a good proxy for the optic disc location. Therefore, some technologies determine the optic disc edge by performing layered calculations on the retina to determine the BMO location, and then directly define the BMO as the optic disc location.

[0046] However, it is not appropriate to directly define the position of the optic disc as the position of the BMO in all cases. Figure 4 This is an OCT image of a myopic human eye. Figure 4 401 indicates the end point of Bruce's membrane, that is, the location of BMO, and 402 indicates the junction of sclera and optic nerve. Figure 4 As shown in the figure, some patients with high myopia experience parapapillary atrophy, which means that the retina, Brucea's membrane, and even the underlying choroid shrink away from the point of contact with the optic nerve, exposing the underlying sclera. Furthermore, parapapillary atrophy does not directly correspond to the health of the optic nerve. Therefore, in this case, using BMO to define the optic disc edge cannot accurately represent the size of the optic disc. Instead, using the termination point of the sclera is more accurate.

[0047] In addition to the above methods, some related technologies use traditional methods or deep learning methods to segment the fundus images captured by the fundus camera to determine the optic disc position. However, this method is not very accurate.

[0048] Figure 5 A schematic diagram of a fundus image captured by a fundus camera, Figure 5 The 501 in the code indicates a video disc. Figure 5 502 in the figure indicates Peripapillary Atrophy (PPA), such as Figure 5 As shown in the figure, for human eyes with paraoptic disc atrophy, since the contrast between the atrophic retinal edge and the underlying tissue is usually significantly higher than the contrast between the scleral edge and ONH, even if the labeled data used for training the model is correctly labeled, the model can still easily mistake the retinal edge for the optic disc edge when making predictions, thereby overestimating the optic disc size.

[0049] The following describes the process of determining the optic cup position in related technologies. The optic cup margin is a closed curve located on the upper surface of the inner limiting membrane (ILM). Because the ILM is very thin, the distinction between the upper surface of the ILM and the ILM itself will not be made here; both are collectively referred to as the ILM. Related technologies typically determine the optic cup position using the following method:

[0050] (1) ONH data are collected in polar coordinates with the center of the BMO as the origin. On each B-scan image, a line segment is formed connecting the BMO position and any point on the ILM. The position of the corresponding point on the ILM with the smallest line segment length is used as the position of the optic cup edge. However, this method does not truly reflect the concept of the total volume of the optic nerve and has poor accuracy.

[0051] (2) The plane where the BMO is located is translated upward along the axial direction by a fixed distance to obtain the shifted plane. The closed curve formed by the intersection of the shifted plane and the ILM surface is the optic cup edge. If the two do not intersect, it can be considered that the optic cup edge does not exist. However, the optic cup edge determined in this way is a subjective definition and has no direct relationship with the total amount of the optic nerve.

[0052] (3) Based on the fundus image or OCT projection image, traditional algorithms or deep learning methods are used to segment the position of the optic cup edge. However, since the optic cup is not the interface between two different physiological structures, there is no clear interface where the optical image brightness changes. Therefore, the results obtained by this method may not be physiologically meaningful. In addition, since there is no clear interface on the fundus image or projection image, the repeatability of the results is also relatively poor.

[0053] Therefore, it is necessary to provide a method for determining fundus data with high accuracy. The following will introduce this method in detail.

[0054] Figure 6 FIG is an application environment diagram of a method for determining fundus data in an embodiment. In an exemplary embodiment, a computer device is provided. The method for determining fundus data provided in the embodiment of the present application can be applied to Figure 6 The computer device shown in FIG. The computer device may be a server, such as Figure 6 As shown, the computer device includes a processor, memory, an input / output (I / O) interface, and a communication interface. The processor, memory, and I / O interface are connected via a system bus, and the communication interface is connected to the system bus via the I / O interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store relevant data. The I / O interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal via a network connection. When executed by the processor, the computer program implements a method for determining fundus data.

[0055] Those skilled in the art will understand that Figure 6The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0056] This embodiment illustrates this method using a server as an example. It is understood that this method can also be applied to a terminal, or to a system comprising a terminal and a server, and implemented through interaction between the terminal and the server. The terminal may be, but is not limited to, various personal computers, laptops, smartphones, and tablet computers. The server may be implemented as a standalone server or a server cluster consisting of multiple servers.

[0057] Figure 7 FIG. 1 is a flow chart of a method for determining fundus data in one embodiment. In an exemplary embodiment, as shown in FIG. Figure 7 As shown, a method for determining fundus data is provided, which is applied to Figure 6 The computer device in the example is used for description, including the following S701 to S704.

[0058] S701 , determining the internal limiting membrane and the edge of the optic disc based on first scanning data including the optic nerve head.

[0059] In this embodiment, the first scan data may include, but is not limited to, a set of B-scan images or volume data. For example, the first scan data may include a set of B-scan images acquired using a raster scan method within a rectangular area perpendicular to the ocular axis. The first scan data may also include a set of B-scan images acquired using a star scan method centered on the center of the ONH. For another example, the first scan data may also include volume data reconstructed from a set of B-scan images.

[0060] Furthermore, the computer device can determine the internal limiting membrane and the optic disc margin based on the first scan data including the optic nerve head. It will be appreciated that the internal limiting membrane is used to represent the location of the internal limiting membrane in three-dimensional space. The optic disc margin is used to represent the location of the optic disc margin in three-dimensional space. For example, the computer device can input the first scan data into a trained analysis model to determine the internal limiting membrane and the optic disc margin using the analysis model.

[0061] In an exemplary embodiment, optionally, the above-mentioned S701 includes: determining the internal limiting membrane and the identifiable layer closest to the optic nerve based on the first scan data including the optic nerve head; and determining the optic disc edge based on a closed curve formed by the identifiable layer.

[0062] Optionally, the computer device may perform layered processing on the first scan data to determine the recognizable layers and internal limiting membrane. For example, the computer device may perform layered processing on the first scan data to determine the recognizable layers and internal limiting membrane using a preset segmentation algorithm. The computer device may also perform layered processing on the first scan data to determine the recognizable layers and internal limiting membrane using a deep learning model. The computer device may also perform layered processing on the first scan data to determine the recognizable layers and internal limiting membrane in response to user input. This embodiment is not limited to this.

[0063] It should be noted that the Bruce membrane physiologically breaks near the optic nerve head, and in this application, the identifiable layer closest to the optic nerve can be understood as the continuous layer where the Bruce membrane is located. For example, based on the physiological structure of the eye in the first scan data, the computer device can identify the continuous layer corresponding to the Bruce membrane as the identifiable layer closest to the optic nerve.

[0064] Optionally, the identifiable layer closest to the optic nerve includes at least one of the Bruce's membrane, the choroid, and the sclera. For example, in the first scan data, where the Bruce's membrane is present, the identifiable layer is located there; where the Bruce's membrane is absent but the choroid is present, the identifiable layer is located on the superior surface of the choroid; where the choroid is absent but the sclera is present, the identifiable layer is located on the superior surface of the sclera; and where the sclera is absent (i.e., where the optic nerve head passes through the sclera), the identifiable layer is located along the line connecting the aforementioned results. This ensures that the identified identifiable layers are continuous, facilitating subsequent accurate determination of the optic disc margin and optic cup margin.

[0065] Figure 8 FIG. 1 is a schematic diagram of a layered result in one embodiment, taking the first scanning data as a set of B-scan images collected at different positions along the axial direction as an example. Figure 8 As shown, a curve 801 located above the B-scan image represents the internal limiting membrane, and a curve 802 located below the B-scan image represents the identifiable layer.

[0066] Furthermore, the computer device can determine the optic disc edge based on the closed curve formed by the recognizable layers. In this embodiment, for example, the computer device can determine the closed curve formed by the recognizable layers based on the image corresponding to the recognizable layers and a trained machine learning model, and use the closed curve formed by the recognizable layers as the optic disc edge of the optic nerve head. The computer device can also perform segmentation based on the grayscale features or structural features of the image corresponding to the recognizable layers to determine the closed curve formed by the recognizable layers, and use the closed curve formed by the recognizable layers as the optic disc edge of the optic nerve head. In some embodiments, the computer device can also perform post-processing such as smoothing on the closed curve formed by the recognizable layers to obtain the optic disc edge.

[0067] In this way, by determining the internal limiting membrane and the identifiable layer closest to the optic nerve based on the first scanning data including the optic nerve head, the optic disc edge can be determined more accurately based on the closed curve formed by the identifiable layer.

[0068] S702 : Based on the surface area formed by the closed curve extending from the optic disc edge to the internal limiting membrane, determine the closed curve when the surface area meets a preset condition as a target curve.

[0069] In this embodiment, the computer device can determine the surface area formed by a closed curve extending from the optic disc edge to the internal limiting membrane. In other words, the optic disc edge and any closed curve on the internal limiting membrane can form a geometric body, and the lateral area of this geometric body is the surface area formed by the closed curve extending from the optic disc edge to the closed curve. Alternatively, the computer device can traverse the internal limiting membrane to obtain multiple closed curves on the internal limiting membrane and calculate the surface area formed by the closed curve extending from the optic disc edge to the internal limiting membrane for each closed curve.

[0070] Furthermore, the computer device can determine that the closed curve when the surface area meets the preset condition is the target curve. In other words, the target curve is a closed curve corresponding to when the surface area meets the preset condition.

[0071] The preset condition is used to constrain the surface area formed by a closed curve extending from the optic disc edge to the internal limiting membrane. Optionally, the preset condition may include a surface area ranked within a first preset range after being sorted by size, a minimum surface area, or a surface area less than a preset area threshold, all of which are not limited in this embodiment. Both the first preset range and the preset area threshold can be set according to actual needs.

[0072] S703: Determine the optic cup edge of the optic nerve head based on the target curve.

[0073] In this embodiment, after the computer device determines the target curve, it can determine the optic cup margin of the optic nerve head based on the target curve. Alternatively, the computer device can directly use the target curve as the optic cup margin of the optic nerve head, or can perform post-processing on the target curve to obtain the optic cup margin of the optic nerve head. The post-processing may include, but is not limited to, at least one of smoothing, enhancement, noise reduction, and filtering.

[0074] Figure 9 is a schematic diagram of the edge of the optic disc and the edge of the optic cup in one embodiment, and Figure 8 As an example of the B-scan image shown in FIG. 1 , the optic disc edge and the optic cup edge on the B-scan image can be seen as follows: Figure 9 shown.

[0075] S704: Determine fundus data according to the optic disc edge and the optic cup edge.

[0076] In this embodiment, the fundus data is data related to the optic nerve head. Optionally, the fundus data includes, but is not limited to, at least one of the following: disc area, cup area, cup-disc ratio (C / D ratio), neuroretinal rim area (RIM area), cup volume, and average retinal nerve fiber layer thickness. The cup-disc ratio may include, but is not limited to, at least one of the following: average cup-to-disc ratio (Avg C / D ratio), vertical cup-to-disc ratio (Vertical C / D ratio), and horizontal cup-to-disc ratio.

[0077] In the above-described fundus data determination method, the internal limiting membrane and optic disc margin are determined based on the first scan data including the optic nerve head. Based on the surface area formed by a closed curve extending from the optic disc margin to the internal limiting membrane, the closed curve whose surface area satisfies a preset condition is determined as a target curve. Therefore, based on the surface area, the surface formed by the closed curve extending from the optic disc margin to the internal limiting membrane can be calculated as a whole. This not only reflects the total volume of all optic nerves, but also results in a highly continuous target curve. Furthermore, based on the target curve, the optic cup margin of the optic nerve head can be determined relatively accurately, thereby determining fundus data with high accuracy based on the optic disc margin and optic cup margin.

[0078] In an exemplary embodiment, the fundus data determination method further includes: performing triangulation processing on a surface formed by a closed curve extending from the optic disc edge to the internal limiting membrane to obtain a plurality of triangles; and determining the surface area based on the sum of the areas of the triangles.

[0079] In this embodiment, the computer device may optionally use a preset triangulation algorithm to triangulate the surface formed by the closed curve extending from the optic disc edge to the internal limiting membrane to obtain a plurality of triangles. The preset triangulation algorithm includes, but is not limited to, a frontier advancing algorithm or a Delaunay triangulation algorithm.

[0080] Figure 10FIG. 1 is a schematic diagram of a triangle segmentation process in one embodiment. Figure 10 As shown, taking the optic disc edge 1001 and a closed curve 1002 on the internal limiting membrane as an example, after performing triangulation processing on the surface extending from the optic disc edge 1001 to the closed curve 1002, the surface can be decomposed into multiple adjacent and non-overlapping triangles. Figure 10 , the computer device can segment the surface to obtain a triangle with vertices Q1, P1, P2, a triangle with vertices Q1, Q2, P2, a triangle with vertices Q2, P2, P3, a triangle with vertices Q2, Q3, P3... and so on, where P1, P2, P3, etc. are points on the optic disc edge 1001, and Q1, Q2, Q3, etc. are points on the closed curve 1002.

[0081] The computer device can then determine the surface area based on the sum of the areas of the triangles. Alternatively, the computer device can use the sum of the areas of the triangles as the surface area. In some embodiments, the computer device can also perform corrections or other processing on the sum of the areas of the triangles to obtain the surface area, which is not a limitation in this embodiment.

[0082] Alternatively, the computer device may determine the area of a triangle based on the following formula (1). In formula (1), taking the vertices of any triangle in three-dimensional space as A, B, and C as an example, 、 、 Represents a vector The components in the x, y and z directions respectively, 、 、 Represents a vector The components in the x-, y-, and z-directions, respectively. The x-, y-, and z-directions are perpendicular to each other in three-dimensional space.

[0083] (1);

[0084] In the above embodiment, since the surface formed by the closed curve extending from the optic disc edge to the internal limiting membrane can be triangulated to obtain a plurality of triangles, the surface formed by the closed curve extending from the optic disc edge to the internal limiting membrane can be decomposed into a plurality of triangles. In this way, the corresponding surface area can be determined efficiently and accurately based on the sum of the areas of the triangles.

[0085] Figure 11 FIG. 1 is a flow chart of determining a target curve in an embodiment. In an exemplary embodiment, as shown in FIG. Figure 11 As shown, S702 includes S1101 to S1103.

[0086] S1101, converting the pixel points of the internal limiting membrane into candidate points in a polar coordinate system.

[0087] In this embodiment, a polar coordinate system can be established based on actual needs. Alternatively, the polar coordinate system can be determined based on any location within the optic disc edge. Further, the computer device can determine the origin of the polar coordinate system based on the centroid of the optic disc edge, thereby establishing the corresponding polar coordinate system.

[0088] In order to more clearly introduce the process of determining the target curve in this application, Figure 12 Provide explanation. Figure 12 A schematic diagram of a polar coordinate system in one embodiment is shown. Taking the center of mass of the optic disc edge as the origin of the polar coordinate system as an example, the following can be established: Figure 12 The polar coordinate system shown.

[0089] Then, the computer device converts the pixel points of the inner limiting membrane into candidate points in the polar coordinate system. Optionally, the computer device can sample all the pixel points on the inner limiting membrane according to a first preset sampling rate, and convert the sampled pixel points into the polar coordinate system to obtain the corresponding candidate points. For example, the computer device performs a polar coordinate transformation on the coordinates of each pixel point on the inner limiting membrane in the x-direction and the y-direction, and the z-coordinate remains unchanged, so as to obtain a curved surface where the inner limiting membrane is located in a polar coordinate system, thereby determining the candidate points in the polar coordinate system. Among them, the z-direction is the axial direction. Please continue to refer to Figure 12 In Figure (a), the candidate points are as follows Figure 12 As shown by the black dots in Figure (a), for example, candidate point C1, candidate point C2, candidate point C3, etc.

[0090] S1102, converting the pixel points at the edge of the optic disc into polar coordinate points in a polar coordinate system; each polar coordinate point corresponds to a polar angle.

[0091] In this embodiment, the computer device may optionally sample all pixels on the edge of the optic disc at a second preset sampling rate and convert the sampled pixels into a polar coordinate system to obtain corresponding polar coordinate points. For example, the computer device performs polar coordinate transformation on the x- and y-direction coordinates of each pixel on the curve corresponding to the optic disc edge, while keeping the z-coordinate unchanged, thereby obtaining a optic disc edge curve in a polar coordinate system, thereby determining the polar coordinate points in the polar coordinate system. Please continue to refer to Figure 12 As shown in FIG. 1( b ), the computer device may convert the pixel points on the edge of the video disc 1201 into polar coordinate points P1 , P2 , P3 , and so on in a polar coordinate system.

[0092] It is understandable that each polar coordinate point corresponds to a polar angle. For example, the polar angle Corresponding to the polar coordinate point P1, polar angle Corresponding to the polar coordinate point P2, polar angle Corresponding to the polar coordinate point P3, and so on.

[0093] S1103 , according to the areas of the triangles corresponding to the polar coordinate points and the candidate points, based on the shortest path method, a closed curve is determined from each candidate point when the surface area meets a preset condition as the target curve.

[0094] In this embodiment, please refer to Figure 10 and Figure 12 After the surface formed by the closed curve extending from the edge of the optic disc to the internal limiting membrane is triangulated to obtain multiple triangles, each polar angle intersects the edge of the optic disc at a polar coordinate point, such as the adjacent polar angles and They intersect with the edge of the optic disc at polar coordinate points P1 and P2 respectively. Similarly, each polar angle intersects with any closed curve at a candidate point, such as the adjacent polar angles and They intersect the edge of the optic cup at candidate points Q1 and Q2, respectively. Therefore, the computer device can determine the area of a triangle formed by the polar coordinate points and the corresponding candidate points. For example, the computer device can determine the area of a triangle whose points are Q1, P1, and P2, and the area of a triangle whose points are Q1, P1, and Q2, and so on.

[0095] Furthermore, in order to determine the closed curve when the surface area satisfies a preset condition, that is, the closed curve when the sum of the areas of multiple triangles satisfies the preset condition, the computer device can transform the above problem into a graph cut problem in a polar coordinate system. Based on the areas of the triangles corresponding to the polar coordinate points and the candidate points, the computer device can determine, from each candidate point, the closed curve when the surface area satisfies the preset condition based on the shortest path method as the target curve. For example, the computer device can treat each candidate point in the polar coordinate system as all nodes in a graph, and based on the shortest path method, determine a shortest path from left to right from all nodes in the graph, where the area of the triangle corresponding to the shortest path satisfies the first preset condition, thereby determining the target curve based on the shortest path.

[0096] In the above embodiment, since the pixel points of the internal limiting membrane are converted into candidate points in a polar coordinate system, and the pixel points of the optic disc edge are converted into polar coordinate points in a polar coordinate system, and each polar coordinate point corresponds to a polar angle, based on the areas of the triangles corresponding to the polar coordinate points and the candidate points, a closed curve whose surface area satisfies a preset condition can be determined from each candidate point based on the shortest path method as the target curve. This converts the problem of determining the surface area that satisfies the preset condition into the problem of determining the shortest path in a polar coordinate system, thereby improving the efficiency of determining the target curve.

[0097] Figure 13 FIG. 1 is a flow chart of another method for determining a target curve in an embodiment. In an exemplary embodiment, as shown in FIG. Figure 13 As shown, S1103 includes S1301 to S1304.

[0098] S1301: Determine a first loss value corresponding to any candidate point based on the area of the first triangle.

[0099] In this embodiment, the computer device may determine a first loss value corresponding to each candidate point. The first loss value may be determined based on the area of the first triangle. The first triangle is determined based on the candidate point, the first polar coordinate point, and the second polar coordinate point.

[0100] Optionally, the computer device may determine the area of the first triangle based on the candidate point, the first polar coordinate point, and the second polar coordinate point, and use the area of the first triangle as the first loss value for the corresponding candidate point. In some embodiments, the computer device may also perform weighted processing on the area of the first triangle to obtain the corresponding first loss value.

[0101] The first polar coordinate point is a polar coordinate point corresponding to the same polar angle as the candidate point; the first polar coordinate point and the second polar coordinate point are polar coordinate points corresponding to two adjacent polar angles along the polar axis direction.

[0102] For example, taking candidate point C1 as an example, candidate point C1 and polar coordinate point P1 correspond to the same polar angle , and the polar coordinate point P1 and the polar coordinate point P2 correspond to adjacent polar angles along the polar axis direction and Therefore, in this case, the first polar coordinate point is polar coordinate point P1, and the second polar coordinate point is polar coordinate point P2. Then, the computer device determines the area of the first triangle whose vertices are candidate point C1, polar coordinate point P1, and polar coordinate point P2. , and the area As the first loss value of candidate point C1, the other first loss values are similar and will not be described here.

[0103] S1302: Determine a second loss value corresponding to a line between the first candidate point and the second candidate point according to the area of the second triangle.

[0104] In this embodiment, the first candidate point and the second candidate point are two different candidate points, and the first candidate point and the second candidate point correspond to two adjacent polar angles along the polar axis. Figure 12 In FIG. 5( a ), when the first candidate points are C1 to C5 , the second candidate points may be C6 to C10 .

[0105] Furthermore, the computer device can determine a second loss value corresponding to a line connecting the first candidate point and the second candidate point. The second loss value can be determined based on the area of the second triangle. The second triangle is determined based on the first candidate point, the second candidate point, and the third polar coordinate point.

[0106] Similarly, the computer device can determine the area of the second triangle based on the first candidate point, the second candidate point, and the third polar coordinate point, and use the area of the second triangle as the second loss value corresponding to the line connecting the first candidate point and the second candidate point. In some embodiments, the computer device can also perform weighted processing on the area of the second triangle to obtain the corresponding second loss value.

[0107] The third polar coordinate point is a polar coordinate point corresponding to the same polar angle as the second candidate point. In some embodiments, the third polar coordinate point is also the second polar coordinate point mentioned above. For example, taking the first candidate point as candidate point C1 and the second candidate point as candidate point C6 as an example, candidate point C1 and candidate point C6 correspond to two adjacent polar angles along the polar axis direction, respectively. and , and the polar coordinate point P2 and the candidate point C6 both correspond to the polar angle Therefore, in this case, the third polar coordinate point is polar coordinate point P2, and the computer device determines the area of the second triangle whose vertices are candidate point C1, candidate point C6, and polar coordinate point P2 in sequence. , and the area As the second loss value of the line between candidate point C1 and candidate point C6, the other second loss values are similar and will not be described in detail here.

[0108] S1303: Determine the shortest path according to the first loss value and the second loss value.

[0109] In this embodiment, according to S1301 and S1302, the computer device can determine the first loss value corresponding to any candidate point and the second loss value of the line connecting any first candidate point and any second candidate point. Furthermore, the computer device can determine the shortest path based on the first loss value and the second loss value.

[0110] Optionally, the computer device may select a candidate point from the candidate points corresponding to each polar coordinate point as a candidate point, and determine a candidate path based on all candidate points and the line connecting two adjacent candidate points. Similarly, the computer device may determine multiple candidate paths. The candidate point corresponding to a polar coordinate point refers to a candidate point that corresponds to the same polar angle as the polar coordinate point.

[0111] For example, please refer to Figure 12 Figure (a) and Figure 12In Figure (b), for polar coordinate point P1, the candidate points corresponding to polar coordinate point P1 include candidate points C1 to candidate point C5. For polar coordinate point P2, the candidate points corresponding to polar coordinate point P2 include candidate points C6 to candidate point C10. The same is true for other polar coordinate points, which will not be repeated here. Assume that candidate point C1 is selected as a candidate point from candidate point C1 to candidate point C5, and candidate point C7 is selected as a candidate point from candidate point C6 to candidate point C10. And so on. Each polar coordinate point can determine the corresponding candidate point. Please continue to refer to Figure 12 As shown in FIG. 1(c), the lines between these candidate points constitute a candidate path 1202.

[0112] Furthermore, the computer device can determine the sum of the first loss value and the second loss value corresponding to each candidate path. The sum of the first loss value and the second loss value corresponding to each candidate path is equal to the sum of the first loss values of all candidate points in the candidate path and the second loss values of the line connecting two adjacent candidate points. It will be understood that the sum of the first loss value and the second loss value corresponding to each candidate path is equal to the surface area formed by extending from the edge of the optic disc to the candidate path.

[0113] Furthermore, the computer device may traverse all candidate paths and, based on the sum of the first loss value and the second loss value corresponding to the candidate paths, select the candidate path with the minimum sum of the first loss value and the second loss value as the shortest path. In this way, the shortest path includes the target point corresponding to each polar coordinate point in the candidate points and the lines connecting the target points, and the sum of the first loss value and the second loss value corresponding to the shortest path is the smallest.

[0114] S1304: Determine a target curve based on the shortest path.

[0115] In this embodiment, the computer device can convert the target point on the shortest path into a Cartesian coordinate system to obtain a corresponding target curve. The computer device can also perform post-processing such as smoothing on the target point on the shortest path and then convert it into a Cartesian coordinate system to obtain a corresponding target curve. This embodiment is not limited to this.

[0116] In the above embodiment, since the first triangle is determined based on the candidate point, the first polar coordinate point, and the second polar coordinate point, the first polar coordinate point and the candidate point correspond to the same polar angle, and the first polar coordinate point and the second polar coordinate point correspond to two adjacent polar angles along the polar axis. Therefore, based on the area of the first triangle, the first loss value corresponding to any candidate point can be determined. Since the second triangle is determined based on the first candidate point, the second candidate point, and the third polar coordinate point, the first candidate point and the second candidate point are two different candidate points, the first candidate point and the second candidate point correspond to two adjacent polar angles along the polar axis, and the third polar coordinate point corresponds to the same polar angle as the second candidate point. Therefore, based on the area of the second triangle, the second loss value corresponding to the line connecting the first and second candidate points can be determined. Furthermore, since the shortest path includes the target point corresponding to each polar coordinate point in the candidate points and the line connecting the target points, the sum of the first and second loss values corresponding to the shortest path is minimized. Therefore, based on the first and second loss values, the shortest path can be efficiently determined, thereby determining a target curve whose surface area meets a preset condition based on the shortest path.

[0117] In an exemplary embodiment, optionally, the radial distance between the first candidate point and the second candidate point satisfies a second preset condition.

[0118] In this embodiment, in order to constrain the smoothness of the obtained shortest path, the computer device constrains the radial distance between the first candidate point and the second candidate point to meet a second preset condition. The second preset condition can be set as needed, for example, the radial distance between the first candidate point and the second candidate point is less than a preset threshold, or the radial distance between the first candidate point and the second candidate point is within a preset range. For example, the polar coordinates of the first candidate point are ( , ), the polar coordinates of the second candidate point are recorded as ( , ),but . dR is a preset value, i and j are two adjacent positive integers.

[0119] In the above embodiment, by constraining the radial distance between the first candidate point and the second candidate point to meet the second preset condition, it is possible to reduce the large jumps between target points in the shortest path and improve the smoothness of the determined shortest path.

[0120] Figure 14 FIG. 1 is a flow chart of determining the edge of the optic disc in one embodiment. In an exemplary embodiment, as shown in FIG. Figure 14 As shown, the above-mentioned "determining the optic disc edge based on the closed curve formed by the identifiable layer" includes S1401 to S1404.

[0121] S1401 , filtering the first scan data based on the identifiable layer to obtain second scan data within a first preset range where the identifiable layer is located.

[0122] In this embodiment, the first preset range is a range based on the identifiable layer, which can be set as needed. For example, the first preset range can include a range between the first sub-range and the second sub-range.

[0123] Optionally, to eliminate the influence of data above the Bruce membrane, the first preset range can include a preset distance from the recognizable layer to the recognizable layer below. For example, the computer device selects the recognizable layer and the second scan data 100 microns below the recognizable layer from the first scan data. This eliminates the influence of data above the Bruce membrane, and the maximum contrast of the resulting image occurs at the junction of the choroid / sclera and the optic nerve, reducing contrast interference from the retinal edge. This allows accurate results to be obtained even for eyes with paraoptic atrophy, thereby improving the accuracy and stability of the resulting first curve.

[0124] S1402: Perform axial projection on the second scanning data to obtain a two-dimensional projection image.

[0125] In this embodiment, axial projection refers to projection along the axial direction, i.e., the optical axis. Optionally, the computer device may perform axial projection on the second scan data using a preset projection algorithm to obtain a two-dimensional projection image. The preset projection algorithm includes, but is not limited to, at least one of a maximum intensity projection (MIP) algorithm, an average intensity projection (AIP) algorithm, or a weighted projection algorithm.

[0126] S1403: Perform segmentation based on the two-dimensional projection image to determine a first curve corresponding to the optic disc edge in the two-dimensional projection image.

[0127] In this embodiment, the computer device may optionally segment the two-dimensional projection image using a preset segmentation algorithm to determine a first curve corresponding to the optic disc edge in the two-dimensional projection image. The preset segmentation algorithm includes, but is not limited to, at least one of a threshold segmentation algorithm and a morphological segmentation algorithm. It is understood that the first curve is the curve corresponding to the optic disc edge on the xy plane.

[0128] S1404 , projecting the first curve axially to the identifiable layer to determine the optic disc edge of the optic nerve head.

[0129] In this embodiment, illustratively, the computer device may project the first curve along the axial direction to obtain a projection result, and determine the position of the optic disc edge in the axial direction based on the intersection of the projection result and the identifiable layer to determine the three-dimensional optic disc edge.

[0130] In the above embodiment, since the first scan data can be filtered based on the identifiable layer, second scan data can be obtained within the first preset range of the identifiable layer that meets actual requirements. Furthermore, since the second scan data can be axially projected to obtain a two-dimensional projection image, segmentation based on the two-dimensional projection image can efficiently determine the first curve corresponding to the optic disc edge in the two-dimensional projection image. Thus, by axially projecting the first curve onto the identifiable layer, the optic disc edge of the optic nerve head can be more accurately determined.

[0131] In an exemplary embodiment, optionally, the above-mentioned S1403 includes: inputting the two-dimensional projection image into a first machine learning model to determine a first classification result of each pixel point in the two-dimensional projection image; performing a first post-processing on the first classification result of each pixel point in the two-dimensional projection image to obtain a first curve.

[0132] In this embodiment, the first machine learning model includes a supervised learning model, a semi-supervised learning model, or an unsupervised learning model. Exemplarily, the machine learning model may include, but is not limited to, a convolutional neural network (CNN) model, a recurrent neural network (RNN) model, a fully convolutional neural network (FCN) model, a radial basis function (RBF) model, a deep belief network (DBN) model, an Elman model, or at least one of a combination thereof. In some embodiments, the first machine learning model includes a deep learning model based on a Unet structure.

[0133] Optionally, the first machine learning model is trained based on multiple two-dimensional projection image samples and a first classification label corresponding to each two-dimensional projection image sample. The first classification label indicates whether each pixel in the two-dimensional projection image sample belongs to the optic disc region. In this way, the first machine learning module is capable of determining whether each pixel in the two-dimensional projection image belongs to the optic disc region.

[0134] Then, the computer device inputs the two-dimensional projection image into the first machine learning model, thereby determining a first classification result for each pixel in the two-dimensional projection image. The first classification result indicates whether the corresponding pixel belongs to the optic disc area. For example, after inputting the two-dimensional projection image into the first machine learning model, the computer device can obtain a binary classification image of the two-dimensional projection image, where one class of pixels in the binary classification image represents pixels within the optic disc area, and the other class of pixels in the binary classification image represents pixels outside the optic disc area.

[0135] Furthermore, the computer device may perform a first post-processing on the first classification results of each pixel in the two-dimensional projection image to obtain a first curve. Optionally, the first post-processing includes, but is not limited to, at least one of connected domain extraction, smoothing, enhancement, noise reduction, and filtering. Furthermore, the computer device may first perform a connected domain analysis based on the first classification results of each pixel in the two-dimensional projection image to obtain a target connected domain, and then obtain the first curve based on a closed curve on the outer envelope of the target connected domain.

[0136] For example, the computer device may extract the largest connected domain from the first classification result of each pixel in the two-dimensional projection image, and use the closed curve outside the largest connected domain as the first curve. In some embodiments, the computer device may further smooth the closed curve outside the largest connected domain to obtain the first curve.

[0137] In the above embodiment, since the two-dimensional projection image can be input into the first machine learning model to determine the first classification result for each pixel in the two-dimensional projection image, and the first classification result is used to indicate whether the corresponding pixel belongs to the optic disc area, by performing a first post-processing on the first classification result for each pixel in the two-dimensional projection image, a first curve corresponding to the optic disc edge can be obtained efficiently and accurately through machine learning.

[0138] In an exemplary embodiment, optionally, any one of the following items is performed in S1402 above:

[0139] (1) Perform mean projection on the second scan data to obtain a two-dimensional projection image.

[0140] Exemplarily, the computer device may determine an average value of pixel values corresponding to all pixels in the axial direction of the second scanning data to obtain a corresponding two-dimensional projection image.

[0141] (2) Performing noise reduction processing on the second scan data to obtain third scan data, and performing maximum projection on the third scan data to obtain a two-dimensional projection image.

[0142] Optionally, the computer device can perform noise reduction processing on the second scan data along the axial direction of the second scan data using a preset noise reduction algorithm. Further optionally, the computer device can perform noise reduction processing on all the second scan data in the axial direction to obtain the third scan data, or can perform noise reduction processing on part of the second scan data in the axial direction to obtain the third scan data, wherein the preset noise reduction algorithm includes but is not limited to a low-pass noise reduction algorithm, a filtering noise reduction algorithm, a transform domain noise reduction algorithm or a deep learning noise reduction algorithm. This embodiment is not limited. It is understandable that the third scan data is still a 3D volume data. Exemplarily, the computer device can perform low-pass noise reduction along the axial direction of the second scan data, and the low-pass window size can be set as required, for example, to 30 microns.

[0143] Furthermore, the computer device performs maximum projection on the third scan data to obtain a two-dimensional projection image. For example, the computer device may determine the maximum value of pixel values corresponding to all pixels in the axial direction along the axial direction of the second scan data to obtain the corresponding two-dimensional projection image.

[0144] In the above embodiment, the signal-to-noise ratio of the two-dimensional projection image can be improved by using the mean projection method. The signal-to-noise ratio of the two-dimensional projection image and the ability to identify the choroid layer / sclera layer can also be improved by performing noise reduction processing on the second scan data to obtain the third scan data, and performing maximum projection on the third scan data to obtain the two-dimensional projection image.

[0145] Figure 15 is a schematic diagram of a two-dimensional projection image and a first curve in one embodiment, as shown in FIG. Figure 15 As shown, Figure 15 Figure (a) shows a two-dimensional projection diagram. Figure 15 Figure (b) shows a first classification result, where the areas belonging to the optic disc are displayed in white, and those not belonging to the optic disc are displayed in black. Figure 15 Figure (c) shows a first curve obtained based on the two-dimensional projection image.

[0146] In an exemplary embodiment, optionally, the above-mentioned "determining the internal limiting membrane and the identifiable layer closest to the optic nerve based on the first scanning data including the optic nerve head" includes: determining the fundus scanning image corresponding to the first scanning data; inputting the fundus scanning image into a second machine learning model to determine a second classification result for each pixel in the fundus scanning image; performing a second post-processing on the second classification result for each pixel in the fundus scanning image to determine the identifiable layer and the internal limiting membrane.

[0147] In this embodiment, optionally, the computer device can obtain fundus scan images sent by other devices, or can obtain fundus scan images from a preset storage space, but this embodiment is not limited to this. The fundus scan images include but are not limited to B-scan images. For example, the computer device can obtain OCT data of the posterior segment of the eye including the ONH position, and perform image reconstruction based on the OCT data, and use the obtained B-scan image as the fundus scan image. In some embodiments, the computer device can also reconstruct a set of B-scan images to obtain volume data, and perform layered processing based on the volume data.

[0148] The second machine learning model includes a supervised learning model, a semi-supervised learning model or an unsupervised learning model. Exemplarily, the machine learning model may include but is not limited to a convolutional neural network (CNN) model, a recurrent neural network (RNN), a full convolutional neural network (FCN) model, a generative adversarial network (GAN) model, a back-propagation (BP) machine learning model, a radial basis function (RBF) model, a deep belief network (DBN) model, an Elman model, or at least one of its combined models. In some embodiments, the second machine learning model includes a deep learning model based on a Unet structure.

[0149] Optionally, the second machine learning model is trained based on multiple fundus scan image samples and a second classification label corresponding to each fundus scan image sample. The second classification label is used to indicate the ocular physiological structure to which each pixel in the fundus scan image sample belongs, for example, marking the pixel as belonging to the choroid layer, sclera layer, retinal pigment epithelium (RPE) layer, and other ocular physiological structures. In this way, the first machine learning module is capable of determining the ocular physiological structure to which each pixel in the fundus scan image belongs based on the fundus scan image.

[0150] Furthermore, the computer device inputs the fundus scan image into the second machine learning model to determine a second classification result for each pixel in the fundus scan image, wherein the second classification result is used to characterize the ocular physiological structure to which the corresponding pixel belongs.

[0151] Furthermore, the computer device performs a second post-processing on the second classification results of each pixel in the fundus scan image to determine the identifiable layers and the internal limiting membrane. Optionally, the second post-processing includes but is not limited to at least one of screening processing, smoothing processing, enhancement processing, or filtering processing. Further optionally, the computer device can analyze the second classification results of each pixel in the fundus scan image to obtain the identifiable layers and the internal limiting membrane. Exemplarily, the computer device analyzes the second classification results to determine a first interface between the choroid and the retinal layer above the choroid, and determines the identifiable layers based on the first interface, and determines a second interface between the vitreous body and the optic nerve layer, and determines the internal limiting membrane based on the second interface.

[0152] In some embodiments, in determining the identifiable layer, at the location where paraoptic disc atrophy exists, if the choroid is not atrophic, the identifiable layer is located at the junction of the choroid and the overlying tissue; if the choroid is atrophic, at the location where the choroid is atrophic, the identifiable layer is located at the junction of the sclera and the overlying tissue.

[0153] In the above embodiment, since the fundus scanning image corresponding to the first scanning data can be determined, the fundus scanning image is input into the second machine learning model to determine the second classification result of each pixel in the fundus scanning image, and the second classification result is used to characterize the physiological structure of the eye to which the corresponding pixel belongs. Therefore, after the second post-processing of the second classification result of each pixel in the fundus scanning image, the identifiable layers and internal limiting membrane can be determined efficiently and accurately through machine learning.

[0154] In an exemplary embodiment, the fundus data optionally includes an optic cup volume. S704 includes: performing plane fitting on the optic cup edge to determine a fitting plane; determining an enclosed space based on the fitting plane, the internal limiting membrane, and the optic cup edge; and determining the optic cup volume of the optic nerve head based on the volume of the enclosed space.

[0155] In this embodiment, after obtaining the optic cup edge, the computer device can determine the optic cup volume based on the volume of the area enclosed by the optic cup edge on the ILM surface. The computer device performs a plane fitting on the optic cup edge to obtain a corresponding fitting plane. The computer device then uses the fitting plane and the ILM surface corresponding to the internal limiting membrane as the upper and lower boundaries of the enclosed space, and uses the annular surface formed by extending the closed curve formed by the optic cup edge in the xy plane along the axial direction as the lateral boundaries of the enclosed space. The x and y directions are perpendicular to the axial direction, i.e., the z direction.

[0156] Furthermore, the computer device may use the volume of the enclosed space as the optic cup volume of the optic nerve head. In some embodiments, the computer device may further process the volume of the enclosed space to obtain the optic cup volume of the optic nerve head.

[0157] In the above embodiment, since the optic cup edge can be plane-fitted to determine the fitting plane, and the closed space is determined based on the fitting plane, the internal limiting membrane and the optic cup edge, the optic cup volume of the optic nerve head can be determined more accurately based on the volume of the closed space.

[0158] In some embodiments, the computer device may optionally store fundus data. This embodiment does not limit the storage method of the computer device.

[0159] In some embodiments, optionally, the computer device can display fundus data. This embodiment does not limit the display method of the computer device.

[0160] Figure 16 A schematic diagram showing an embodiment of the present invention is shown in FIG. Figure 16 As shown, the computer device can show the position of the edge of the optic disc and the edge of the optic cup on the projection graphic. In one embodiment, the transparency of the projected image can also be adjusted.

[0161] Furthermore, the computer device can display fundus data. For example, Figure 16 As shown, the computer device can display the average cup-to-disc ratio, average retinal nerve fiber layer thickness, optic cup area, optic cup volume, rim area, vertical cup-to-disc ratio, and horizontal cup-to-disc ratio, where mm represents millimeters and μm represents micrometers.

[0162] In order to more clearly introduce the fundus data determination method of this application, Figure 17 Provide explanation. Figure 17 FIG. 1 is a process diagram of a method for determining fundus data in one embodiment. Figure 17 As shown, the computer device can execute the fundus data determination method according to the following process.

[0163] S1701, determining a fundus scan image corresponding to first scan data.

[0164] S1702: Input the fundus scan image into a second machine learning model to determine a second classification result for each pixel in the fundus scan image.

[0165] S1703 , performing a second post-processing on the second classification result of each pixel point in the fundus scan image to determine identifiable layers and internal limiting membranes.

[0166] S1704: Filter the first scan data based on the identifiable layer to obtain second scan data within a first preset range where the identifiable layer is located.

[0167] S1705: Perform axial projection on the second scan data to obtain a two-dimensional projection image. The two-dimensional projection image can be obtained by performing mean projection on the second scan data, or by performing noise reduction processing on the second scan data to obtain third scan data, and then performing maximum projection on the third scan data to obtain the two-dimensional projection image.

[0168] S1706: Input the two-dimensional projection image into a first machine learning model to determine a first classification result for each pixel in the two-dimensional projection image.

[0169] S1707 , performing a first post-processing on the first classification result of each pixel point in the two-dimensional projection image to obtain a first curve corresponding to the optic disc edge.

[0170] S1708 , projecting the first curve axially to the identifiable layer to determine the optic disc edge of the optic nerve head.

[0171] S1709: Convert the pixel points of the inner limiting membrane into candidate points in the polar coordinate system.

[0172] S1710: Convert the pixel points at the edge of the optic disc into polar coordinate points in a polar coordinate system.

[0173] S1711: Determine a first loss value corresponding to any candidate point based on the area of the first triangle.

[0174] S1712: Determine a second loss value corresponding to a line between the first candidate point and the second candidate point according to the area of the second triangle.

[0175] S1713: Determine the shortest path according to the first loss value and the second loss value.

[0176] S1714: Determine the target curve based on the shortest path.

[0177] S1715, determining the optic cup edge of the optic nerve head based on the target curve.

[0178] S1716, determining fundus data based on the optic disc edge and the optic cup edge.

[0179] The processes of S1701 to S1716 can refer to the above embodiment and will not be described again here.

[0180] Figure 18This is a comparison chart of the effects of the related art and the present application. It is assumed that the continuity of the surface area formed by the closed curve extending from the edge of the optic disc to the internal limiting membrane is not considered. Instead, a plane with the minimum area is independently solved in each meridian angle fraction in the polar coordinates, and then the sum of all minimum areas is calculated. The surface obtained in this way is actually likely to be fragmented in 3D space, and the area of this fragmented surface cannot best represent the total area of the optic nerve.

[0181] Furthermore, when calculating the optic cup volume, if the original calculated optic cup edge position is used, the physiological significance of the calculated optic cup volume result is no longer clear because the original optic cup edge position has no spatial continuity.

[0182] like Figure 18 As shown, Figure 18 Figure (a) shows the optic disc edge and optic cup edge without considering the continuity of the surface area formed by the closed curve extending from the optic disc edge to the internal limiting membrane. Figure 18 Figure (b) shows the optic disc and optic cup edges obtained in this application. The outer closed curve represents the optic disc position. The inner closed curve represents the optic cup position. It can be seen that if the continuity of the surface area is not considered, the obtained optic cup position will show abrupt changes on the left side of the ONH, making its physiological significance unclear. However, the optic cup position obtained in this application is more continuous and smooth.

[0183] It can be seen that since the present application minimizes the area of a continuous surface, it can most realistically reflect the total amount of all optic nerves, and the result obtained is also a closed loop with a high degree of continuity, making the subsequent calculation of the optic cup volume more stable and physiologically meaningful, and the accuracy and repeatability of quantitative indicators such as the C / D ratio of the ONH are higher.

[0184] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0185] Based on the same inventive concept, embodiments of the present application further provide a fundus data determination device for implementing the aforementioned fundus data determination method. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more fundus data determination device embodiments provided below can be found in the limitations of the fundus data determination method described above and will not be further elaborated here.

[0186] Figure 19 FIG. 1 is a block diagram of a fundus data determination device in one embodiment. In an exemplary embodiment, as shown in FIG. Figure 19 As shown, a fundus data determination device 1900 is provided, comprising: a first determination module 1901, a second determination module 1902, a third determination module 1903 and a fourth determination module 1904, wherein:

[0187] The first determining module 1901 is configured to determine the internal limiting membrane and the edge of the optic disc based on first scanning data containing the optic nerve head.

[0188] The second determining module 1902 is configured to determine, based on a surface area formed by a closed curve extending from the optic disc edge to the internal limiting membrane, a closed curve whose surface area satisfies a preset condition as a target curve.

[0189] The third determining module 1903 is configured to determine the optic cup edge of the optic nerve head based on the target curve.

[0190] The fourth determining module 1904 is configured to determine fundus data according to the optic disc edge and the optic cup edge.

[0191] Each module in the fundus data determination device described above may be implemented in whole or in part through software, hardware, or a combination thereof. Each module may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in the computer device in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0192] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0193] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0194] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0195] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), quantum computing-based data processing logic devices, artificial intelligence (AI) processors, and the like.

[0196] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, 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 application.

[0197] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for determining fundus data, characterized in that: The method comprises: determining an internal limiting membrane and an optic disc margin based on first scan data including the optic nerve head; Based on a surface area formed by a closed curve extending from the optic disc edge to the internal limiting membrane, determining a closed curve when the surface area meets a preset condition as a target curve; determining an optic cup edge of the optic nerve head based on the target curve; The fundus data is determined based on the optic disc edge and the optic cup edge.

2. The method according to claim 1, characterized in that The method of determining the internal limiting membrane and the optic disc edge based on the first scan data including the optic nerve head comprises: determining the internal limiting membrane and an identifiable layer proximal to the optic nerve based on first scan data including the optic nerve head; The optic disc edge is determined based on a closed curve formed by the identifiable layers.

3. The method according to claim 2, characterized in that The identifiable layer closest to the optic nerve includes at least one of Bruce's membrane, choroid, and sclera.

4. The method according to claim 1, wherein The method further comprises: performing triangulation processing on a surface formed by a closed curve extending from the edge of the optic disc to the internal limiting membrane to obtain a plurality of triangles; The surface area is determined according to the sum of the areas of the triangles.

5. The method according to claim 4, characterized in that The step of determining, based on a surface area formed by a closed curve extending from the optic disc edge to the internal limiting membrane, a closed curve when the surface area satisfies a preset condition as a target curve comprises: Converting the pixel points of the inner limiting membrane into candidate points in a polar coordinate system; Converting the pixel points at the edge of the optic disc into polar coordinate points in the polar coordinate system; each polar coordinate point corresponds to a polar angle; According to the areas of the triangles corresponding to the polar coordinate points and the candidate points, based on the shortest path method, a closed curve is determined from each of the candidate points when the surface area meets the preset condition as the target curve.

6. The method according to claim 5, characterized in that The step of determining, based on the shortest path method and the areas of the triangles corresponding to the polar coordinate points and the candidate points, a closed curve when the surface area satisfies the preset condition from each of the candidate points as the target curve includes: Determine, based on the area of the first triangle, a first loss value corresponding to any candidate point; the first triangle is determined based on the candidate point, a first polar coordinate point, and a second polar coordinate point, where the first polar coordinate point is a polar coordinate point corresponding to the same polar angle as the candidate point; and the first polar coordinate point and the second polar coordinate point are polar coordinate points corresponding to two adjacent polar angles along the polar axis; Determine, based on the area of the second triangle, a second loss value corresponding to a line between the first candidate point and the second candidate point; the second triangle is determined based on the first candidate point, the second candidate point, and a third polar coordinate point; the first candidate point and the second candidate point are two different candidate points, and the first candidate point and the second candidate point correspond to two adjacent polar angles along the polar axis, and the third polar coordinate point is a polar coordinate point corresponding to the same polar angle as the second candidate point; Determine a shortest path based on the first loss value and the second loss value; the shortest path includes the target point corresponding to each of the polar coordinate points in the candidate points and the connecting lines between the target points, and the sum of the first loss value and the second loss value corresponding to the shortest path is the smallest; The target curve is determined according to the shortest path.

7. The method according to claim 6, characterized in that A radial distance between the first candidate point and the second candidate point satisfies a second preset condition.

8. The method according to claim 2 or 3, characterized in that The determining the optic disc edge based on the closed curve formed by the identifiable layers comprises: Filtering the first scan data based on the identifiable layer to obtain second scan data within a first preset range where the identifiable layer is located; Performing axial projection on the second scanning data to obtain a two-dimensional projection image; performing segmentation based on the two-dimensional projection image, and determining a first curve corresponding to the optic disc edge in the two-dimensional projection image; The first curve is projected axially onto the identifiable layer to determine the optic disc edge of the optic nerve head.

9. The method according to claim 8, characterized in that The performing segmentation based on the two-dimensional projection image to determine a first curve corresponding to the optic disc edge in the two-dimensional projection image includes: Inputting the two-dimensional projection image into a first machine learning model to determine a first classification result for each pixel in the two-dimensional projection image; the first classification result is used to indicate whether the corresponding pixel belongs to the optic disc area; A first post-processing is performed on the first classification result of each pixel point in the two-dimensional projection image to obtain the first curve.

10. The method according to any one of claims 1 to 7, characterized in that The fundus data includes an optic cup volume; and determining the fundus data according to the optic disc edge and the optic cup edge includes: Performing plane fitting on the edge of the optic cup to determine a fitting plane; determining a closed space based on the fitting plane, the internal limiting membrane, and the optic cup edge; The optic cup volume of the optic nerve head is determined according to the volume of the closed space.

Images