Fundus image structure recognition method and medical apparatus

Abstract

The present disclosure relates to the technical field of image recognition, and particularly relates to a fundus image structure recognition method and medical equipment. The method comprises: acquiring a three-dimensional OCT image of a to-be-scanned part, wherein the OCT image comprises at least N frames of B-scan images with continuous structures, and N is a positive integer; sampling along a B-scan scanning direction of the three-dimensional OCT image to form a set of sampling slices; outputting structure boundaries of each slice in the set of sampling slices based on the set of sampling slices; and mapping information of the structure boundaries of each slice to the three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary. The present disclosure can improve the accuracy of fundus image structure recognition while reducing the number of samplings and the limitation of three-dimensional OCT images.

Description

Technical Field

[0001] This disclosure relates to the field of image recognition technology, and in particular to a method and medical device for recognizing fundus image structures. Background Technology

[0002] With the development of science and technology, image recognition technology is continuously facilitating users' production and daily lives. For example, morphological changes in the posterior segment of the eye are an important basis for the diagnosis of eye diseases. Optical coherence tomography (OCT) images can capture these subtle changes, and retinal structure segmentation based on OCT images is of great significance for clinical judgment. However, with the continuous expansion of scanning and diagnostic scope, and the widespread application of more information technology in diagnosis and treatment, there is a need to effectively identify more tissue structures and more complex pathological changes. In actual diagnosis and treatment scenarios, the requirements for the immediacy and robustness of information processing algorithms are also constantly increasing. Summary of the Invention

[0003] This disclosure provides a method and medical device for recognizing fundus image structures, which can improve the accuracy of fundus image structure recognition while reducing the number of samplings and the limitations of three-dimensional OCT images. The technical solution of this disclosure is as follows:

[0004] According to a first aspect of the present disclosure, a method for recognizing fundus image structures is provided, comprising: Acquire a three-dimensional OCT image of the area to be scanned, wherein the OCT image includes at least N frames of structurally continuous brightness scan B-scan images, where N is a positive integer; Samples are taken along the B-scan scanning direction of the three-dimensional OCT image to form a set of sampled slices; Based on the sampled slice set, output the structural boundary of each slice in the sampled slice set; The structural boundary information of each slice is mapped onto the three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary.

[0005] According to some embodiments, the method further includes: Based on the three-dimensional OCT image or the current eye image, obtain the transition information during the acquisition process of the three-dimensional OCT image; Based on the transition information, determine the B-scan images corresponding to all frames before and after the transition in the at least N frame structurally continuous B-scan images.

[0006] According to some embodiments, the sampling along the B-scan scanning direction of the three-dimensional OCT image to form a set of sampling slices includes: Along the B-scan scanning direction of the three-dimensional OCT image, a portion of the B-scan image is extracted at equal intervals. With the jump information present, B-scan images corresponding to all frames before and after the jump are extracted from the B-scan images of at least N consecutive frames to form the sampling slice set.

[0007] According to some embodiments, mapping the structural boundary information of each slice onto the three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary includes: Based on the first slice subset, boundary assignment processing is performed on all unsampled frames in the non-jump intervals of the three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary. The first slice subset includes at least one slice whose structural boundary includes the boundary of each layer of tissue structure in the fundus. The non-jump interval is obtained by dividing at least N frames of structurally continuous B-scan images according to the jump information, wherein the B-scan images of the corresponding frames before and after the jump do not exist in the same non-jump interval at the same time.

[0008] According to some embodiments, mapping the structural boundary information of each slice onto the three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary includes: Outlier processing is performed on the second slice subset to obtain the third slice subset, wherein the second slice subset includes at least one slice whose structural boundary includes the view disk structural boundary; Based on the third slice subset, the unsampled frames of the at least N frame structurally continuous B-scan images are subjected to boundary assignment processing to form a three-dimensional OCT volume data structure boundary.

[0009] According to some embodiments, the outlier processing of the second slice subset to obtain the third slice subset includes: Based on the second slice subset, multiple sets of corresponding En Face images and the identified boundaries are formed; Based on the continuity relationship of the multiple sets of corresponding En Face images and the identified boundaries, the B-scan images containing outliers are filtered out to obtain the third slice subset.

[0010] According to some embodiments, mapping the structural boundary information of each slice onto the three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary includes: Based on the structural boundary information of each slice, boundary assignment processing is performed on the unsampled frames of the at least N structurally continuous B-scan images to form a three-dimensional OCT volume data structure boundary. The boundary assignment processing includes shifting the boundary coordinates of each slice in the at least one slice by the relative offset distance between frames to obtain the boundary coordinates of the unsampled frames, or obtaining the boundary coordinates of the unsampled frames through interpolation processing. The interpolation processing includes at least one of linear interpolation processing, piecewise quadratic interpolation processing, and cubic spline interpolation processing.

[0011] According to a second aspect of the present disclosure, a medical device is provided for performing the method as described in any of the preceding claims, comprising: OCT scanning unit; Control and image processing unit; Display unit; The control and image processing unit is used to identify the boundary information of the OCT volume data structure using any of the methods described above, and the display unit is used to display all or part of the boundary information of the OCT volume data structure.

[0012] According to a fourth aspect of the present disclosure, a fundus image structure recognition device is provided, comprising: An image acquisition unit is used to acquire a three-dimensional OCT image of the area to be scanned, wherein the OCT image includes at least N frames of structurally continuous B-scan images, where N is a positive integer; A set forming unit is used to sample along the B-scan scanning direction of the three-dimensional OCT image to form a set of sampled slices; A boundary output unit is used to output the structural boundary of each slice in the sampled slice set based on the sampled slice set; A boundary forming unit is used to map the structural boundary information of each slice onto the three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary.

[0013] According to a fourth aspect of the present disclosure, an electronic device is provided, comprising: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the fundus image structure recognition method described in any one of the preceding aspects.

[0014] According to a fifth aspect of the present disclosure, a storage medium is provided that, when instructions in the storage medium are executed by a processor of an electronic device, enables the electronic device to perform the fundus image structure recognition method described in any one of the preceding aspects.

[0015] According to a sixth aspect of the present disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the method described in any one of the preceding aspects.

[0016] The technical solutions provided by the embodiments of this disclosure have at least the following beneficial effects: In some or related embodiments, a three-dimensional OCT image of the area to be scanned is acquired, the OCT image comprising at least N structurally continuous B-scan images, where N is a positive integer; sampling is performed along the B-scan scanning direction of the three-dimensional OCT image to form a set of sampled slices; based on the set of sampled slices, the structural boundaries of each slice in the set of sampled slices are output; the information of the structural boundaries of each slice is mapped onto the three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary. Therefore, fundus image structure recognition can be performed with only one sampling, eliminating the need for two rounds of sampling for coarse localization and fine segmentation, reducing the number of samplings, reducing the complexity of fundus image structure recognition, and eliminating the need to restrict the acquired three-dimensional OCT image, reducing the situation where only eye OCT images centered on the optic disc can be processed, reducing the situation where image limitations prevent fundus image structure recognition, and reducing the situation where the recognition robustness is poor due to the use of a star-shaped method to synthesize slices in two rounds of sampling. This can improve both the accuracy and efficiency of fundus image structure recognition. That is, by using this disclosure, the processing speed of a large amount of volume data information can be greatly accelerated through a single sampling and interpolation. At the same time, during the sampling process, jump information is identified, and the accuracy of interpolated data is improved; thus, the efficient identification of complex structural information is satisfied.

[0017] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure, and are not intended to unduly limit this disclosure.

[0019] Figure 1 This is a flowchart of the first fundus image structure recognition method provided in the embodiments of this disclosure; Figure 2 This is an example schematic diagram of a three-dimensional OCT image provided in an embodiment of this disclosure; Figure 3 This is a schematic diagram illustrating an example of a cross-sectional structure of the macula and optic disc provided in an embodiment of this disclosure; Figure 4 This is a schematic diagram illustrating an example of a structural boundary provided in an embodiment of this disclosure; Figure 5 This is a flowchart of the second fundus image structure recognition method provided in the embodiments of this disclosure; Figure 6 This is a schematic diagram illustrating an example of a viewing disc structure provided in an embodiment of this disclosure; Figure 7 This is a schematic diagram illustrating an example of hierarchical boundary interpolation provided in an embodiment of this disclosure; Figure 8 This is an example schematic diagram of the mapping of the visual disk structure boundary on a B-scan image provided in an embodiment of this disclosure; Figure 9 This is a block diagram of a medical device provided in an embodiment of this disclosure; Figure 10 This is a block diagram illustrating a fundus image structure recognition device according to an exemplary embodiment; Figure 11 This is an example schematic diagram of an electronic device according to an exemplary embodiment. Detailed Implementation

[0020] To enable those skilled in the art to better understand the technical solutions of this disclosure, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0021] This disclosure provides a method for recognizing fundus image structures and a medical device. In some embodiments, the terms "fundus image structure recognition method" and "information processing method" and "communication method" can be used interchangeably; the terms "fundus image structure recognition device" and "information processing device" and "communication device" can be used interchangeably; and the terms "information processing system" and "communication system" can be used interchangeably.

[0022] This disclosure is not exhaustive, but merely illustrative of some embodiments, and is not intended to limit the scope of protection of this disclosure. Unless otherwise specified, each step in a particular embodiment can be implemented as an independent embodiment, and the steps can be arbitrarily combined. For example, a solution after removing some steps in a particular embodiment can also be implemented as an independent embodiment, and the order of the steps in a particular embodiment can be arbitrarily interchanged. Furthermore, the optional implementation methods in a particular embodiment can be arbitrarily combined; moreover, the embodiments can be arbitrarily combined, for example, some or all steps of different embodiments can be arbitrarily combined, and a particular embodiment can be arbitrarily combined with the optional implementation methods of other embodiments.

[0023] In each of the disclosed embodiments, unless otherwise specified or in case of logical conflict, the terminology and / or descriptions of the embodiments are consistent and can be referenced by each other. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0024] The terminology used in the embodiments of this disclosure is for the purpose of describing particular embodiments only and is not intended to limit the scope of this disclosure.

[0025] In this disclosure, unless otherwise stated, elements expressed in the singular form, such as "a," "an," "the," "the," "the," "the," "the," "the," "this," etc., can mean "one and only one," or "one or more," "at least one," etc. For example, when using articles such as "a," "an," "the," etc. in translation, the noun following the article can be understood as either a singular or a plural expression.

[0026] In the embodiments disclosed herein, "multiple" refers to two or more.

[0027] In some embodiments, the terms “at least one of,” “one or more,” “a plurality of,” and “multiple” may be used interchangeably.

[0028] The prefixes "first," "second," etc., used in the embodiments of this disclosure are merely for distinguishing different descriptive objects and do not impose restrictions on the position, order, priority, quantity, or content of the descriptive objects. The description of the descriptive objects is found in the claims or the context of the embodiments, and the use of prefixes should not constitute unnecessary restrictions. For example, if the descriptive object is a "field," the ordinal numbers preceding "field" in "first field" and "second field" do not restrict the position or order of the "fields." "First" and "second" do not restrict whether the "fields" they modify are in the same message, nor do they restrict the order of "first field" and "second field." Similarly, if the descriptive object is a "level," the ordinal numbers preceding "level" in "first level" and "second level" do not restrict the priority between "levels." Furthermore, the number of descriptive objects is not limited by ordinal numbers and can be one or more. For example, in "first device," the number of "devices" can be one or more. Furthermore, the objects modified by different prefixes can be the same or different. For example, if the object being described is "device", then "first device" and "second device" can be the same device or different devices, and their types can be the same or different. Similarly, if the object being described is "information", then "first information" and "second information" can be the same information or different information, and their content can be the same or different.

[0029] In some embodiments, "terminal" or "terminal device" may be referred to as "user equipment (UE)," "user terminal," "mobile station (MS)," "mobile terminal (MT)," "subscriber station," "mobile unit," "subscriber unit," "wireless unit," "remote unit," "mobile device," "wireless device," "wireless communication device," "remote device," "mobile subscriber station," "access terminal," "mobile terminal," "wireless terminal," "remote terminal," "handset," "user agent," "mobile client," "client," etc.

[0030] In some embodiments, data, information, etc., may be obtained with the user's consent.

[0031] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0032] Figure 1 This is a flowchart of the first fundus image structure recognition method provided in the embodiments of this disclosure, such as... Figure 1 As shown, this fundus image structure recognition method can be used in scenarios involving the recognition of morphological changes in the posterior segment of the eye, and includes the following steps: In step S11, a three-dimensional OCT image of the area to be scanned is acquired. The OCT image includes at least N frames of structurally continuous B-scan images, where N is a positive integer. In some embodiments, the executing entity of this disclosure may be, for example, an electronic device. This electronic device does not specifically refer to a particular fixed electronic device. For example, when the device identifier changes, the electronic device may also change accordingly. For example, when the structure of the electronic device changes, the electronic device may also change accordingly. Furthermore, the executing entity of this disclosure may also be a server, which may be a single server or a server cluster; this disclosure does not limit this.

[0033] In some embodiments, the area to be scanned may be, for example, a region where structural information needs to be determined. This area may be, for example, the eye. The area to be scanned is not specifically a fixed location. For example, the area to be scanned may be the right eye or the left eye. For example, when a user identifier changes, the corresponding eye may also change accordingly, and the area to be scanned may also change accordingly.

[0034] In some embodiments, Figure 2 This is an example schematic diagram of a three-dimensional OCT image provided in an embodiment of this disclosure, such as... Figure 2 As shown, the axial scan of a given tissue structure by OCT is called an amplitude scan (A-scan), followed by a transverse A-scan sequence to form a cross-sectional image, called a B-scan. The OCT volume data consists of multiple B-scan frames, with the scanning range covering a specific region of the retina. An OCT scan is as follows: Figure 2 As shown. Axial projection of the OCT volume data can generate a frontal image (Enface) of the structure, such as... Figure 2 As shown, the positional correspondence between the En face and the OCT volume data is as follows: the i-th row and j-th column of the En face image corresponds to all or part of the pixel statistics of the j-th column of the i-th frame B-scan of the OCT volume data. The statistics can be, for example, mean statistics, maximum statistics, etc. Whether all or part of the pixels are projected depends on the choice of the projection hierarchy, i.e., projecting onto the entire retina (all pixels) or projecting onto a specific hierarchy (partial pixels).

[0035] In some embodiments, a three-dimensional OCT image may include, for example, at least N frames of structurally continuous B-scan images, where N is a positive integer. This three-dimensional OCT image does not refer to a specific fixed image. For example, the three-dimensional OCT image may change accordingly when N changes.

[0036] In some embodiments, for example, a three-dimensional OCT image of the area to be scanned can be acquired, the OCT image comprising at least N frames of structurally continuous B-scan images, where N is a positive integer.

[0037] In step S12, sampling is performed along the B-scan scanning direction of the three-dimensional OCT image to form a set of sampling slices; The sampling method in this embodiment is not limited. For example, the sampling method can also change when the sampling interval changes. For example, the sampling method can also change when the jump parameters of the 3D OCT image change.

[0038] In some embodiments, the sample slice set may be a collection of at least one sample slice. This sample slice may, for example, be a frame of a B-scan image in a 3D OCT image. The sample slice set is not specifically a fixed set. For example, the sample slice set may change when the number of slices in it changes. Similarly, the sample slice set may change when a particular sample slice within it changes.

[0039] In some embodiments, sampling can be performed along the B-scan scanning direction of the 3D OCT image to form a set of sampled slices. The sampling method is not limited. For example, it can include equally spaced sampling, or it can be determined according to a sampling strategy. This sampling strategy could, for example, be based on image recognition results.

[0040] In step S13, based on the sampled slice set, the structural boundaries of each slice in the sampled slice set are output; In some embodiments, structural boundaries may include, for example, hierarchical structural boundaries and visual disk structural boundaries. The structural boundaries of each slice do not specifically refer to a fixed boundary. For example, when a slice changes, its structural boundaries may also change accordingly. For example, when the method of identifying structural boundaries changes, the structural boundaries of a slice may also change accordingly.

[0041] According to some embodiments, Figure 3 This is an example schematic diagram of the cross-sectional structure of the macula and optic disc provided in an embodiment of this disclosure, such as... Figure 3 As shown, the macula and optic disc are two key functional structures in the posterior pole of the eyeball, possessing unique histological characteristics and physiological functions. The central fovea of ​​the macula is rich in high-density cone cells, responsible for fine vision and color perception. The optic disc, also known as the optic nerve head, is where the axons of retinal ganglion cells converge to form the optic nerve, which exits the eyeball. As a nerve conduction hub, the optic disc is responsible for transmitting visual information to the brain.

[0042] According to some embodiments, Figure 4 This is an example schematic diagram of a structural boundary provided in an embodiment of this disclosure, such as... Figure 4As shown, the hierarchical structure refers to multiple tissue layers distributed axially in the posterior segment of the eye, including one or more of the following structures: nerve fiber layer, ganglion cell layer + inner plexiform layer, nuclear layer, outer plexiform layer, outer nuclear layer + external membrane + myoid band, ellipsoid band, lower boundary of ellipsoid band - pigment epithelium layer, pigment epithelium layer, Bruch's membrane, choroid, and sclera. The boundary of the hierarchical structure refers to the boundary of the above tissue layers. The optic disc structure refers to the area where nerve fibers converge on the retina and exit the eyeball—appearing as an elliptical disc-shaped structure on the Enface image and as a hierarchical structural break on the B-scan image. The boundary of the optic disc structure refers to the left and right boundaries of the hierarchical structural break on the B-scan image, which are mapped from the outline of the disc-shaped area on the Enface image. The optic disc structure can be, for example, as shown in... Figure 4 As shown by the white marking line in the image.

[0043] In some embodiments, the structural boundaries of each slice in the sampled slice set can be output based on the sampled slice set.

[0044] In step S14, the structural boundary information of each slice is mapped onto the three-dimensional OCT image to form the three-dimensional OCT volume data structure boundary.

[0045] In some embodiments, the boundary of the three-dimensional OCT volume data structure can be, for example, the structural boundary corresponding to the acquired three-dimensional OCT image. This three-dimensional OCT volume data structure boundary does not specifically refer to a fixed structural boundary. For example, when the three-dimensional OCT image changes, the boundary of the three-dimensional OCT volume data structure can also change accordingly.

[0046] According to some embodiments, the structural boundary information of each slice can be mapped onto a 3D OCT image to form the 3D OCT volume data structure boundary. Specifically, for example, the structural boundaries of each B-scan image in the 3D OCT image can be obtained through the structural boundary information of each slice, thereby obtaining the structural boundaries of all B-scan images and thus obtaining the 3D OCT volume data structure boundary.

[0047] According to some embodiments, mapping the structural boundary information of each slice to a three-dimensional OCT image may include, for example, switching each slice containing the structural boundary to an image slice corresponding to a threshold of the three-dimensional OCT image, or simply adding the boundary information to the corresponding image slice of the three-dimensional OCT image.

[0048] In some or related embodiments, a three-dimensional OCT image of the area to be scanned is acquired. The OCT image includes at least N frames of structurally continuous B-scan images, where N is a positive integer. Sampling is performed along the B-scan scanning direction of the three-dimensional OCT image to form a set of sampled slices. Based on the set of sampled slices, the structural boundaries of each slice in the set are output. The information of the structural boundaries of each slice is mapped onto the three-dimensional OCT image to form the three-dimensional OCT volume data structure boundary. Therefore, fundus image structure recognition can be performed with only one sampling, eliminating the need for two rounds of sampling for coarse localization and fine segmentation. This reduces the number of samplings, thereby reducing the complexity of fundus image structure recognition and the complexity of the processing steps. Furthermore, it eliminates the need to restrict the acquired three-dimensional OCT image, meaning the acquired three-dimensional OCT image is independent of the imaging center. This reduces the situation where only eye OCT images with the optic disc as the imaging center can be processed, reducing the possibility of image limitations preventing fundus image structure recognition. It also reduces the situation where using a star-shaped method to synthesize slices in both rounds of sampling results in poor recognition robustness. This improves both the accuracy and efficiency of fundus image structure recognition.

[0049] Figure 5 This is a flowchart of the second fundus image structure recognition method provided in the embodiments of this disclosure, such as... Figure 5 As shown, this fundus image structure recognition method can be used in the diagnosis and research of ophthalmic diseases, and includes the following steps: In step S21, a three-dimensional OCT image of the area to be scanned is acquired. The OCT image includes at least N frames of structurally continuous B-scan images, where N is a positive integer. The relevant processes are as described above and will not be repeated here.

[0050] In step S22, jump information during the acquisition of the three-dimensional OCT image is obtained based on the three-dimensional OCT image or the current eye image; According to some embodiments, the current eye image may be, for example, the eye image acquired during the execution of the fundus image structure recognition method. This current eye image may also be referred to as a real-time eye image. The current eye image does not specifically refer to a fixed image. For example, if the acquisition time point or acquisition device of the current eye image changes, the current eye image may also change accordingly. The current eye image may, for example, be different from a three-dimensional OCT image; specifically, it may be acquired using a scanning method different from OCT scanning. The acquisition device for the current eye image may, for example, have a light source different from the OCT scanning method, and the acquisition device may, for example, be coupled to an OCT scanning system. The three-dimensional OCT image may, for example, be a cross-sectional view, a section view, or a combination of cross-sectional and section views, and the current eye image may, for example, be a frontal view of the eye.

[0051] In some embodiments, the jump information may be jump information corresponding to each frame of B-scan slice. This jump information may also include jump parameters. These jump parameters can be read by the lens and can be used to determine whether a large jump occurs in the scanned area during the scanning of that frame of B-scan slice, resulting in a large jump in the measurement position. The jump parameter may be set to 0 or 1, where 0 may indicate the presence of a jump and 1 may indicate the absence of a jump; or 1 may indicate the presence of a jump and 0 may indicate the absence of a jump. This disclosure does not limit this aspect.

[0052] In step S23, based on the jump information, the B-scan images corresponding to all frames before and after the jump are determined in at least N structurally continuous B-scan images; According to some embodiments, the transition information can be used to determine the B-scan images corresponding to all frames before and after all transitions in at least N structurally continuous B-scan images.

[0053] For example, a transition frame is defined as the B-scan images corresponding to all frames before and after a transition within a B-scan image structure of at least N consecutive frames. These corresponding B-scan images do not refer to a specific fixed image. For instance, when the transition information changes, these corresponding B-scan images can also change accordingly.

[0054] In step S24, along the B-scan scanning direction of the three-dimensional OCT image, a portion of the B-scan image is extracted at equal intervals. In the case of jump information, at least N frames of structurally continuous B-scan images are extracted, corresponding to all frames before and after the jump, to form a set of sampling slices. The relevant descriptions can be as described above, and will not be repeated here.

[0055] According to some embodiments, the equally spaced extraction method can be, for example, equally spaced sampling. For example, it can be along the B-scan scanning direction of the 3D OCT image, extracting sampling slices at equally spaced intervals within at least N structurally continuous B-scan images. The sampling interval can be, for example, [3, 5]. Specifically, for example, sampling can be performed once every 3 B-scan images. For example, if N is 10, the at least N structurally continuous B-scan images can be, for example, 10 structurally continuous B-scan images, the sampling interval can be, for example, 3 frames, and the extracted B-scan images can be, for example, the 1st B-scan image, the 4th B-scan image, the 7th B-scan image, and the 10th B-scan image. The above 10 frames are only illustrative and do not represent the actual number of frames acquired.

[0056] In some embodiments, in the absence of transition information, a portion of the B-scan image extracted at equal intervals constitutes the sampled slice set.

[0057] According to some embodiments, when jump information is available, B-scan images corresponding to all frames before and after the jump are extracted from at least N structurally consecutive B-scan images. For example, in 10 structurally consecutive B-scan images, when the 3rd and 9th B-scan images are determined to be jump frames based on jump information, the 3rd and 9th B-scan images can be added to the sampling slice set. The sampling slice set is {1st B-scan image, 3rd B-scan image, 4th B-scan image, 7th B-scan image, 9th B-scan image, 10th B-scan image}.

[0058] In some embodiments, the sampled slice set includes a portion of the B-scan image obtained by sampling at equal intervals and a jump frame determined based on jump information.

[0059] According to some embodiments, before sampling, for example, an inter-frame alignment method can be obtained, and the position transformation of each B-scan image can be performed using the inter-frame alignment method to eliminate the jump information between B-scan sequences. The multi-frame B-scan images after position transformation can be directly sampled in an equally spaced manner to obtain a set of sampled slices.

[0060] In step S25, based on the sampled slice set, the structural boundaries of each slice in the sampled slice set are output; The relevant descriptions can be as described above, and will not be repeated here.

[0061] According to some embodiments, based on a set of sampled slices, the structural boundaries of each slice in the set of sampled slices are output, including: The sampled slice set is input into an image segmentation model for recognition, and the model outputs a category probability map of structural boundaries. The structural boundaries of each slice in the sampled slice set are obtained through the category probability map, probability map thresholding, and edge detection algorithms. Therefore, the structural boundaries of each slice in the sampled slice set can be obtained based on the category probability map, improving the accuracy of structural boundary acquisition and enhancing the accuracy of structural boundary acquisition in 3D OCT volumetric data.

[0062] According to some embodiments, for example, a set of sampled slices can be input into a trained image segmentation model for recognition. The output of the trained image segmentation model is the structural boundary of each slice in the sampled slice set. Specifically, this can be the category probability information of the structural boundary of each slice in the sampled slice set. This category probability information may include, for example, a category probability value and a category probability map. Based on the category probability information, the category probability threshold, and the edge detection algorithm, the structural boundary of each slice in the sampled slice set can be obtained. Therefore, the boundary coordinate position of each slice structure can be obtained, and based on the boundary coordinate position, the structural boundary of each slice can be obtained, thereby displaying the visual disk structure and different hierarchical structures.

[0063] The structural boundaries of each slice include at least the boundaries of the various layers of retinal tissue and the optic disc structure. The output structural boundaries include, but are not limited to, output boundary information and regional information of the overall structure. Output boundary information may, for example, output the dividing lines of the boundary regions. The regional information of the overall structure may be used to represent structural boundaries, for example, by using color overlays of different structural regions to indicate their boundaries.

[0064] According to some embodiments, such as Figure 6 As shown, the white area represents the optic disc structure, also known as the optic nerve disc; the optic disc structure can include the optic nerve mammary sulcus. The white area represents the boundary region of the optic disc structure. An image segmentation model can be used to identify the optic disc structure boundary and generate an En face image that meets the requirements.

[0065] In step S26, the structural boundary information of each slice is mapped onto the three-dimensional OCT image to form the three-dimensional OCT volume data structural boundary.

[0066] The relevant descriptions can be as described above, and will not be repeated here.

[0067] According to some embodiments, the structural boundary information of each slice is mapped onto a three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary, including: Based on the structural boundary information of each slice, boundary assignment processing is performed on the unsampled frames of at least N structurally continuous B-scan images to form a three-dimensional OCT volume data structure boundary. The boundary assignment processing includes shifting the boundary coordinates of each slice in at least one slice by the relative offset distance between frames to obtain the boundary coordinates of the unsampled frames, or obtaining the boundary coordinates of the unsampled frames through interpolation processing. The interpolation processing includes at least one of linear interpolation processing, piecewise quadratic interpolation processing, and cubic spline interpolation processing.

[0068] According to some embodiments, the structural boundary information of each slice is mapped onto a three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary, including: Based on the first slice subset, boundary assignment processing is performed on all unsampled frames in non-jump intervals of the three-dimensional OCT image to form the three-dimensional OCT volume data structure boundary. The first slice subset includes at least one slice whose structural boundary includes the boundary of each layer of tissue structure of the fundus. The non-jump interval is obtained by dividing at least N consecutive structural B-scan images based on jump information. The B-scan images of corresponding frames before and after the jump do not exist simultaneously in the same non-jump interval. Therefore, the structural boundary of the unsampled slice can be obtained by acquiring the sampled slice of the structural boundary, which can reduce the robustness of eye-tracking jump frames on the structural recognition results, reduce the amount of image processing, improve the accuracy of structural boundary acquisition, and improve the robustness of structural recognition.

[0069] According to some embodiments, the non-jump interval is obtained by dividing at least N structurally continuous B-scan images according to jump information, wherein the B-scan images of corresponding frames before and after the jump do not exist simultaneously in the same non-jump interval. For example, the B-scan images of corresponding frames before and after the jump can be the 3rd frame and the 4th frame, the 3rd frame B-scan image can exist in the first non-jump interval, and the 4th frame B-scan image can exist in the second non-jump interval.

[0070] According to some embodiments, for example, based on jump parameters, N frames of B-scan images of OCT volume data can be divided into M non-jump intervals. Within a single non-jump interval, the hierarchical boundaries of adjacent unsampled frames are assigned values ​​based on the hierarchical boundaries of the sampled B-scan slices of each frame.

[0071] According to some embodiments, the boundary assignment method is not limited. The boundary assignment process includes at least one of the following: shifting the boundary coordinates of each slice in at least one slice by the relative offset distance between frames as the structural boundary of the unsampled frame; and interpolation processing, which includes at least one of linear interpolation, piecewise quadratic interpolation, and cubic spline interpolation.

[0072] According to some embodiments, when boundary assignment processing includes interpolation processing, the specific interpolation processing method can be determined based on structural identification requirements, for example, piecewise quadratic interpolation processing can be selected based on smoothness requirements.

[0073] According to some embodiments, for example, due to the sudden movement of the human eyeball, the entire shooting part is displaced, so there will be an overall jump between the first part of the image frame and the second part of the image frame. In this case, the two or more adjacent frames between the two stages are the jump frames. Jump frames generally appear in pairs to indicate the boundary between the two stages.

[0074] For example, a 3D OCT image may have a total of 10 frames. If a transition occurs, resulting in two states, frames 1-4 are in state one and frames 5-10 are in state two, then the transition frame is frame 4-5.

[0075] like Figure 7 As shown, #1 and #4 can be sampled B-scan images, and the first boundary position results output by #1 and #4 are obtained respectively through image segmentation models. #2 and #3 are unsampled B-scan images, and the interpolation results of the first boundary on #2 and #3 can be obtained, for example, through linear interpolation. Finally, the boundary positions on all B-scan images can form a three-dimensional boundary surface.

[0076] For example, the B-scan images of frames 1 and 4 are sampled, and the first boundary position is obtained after image segmentation. Let (1, 100, 10) represent the vertical coordinate of the first boundary in the 100th column of the B-scan image of frame 1, which is 10. Similarly, let (4, 100, 40) represent the vertical coordinate of the boundary in the 100th column of the B-scan image of frame 4, which is 40. When obtaining the vertical coordinate of the first boundary in the 100th column of the B-scan images of frames 2 and 3, according to the principle of linear interpolation, the boundary results for frames 2 and 3 are (2, 100, 20) and (3, 100, 30), respectively.

[0077] According to some embodiments, the structural boundary information of each slice is mapped onto a three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary, including: Outlier processing is performed on the second slice subset to obtain the third slice subset, wherein the second slice subset includes at least one slice whose structural boundary includes the view disk structural boundary; Based on the third slice subset, boundary assignment processing is performed on unsampled frames of at least N structurally continuous B-scan images to form the three-dimensional OCT volume data structure boundary. Therefore, outlier processing can be performed on at least one slice of the optic disc structure boundary, improving the accuracy of the third slice subset acquisition and the accuracy of optic disc structure boundary assignment, thus improving the robustness of the volume data results.

[0078] In some embodiments, the boundary assignment process includes at least one of the following: shifting the boundary coordinates of each slice in at least one slice by the relative offset distance between frames as the structural boundary of the unsampled frame and interpolation processing, wherein the interpolation processing includes at least one of linear interpolation processing, piecewise quadratic interpolation processing and cubic spline interpolation processing.

[0079] According to some embodiments, outlier handling is performed on the second slice subset to obtain the third slice subset, including: Based on the second slice subset, multiple sets of corresponding En Face images and the identified boundaries are formed; Based on the continuity relationship of multiple sets of corresponding En Face images and the identified boundaries, the B-scan images containing outliers are filtered out to obtain the third slice subset.

[0080] According to some embodiments, the method may further include, for example, performing outlier processing after identifying and filtering out B-scan images containing outliers based on the En Face image. The filtering of B-scan images containing outliers may include, for example, mapping the visual disk structure boundaries of each slice in the second slice subset to feature positions on the En Face image, and using these feature positions as feature points or feature point sets; based on the continuity of feature points or feature point sets corresponding to different slices in the second slice subset, filtering out B-scan images containing outliers.

[0081] Specifically, for example, the attachment points or attachment regions between the Bruch's membrane and the optic disc can be identified as feature points or feature point sets. For example, the attachment points between the Bruch's membrane and the optic disc can be identified as feature points or feature point sets, or the attachment regions between the Bruch's membrane and the optic disc can be identified as feature points or feature point sets.

[0082] According to some embodiments, the visual disk region exhibits the following imaging characteristics along the scanning direction in OCT volumetric data: 1) the visual disk appears within a continuous B-scan slice interval; 2) the visual disk boundary coordinate values ​​change gradually within this interval. Therefore, outlier removal is performed on the visual disk boundary results of the second slice subset. Outlier removal can be based on statistical methods, for example, and this disclosure does not limit this approach. After removing outliers, the hierarchical boundaries of adjacent unsampled frames are assigned values ​​based on the visual disk structure boundaries of the sampled B-scan slices. The assignment methods include, but are not limited to, linear interpolation of the boundary coordinates of adjacent sampled frames.

[0083] like Figure 8 As shown, the boundary of the visual disk structure on a B-scan image is mapped to two two-dimensional coordinate points on the En face image. Let the horizontal direction of the En face image be the X direction and the vertical direction be the Y direction. In some embodiments, the visual disk structure appears as a single disk-shaped connected region on the En face image. Therefore, for a preset set of coordinate points, the coordinate points are concentrated in both the X and Y directions, i.e., the positional variance is small. If an outlier appears in either the X or Y direction, it is identified as an anomaly and is filtered out. Outlier identification methods include the standard deviation threshold method and the median absolute deviation method. After filtering out outliers, interpolation and connection processing are performed based on the remaining boundary coordinate points to obtain the three-dimensional OCT volume data structure boundary.

[0084] In some embodiments, jump information during the acquisition process of 3D OCT images is obtained based on 3D OCT images or current eye images. Based on the jump information, B-scan images corresponding to all frames before and after the jump are determined in at least N structurally continuous B-scan images. Along the B-scan scanning direction of the 3D OCT image, a portion of B-scan images is extracted at equal intervals. With jump information, B-scan images corresponding to all frames before and after the jump are extracted in at least N structurally continuous B-scan images to form a sampling slice set. Therefore, the sampling slice set can be determined by the jump information, thereby improving the accuracy of sampling slice set acquisition and the accuracy of boundary structure recognition.

[0085] A block diagram illustrating a medical device according to an exemplary embodiment. (Refer to...) Figure 9 The medical device 900 includes: a medical device, including: OCT scanning unit 901; Control and image processing unit 902; Display unit 903; The control and image processing unit 902 is used to identify the boundary information of the OCT volume data structure using the adopted method, and the display unit 903 is used to display all or part of the boundary information of the OCT volume data structure.

[0086] According to some embodiments, the display unit can, for example, display all OCT volume data structure boundary information, or it can also display partial OCT volume data structure boundary information. Specifically, for example, it can display the corresponding region information of a 3D OCT image, that is, it can display the 3D boundary information of the corresponding region. For example, it can also display the corresponding boundary information of any Enface image, that is, the 2D boundary information of the corresponding region on any plane.

[0087] A block diagram of a fundus image structure recognition device according to an exemplary embodiment is shown. (Refer to...) Figure 10 The device 1000 includes: The image acquisition unit 1001 is used to acquire a three-dimensional OCT image of the area to be scanned. The OCT image includes at least N frames of structurally continuous B-scan images, where N is a positive integer. The set forming unit 1002 is used to sample along the B-scan scanning direction of the three-dimensional OCT image to form a set of sampled slices; Boundary output unit 1003 is used to output the structural boundary of each slice in the sampled slice set based on the sampled slice set; Boundary forming unit 1004 is used to map the structural boundary information of each slice onto the three-dimensional OCT image to form the three-dimensional OCT volume data structure boundary.

[0088] According to some embodiments, the assembly forming unit 1002 is further configured to: Based on the 3D OCT image or the current eye image, obtain the transition information during the 3D OCT image acquisition process; Based on the transition information, determine the B-scan images corresponding to all frames before and after the transition in at least N structurally continuous B-scan images.

[0089] According to some embodiments, the set-forming unit 1002, when sampling along the B-scan scanning direction of the three-dimensional OCT image to form a set of sampled slices, is specifically used for: Along the B-scan scanning direction of the 3D OCT image, a portion of the B-scan image is extracted at equal intervals. In the case of jump information, B-scan images corresponding to all frames before and after all jumps are extracted from at least N structurally continuous B-scan images to form a sampling slice set.

[0090] According to some embodiments, the boundary forming unit 1004 is used to map the structural boundary information of each slice onto a three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary, specifically for: Based on the first slice subset, boundary assignment processing is performed on all unsampled frames in non-jump intervals of the three-dimensional OCT image to form the three-dimensional OCT volume data structure boundary. The first slice subset includes at least one slice whose structural boundary includes the boundary of each layer of tissue structure of the fundus. The non-jump interval is obtained by dividing at least N frames of structurally continuous B-scan images according to the jump information. The B-scan images of the corresponding frames before and after the jump do not exist in the same non-jump interval at the same time.

[0091] According to some embodiments, the boundary forming unit 1004 is used to map the structural boundary information of each slice onto a three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary, specifically for: Outlier processing is performed on the second slice subset to obtain the third slice subset, wherein the second slice subset includes at least one slice whose structural boundary includes the view disk structural boundary; Based on the third slice subset, boundary assignment processing is performed on the unsampled frames of at least N structurally continuous B-scan images to form the boundary of the three-dimensional OCT volume data structure.

[0092] According to some embodiments, the boundary forming unit 1004 is used to perform outlier processing on the second slice subset, and when obtaining the third slice subset, it is specifically used for: Based on the second slice subset, multiple sets of corresponding En Face images and the identified boundaries are formed; Based on the continuity relationship of multiple sets of corresponding En Face images and the identified boundaries, the B-scan images containing outliers are filtered out to obtain the third slice subset.

[0093] According to some embodiments, the boundary forming unit 1004 is used to map the structural boundary information of each slice onto a three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary, specifically for: Based on the structural boundary information of each slice, boundary assignment processing is performed on the unsampled frames of at least N structurally continuous B-scan images to form a three-dimensional OCT volume data structure boundary. The boundary assignment processing includes shifting the boundary coordinates of each slice in at least one slice by the relative offset distance between frames to obtain the boundary coordinates of the unsampled frames, or obtaining the boundary coordinates of the unsampled frames through interpolation processing. The interpolation processing includes at least one of linear interpolation processing, piecewise quadratic interpolation processing, and cubic spline interpolation processing.

[0094] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0095] In some or related embodiments, an image acquisition unit is used to acquire a three-dimensional OCT image of the area to be scanned. The OCT image includes at least N frames of structurally continuous B-scan images, where N is a positive integer. A set formation unit is used to sample along the B-scan scanning direction of the three-dimensional OCT image to form a set of sampled slices. A boundary output unit is used to output the structural boundaries of each slice in the sampled slice set based on the sampled slice set. A boundary formation unit is used to map the information of the structural boundaries of each slice onto the three-dimensional OCT image to form the three-dimensional OCT volume data structure boundary. Therefore, fundus image structure recognition can be performed with only one sampling, eliminating the need for two rounds of sampling for coarse localization and fine segmentation. This reduces the number of samplings, thereby reducing the complexity of fundus image structure recognition. Furthermore, it eliminates the need to restrict the acquired three-dimensional OCT image, reducing the possibility of only being able to process eye OCT images centered on the optic disc. This also reduces the possibility of image limitations preventing fundus image structure recognition and reduces the possibility of poor robustness due to using a star-shaped method to synthesize slices in both rounds of sampling. This improves both the accuracy and efficiency of fundus image structure recognition.

[0096] Figure 11 A schematic block diagram of an example electronic device 1100 that can be used to implement embodiments of the present disclosure is shown. The electronic device 1100 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0097] like Figure 11 As shown, the electronic device 1100 includes a computing unit 1101, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1102 or a computer program loaded from a storage unit 1108 into a random access memory (RAM) 1103. The RAM 1103 may also store various programs and data required for the operation of the electronic device 1100. The computing unit 1101, ROM 1102, and RAM 1103 are interconnected via a bus 1104. An input / output (I / O) interface 1105 is also connected to the bus 1104.

[0098] Multiple components in electronic device 1100 are connected to I / O interface 1105, including: input unit 1106, such as keyboard, mouse, etc.; output unit 1107, such as various types of displays, speakers, etc.; storage unit 1108, such as disk, optical disk, etc.; and communication unit 1109, such as network card, modem, wireless transceiver, etc. Communication unit 1109 allows electronic device 1100 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0099] The computing unit 1101 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 1101 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 1101 performs the various methods and processes described above. For example, in some embodiments, the above methods can be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 1108. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 1100 via ROM 1102 and / or communication unit 1109. When the computer program is loaded into RAM 1103 and executed by the computing unit 1101, one or more steps of the methods described above can be performed. Alternatively, in other embodiments, the computing unit 1101 can be configured to perform the above methods by any other suitable means (e.g., by means of firmware).

[0100] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0101] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0102] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0103] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0104] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), the Internet, and blockchain networks.

[0105] Computer systems can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is established by computer programs running on the respective computers and having a client-server relationship with each other. A server can be a cloud server, also known as a cloud computing server or cloud host, a hosting product within the cloud computing service system that addresses the shortcomings of traditional physical hosts and VPS (Virtual Private Server, or simply "VPS") services, such as high management difficulty and weak business scalability. Servers can also be servers for distributed systems or servers incorporating blockchain technology.

[0106] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0107] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for recognizing structures in fundus images, characterized in that, include: Acquire a three-dimensional OCT image of the area to be scanned, wherein the OCT image includes at least N structurally continuous B-scan images, where N is a positive integer; Samples are taken along the B-scan scanning direction of the three-dimensional OCT image to form a set of sampled slices; Based on the sampled slice set, output the structural boundary of each slice in the sampled slice set; The structural boundary information of each slice is mapped onto the three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary. The step of mapping the structural boundary information of each slice onto the three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary includes: Based on the first slice subset, boundary assignment processing is performed on all unsampled frames within the non-jump intervals of the three-dimensional OCT image to form the three-dimensional OCT volume data structure boundary. The first slice subset includes at least one slice whose structural boundary includes the boundary of each layer of tissue structure in the fundus. The non-jump interval is obtained by dividing at least N structurally continuous B-scan images according to the jump information. The B-scan images of the corresponding frames before and after the jump do not exist in the same non-jump interval at the same time.

2. The method according to claim 1, characterized in that, The method further includes: Based on the three-dimensional OCT image or the current eye image, obtain the transition information during the acquisition process of the three-dimensional OCT image; Based on the transition information, determine the B-scan images corresponding to all frames before and after the transition in the at least N frame structurally continuous B-scan images.

3. The method according to claim 2, characterized in that, The sampling along the B-scan scanning direction of the three-dimensional OCT image to form a set of sampling slices includes: Along the B-scan scanning direction of the three-dimensional OCT image, a portion of the B-scan image is extracted at equal intervals. With the jump information present, B-scan images corresponding to all frames before and after the jump are extracted from the B-scan images of at least N consecutive frames to form the sampling slice set.

4. The method according to claim 1, characterized in that, The step of outputting the structural boundaries of each slice in the sampled slice set includes: The sampled slice set is input into an image segmentation model for recognition, and the class probability map of the structural boundary is output. The structural boundary of each slice in the sampled slice set is obtained through the class probability map, probability map thresholding, and edge detection algorithm.

5. The method according to claim 1, characterized in that, The step of mapping the structural boundary information of each slice onto the three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary also includes: Outlier processing is performed on the second slice subset to obtain the third slice subset, wherein the second slice subset includes at least one slice whose structural boundary includes the view disk structural boundary; Based on the third slice subset, the unsampled frames of the at least N frame structurally continuous B-scan images are subjected to boundary assignment processing to form a three-dimensional OCT volume data structure boundary.

6. The method according to claim 5, characterized in that, The step of performing outlier processing on the second slice subset to obtain the third slice subset includes: Based on the second slice subset, multiple sets of corresponding En Face images and the identified boundaries are formed; Based on the continuity relationship of the multiple sets of corresponding En Face images and the identified boundaries, the B-scan images containing outliers are filtered out to obtain the third slice subset.

7. The method according to any one of claims 1 to 6, characterized in that, The step of mapping the structural boundary information of each slice onto the three-dimensional OCT image to form a three-dimensional OCT volume data structure boundary includes: Based on the structural boundary information of each slice, boundary assignment processing is performed on the unsampled frames of the at least N structurally continuous B-scan images to form a three-dimensional OCT volume data structure boundary. The boundary assignment processing includes shifting the boundary coordinates of each slice in the at least one slice by the relative offset distance between frames to obtain the boundary coordinates of the unsampled frames, or obtaining the boundary coordinates of the unsampled frames through interpolation processing. The interpolation processing includes at least one of linear interpolation processing, piecewise quadratic interpolation processing, and cubic spline interpolation processing.

8. A medical device, characterized in that, include: OCT scanning unit; Control and image processing unit; Display unit; The control and image processing unit is used to identify the boundary information of the OCT volume data structure using the method of any one of claims 1 to 7, and the display unit is used to display all or part of the boundary information of the OCT volume data structure.

9. A storage medium storing instructions, characterized in that, When the instructions are executed on an electronic device, the electronic device causes the electronic device to perform the fundus image structure recognition method as described in any one of claims 1 to 7.

Images