Anterior segment parameter measurement method and device, electronic device, and storage medium

Abstract

The present disclosure relates to the field of computer vision and ophthalmology, and provides an anterior segment parameter measurement method and device, electronic device, and storage medium. The measurement method includes: based on a generative adversarial network, the panoramic AS-OCT image is repaired to eliminate the central axis and artifacts in the panoramic AS-OCT image; based on an image segmentation model, the anterior segment structures in the repaired panoramic AS-OCT image are segmented to obtain a corresponding segmented image; using a preset positioning model, the key structure of the repaired panoramic AS-OCT image is positioned to obtain a corresponding positioning result; based on the segmented image and the positioning result, the anterior segment parameters involved in the panoramic AS-OCT image are measured to obtain the corresponding biological parameter measurement results. The present disclosure can greatly improve the automation, accuracy, and efficiency of the measurement of anterior segment biological parameters in AS-OCT images, and provide strong support for the diagnosis and treatment of ophthalmic diseases.

Description

Technical Field

[0001] The present disclosure relates to the fields of computer vision and ophthalmic medical technology, and in particular to an anterior segment parameter measurement method and device, electronic equipment, and storage medium. Background Art

[0002] The anterior segment refers to the front half of the eyeball, including the cornea, anterior chamber, posterior chamber, chamber angle, lens suspensory ligament and lens. These structures work together to achieve visual accommodation of the eye and are the basis of normal visual function. Problems with the anterior segment structure can cause different ophthalmic diseases and hazards. For example, a smaller anterior chamber angle can cause excessive aqueous humor to squeeze the sclera backwards, causing irreversible visual damage such as glaucoma and cataracts. Ciliary body problems may lead to lens dysfunction and affect visual adaptation. Therefore, accurately distinguishing the various parts and structures in the anterior segment image is crucial for accurately diagnosing ophthalmic diseases and formulating and implementing reasonable treatment plans.

[0003] Anterior Segment Optical Coherence Tomography (AS-OCT) is an important tool for diagnosing ophthalmic diseases. It was first introduced by Laztt et al. in 1994. Its initial working wavelength was 830 nanometers, similar to retinal OCT. However, in order to overcome the limitation of tissue penetration, especially in the sclera, AS-OCT later adopted a wavelength of 1310 nanometers, which improved the sclera penetration and enabled real-time imaging at 8 frames per second. This tomography technology is a non-invasive ophthalmic imaging technology that ensures patient comfort and is not affected by radiation. It can observe ocular structures such as the cornea, anterior chamber, iris, and lens, and provide valuable information for the diagnosis and treatment of ocular diseases, such as: providing lens shape, density, and position information for surgery, which helps to determine the type of cataract; providing corneal thickness, morphology and other data to assist in the diagnosis of corneal diseases; providing anterior chamber angle and iris morphology information to assist in the planning of posterior chamber intraocular lens implantation and glaucoma diagnosis in phakic eyes, etc. At present, the AS-OCT image data obtained from hospitals have obvious central axes and artifacts, which easily interfere with segmentation and affect the segmentation effect.

[0004] Since AS-OCT images have some inherent defects during imaging, such as the vertical line of the central axis, which will have a great impact on the segmentation and measurement operations, some image enhancement methods are needed to make the subsequent segmentation and measurement more convenient. For some ophthalmic modality images, some image enhancement methods have been used.

[0005] For example, the prior art proposes a fundus color image enhancement method based on the Retinex theory, analyzes the problems existing in the existing Retinex image enhancement method, and proposes an improved fundus color image enhancement method based on the Retinex theory. This method has better effects than other image processing methods in terms of protecting image color, optimizing vascular contrast, and improving image details. Objective rating indicators show that this method has the advantage of color fidelity over existing fundus image enhancement methods. The prior art also obtains enhanced Retinex fundus images by conducting experiments on fundus images based on traditional Retinex methods, which makes the image more prominent in terms of edge preservation and detail optimization, and also has a positive effect on the processing of redundant fundus vascular details and lesion details. However, in the above method, the color distortion of the fundus image is serious, the amount of data required for calculation is large, the running speed is relatively slow, and there are defects of time-consuming. Medical image analysis is a branch of the intersection of image processing technology and biomedical engineering. Traditional machine learning algorithms can be roughly divided into three parts: image preprocessing, image recognition, and image feature extraction when processing 2D medical images, and medical image features need to be manually extracted, resulting in low recognition accuracy.

[0006] Since the images under the AS-OCT modality contain rich anterior segment structural information, the prior art has also actively explored methods under this modality, and the measurement of biological parameters under the AS-OCT image modality has also made certain progress. For example, the prior art provides a semi-automatic angle evaluation program to calculate various anterior chamber angle parameters. For the fully automatic AS-OCT anterior segment biological parameter measurement system, the prior art also proposes a high-definition OCT parameter calculation method based on Schwalbe line detection. The prior art also proposes a label migration system that combines the segmentation, measurement and detection of AS-OCT structures, restores the main anterior chamber angle parameters based on the anterior segment structure, and uses it as a feature for detecting anterior chamber angle closure. However, the above methods are usually inefficient and often still require manual operation. Summary of the invention

[0007] The present disclosure aims to solve at least one of the problems existing in the prior art and provides an anterior segment parameter measurement method and device, an electronic device, and a storage medium.

[0008] In one aspect of the present disclosure, a method for measuring anterior segment parameters is provided, the method comprising:

[0009] Based on a generative adversarial network, the panoramic AS-OCT image is repaired to eliminate the central axis and artifacts in the panoramic AS-OCT image;

[0010] Based on the image segmentation model, the structures of the anterior segment of the eye in the restored panoramic AS-OCT image are segmented to obtain a corresponding segmented image;

[0011] Using a preset positioning model, key structure positioning is performed on the restored panoramic AS-OCT image to obtain a corresponding positioning result;

[0012] Based on the segmented image and the positioning result, the anterior segment parameters involved in the panoramic AS-OCT image are measured to obtain corresponding biological parameter measurement results.

[0013] Optionally, the repairing of the panoramic AS-OCT image based on the generative adversarial network to eliminate the central axis and artifacts in the panoramic AS-OCT image includes:

[0014] Using specific texture features of the mid-axis defect in the panoramic AS-OCT image, generating a corresponding mask;

[0015] Taking the mask and its corresponding mask image as input, roughly repairing the panoramic AS-OCT image using a generative model based on the generative adversarial network consisting of a convolution head, a Transformer block, and a convolution tail;

[0016] The Conv-U-Net neural network is used to refine and repair the roughly repaired panoramic AS-OCT image.

[0017] Optionally, segmenting the anterior segment structures in the restored panoramic AS-OCT image based on the image segmentation model to obtain a corresponding segmented image includes:

[0018] Using SAM and cue points, segmenting the iris structure and corneal structure in the restored panoramic AS-OCT image;

[0019] Obtaining initial segmentation annotations corresponding to the anterior chamber structure and the lens structure in the restored panoramic AS-OCT image through the SAM and the prompt points, using the initial segmentation annotations as supervision data of the SAM, fine-tuning the SAM, and using the fine-tuned SAM to segment the anterior chamber structure and the lens structure in the restored panoramic AS-OCT image;

[0020] Wherein, when fine-tuning the SAM, a serial adapter method is used to adjust the image encoder of the SAM.

[0021] Optionally, the using of a preset positioning model to locate key structures of the restored panoramic AS-OCT image to obtain corresponding positioning results includes:

[0022] The YOLOv8 model is used to detect and locate the scleral protrusion in the repaired panoramic AS-OCT image to obtain the coordinates corresponding to the scleral protrusion.

[0023] Optionally, based on the segmented image and the positioning result, measuring the anterior segment parameters involved in the panoramic AS-OCT image to obtain corresponding biological parameter measurement results includes:

[0024] Using the segmented image and the coordinates corresponding to the scleral protrusion, applying the Canny edge detection algorithm to obtain edge information of each segmented area in the segmented image;

[0025] Based on the edge information of each segmented area and the definition of the anterior segment parameter measurement, the key point coordinates are identified; the key point coordinates include the scleral protrusion, the highest point of the anterior chamber structure, the highest point of the corneal structure, the end point of the iris structure, and the highest point of the lens structure;

[0026] Based on the key point coordinates, biological parameter measurement results corresponding to the anterior segment parameters involved in the panoramic AS-OCT image are determined.

[0027] Optionally, the identifying key point coordinates based on the edge information of each segmented area and the definition of the anterior segment parameter measurement includes:

[0028] Connecting the scleral protrusions located on both sides of the image and belonging to the same eyeball to obtain corresponding scleral protrusion connection lines;

[0029] Obtaining the perpendicular bisector of the scleral process connection line as the central midline of the anterior segment;

[0030] The highest point and the lowest point of the corneal structure, the highest point of the anterior chamber structure, and the highest point and the lowest point of the lens structure are determined respectively by using the central midline and the segmented areas corresponding to the corneal structure, the anterior chamber structure, and the lens structure.

[0031] Optionally, the anterior segment parameters involved in the panoramic AS-OCT image include central corneal thickness, anterior chamber depth, anterior chamber width, lens thickness, anterior chamber area, iris area, and iris endpoint width;

[0032] Determining the biological parameter measurement result corresponding to the anterior segment parameter involved in the panoramic AS-OCT image based on the key point coordinates includes:

[0033] Calculating the height difference between the highest point of the corneal structure and the lowest point of the corneal structure to obtain the central corneal thickness;

[0034] Calculating the length of the central midline between the lowest point of the corneal structure and the highest point of the lens structure to obtain the anterior chamber depth;

[0035] Calculate the length of the scleral protrusion connection line to obtain the anterior chamber width;

[0036] Calculating the height difference between the highest point of the lens structure and the lowest point of the lens structure to obtain the lens thickness;

[0037] Calculating the area of ​​the segmented region corresponding to the anterior chamber structure to obtain the anterior chamber area;

[0038] Calculating the area of ​​the segmented region corresponding to the iris structure to obtain the iris area;

[0039] The endpoints on both sides of the iris structure are determined using a line parallel to the scleral prominence connection line, and the length of a line connecting the endpoints on both sides of the iris structure is calculated to obtain the iris endpoint width.

[0040] Another aspect of the present disclosure provides an anterior segment parameter measuring device, the measuring device comprising:

[0041] An image restoration module, used to restore the panoramic AS-OCT image based on a generative adversarial network, and eliminate the central axis and artifacts in the panoramic AS-OCT image;

[0042] A key structure segmentation module, used to segment the anterior segment structures in the restored panoramic AS-OCT image based on an image segmentation model to obtain a corresponding segmented image;

[0043] A key structure positioning module is used to use a preset positioning model to perform key structure positioning on the restored panoramic AS-OCT image to obtain a corresponding positioning result;

[0044] The biological parameter measurement module is used to measure the anterior segment parameters involved in the panoramic AS-OCT image based on the segmented image and the positioning result to obtain the corresponding biological parameter measurement results.

[0045] Another aspect of the present disclosure provides an electronic device, including:

[0046] at least one processor; and,

[0047] a memory communicatively connected to at least one processor; wherein,

[0048] The memory stores instructions that can be executed by at least one processor. The instructions are executed by the at least one processor so that the at least one processor can execute the anterior segment parameter measurement method described above.

[0049] Another aspect of the present disclosure provides a computer-readable storage medium storing a computer program, which implements the anterior segment parameter measurement method described above when executed by a processor.

[0050] Compared with the prior art, the present invention realizes the automated measurement of anterior segment parameters, avoids the tedious process of manual measurement, reduces human errors, and significantly improves measurement efficiency and accuracy; it also realizes the comprehensive measurement of anterior segment parameters, and provides comprehensive and accurate data support for ophthalmic clinical diagnosis and treatment; it has a high degree of automation and can quickly process a large number of panoramic AS-OCT images, which is suitable for large-scale clinical screening and diagnosis, and improves the work efficiency of ophthalmic medical treatment. At the same time, accurate measurement of anterior segment parameters helps to detect and diagnose ophthalmic diseases at an early stage, improves the treatment effect of patients, and has significant clinical application value and social benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] One or more embodiments are exemplarily described by pictures in the corresponding drawings, and these exemplifications do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0052] Figure 1 A flow chart of a method for measuring anterior segment parameters provided by one embodiment of the present disclosure;

[0053] Figure 2 A flowchart of image restoration provided by another embodiment of the present disclosure;

[0054] Figure 3 A flowchart of key structure segmentation provided for another embodiment of the present disclosure;

[0055] Figure 4 A flowchart of key structure positioning provided for another embodiment of the present disclosure;

[0056] Figure 5 A diagram showing the definition of some biological parameters provided in another embodiment of the present disclosure;

[0057] Figure 6 A schematic diagram of the overall framework of an anterior segment parameter measurement device provided by another embodiment of the present disclosure;

[0058] Figure 7 A schematic structural diagram of an electronic device provided in another embodiment of the present disclosure. DETAILED DESCRIPTION

[0059] At present, there are often obvious central axes and artifacts in the images obtained by AS-OCT technology. These interferences will affect the image segmentation effect, thereby reducing the accuracy of biological parameter measurement. The existing manual measurement methods for anterior segment parameters are not only time-consuming and labor-intensive, but also have subjective errors and are difficult to meet clinical needs. Although the prior art also provides some automatic measurement methods for anterior segment parameters, these automatic measurement methods still have deficiencies in processing image enhancement, key structure segmentation and parameter measurement, resulting in low accuracy and efficiency in biological parameter measurement. In addition, when faced with complex anterior segment structures, the prior art often requires manual intervention in the biological parameter measurement process, and fully automated measurement cannot be achieved.

[0060] In view of the above problems, an embodiment of the present disclosure proposes an anterior segment parameter measurement method for automatically measuring the anterior segment parameters in AS-OCT images.

[0061] In order to make the purpose, technical scheme and advantages of the embodiments of the present disclosure clearer, the embodiments of the present disclosure will be described in detail below in conjunction with the accompanying drawings. However, it can be understood by those skilled in the art that in each embodiment of the present disclosure, many technical details are proposed in order to enable readers to better understand the present disclosure. However, even without these technical details and various changes and modifications based on the following embodiments, the technical scheme claimed for protection in the present disclosure can also be implemented. The division of the following embodiments is for the convenience of description and should not constitute any limitation on the specific implementation of the present disclosure. The various embodiments can be combined and referenced with each other without contradiction.

[0062] One embodiment of the present disclosure relates to a method for measuring anterior segment parameters, the process of which is as follows: Figure 1 As shown, including:

[0063] Step S110: Based on the generative adversarial network, the panoramic AS-OCT image is repaired to eliminate the central axis and artifacts in the panoramic AS-OCT image.

[0064] Specifically, step S110 is mainly used for image enhancement. By introducing a generative adversarial network (GAN) to repair the panoramic AS-OCT image, the central axis and artifacts in the panoramic AS-OCT image can be effectively eliminated, thereby effectively improving the image clarity and quality and reducing the interference of noise on subsequent image segmentation and parameter measurement.

[0065] Step S120 , based on the image segmentation model, segment the anterior segment structures in the restored panoramic AS-OCT image to obtain a corresponding segmented image.

[0066] Specifically, step S120 is mainly used for structural segmentation. Step S120 can fine-tune an image segmentation model such as a Segment Anything Model (SAM), and use the fine-tuned image segmentation model to accurately segment key structures such as the anterior segment of the eye, such as the iris structure, corneal structure, anterior chamber structure, and lens structure in the restored panoramic AS-OCT image, so as to improve the image segmentation accuracy and effectively solve the problem of low segmentation accuracy of traditional image segmentation methods.

[0067] Step S130, using a preset positioning model, positioning key structures of the restored panoramic AS-OCT image to obtain corresponding positioning results.

[0068] Specifically, step S130 is mainly used for structural positioning. Step S130 can select the preset positioning model as the YOLOv8 model, and use the YOLOv8 model to efficiently and accurately locate the key structures in the AS-OCT image to improve the recognition accuracy and speed of the key structures, provide a reliable basis for subsequent biological parameter measurement, and thus improve the accuracy of biological parameter measurement.

[0069] Step S140, based on the segmented image and the positioning result, the anterior segment parameters involved in the panoramic AS-OCT image are measured to obtain corresponding biological parameter measurement results.

[0070] Specifically, step S140 is mainly used to measure the anterior segment parameters of the eye. It can automatically measure the anterior segment parameters of the panoramic AS-OCT image, such as scleral spur (SS), anterior chamber depth (ACD), central corneal thickness (CCT), anterior chamber width (ACW), anterior chamber area (ACArea), iris endpoint width (IEW), iris area (I-Area), lens thickness (LT), etc., so as to reduce manual operation and improve the efficiency and accuracy of parameter measurement.

[0071] Compared with the prior art, the anterior segment parameter measurement method provided in the embodiment of the present invention realizes the automatic measurement of the anterior segment parameters, avoids the tedious process of manual measurement, reduces human errors, and significantly improves the measurement efficiency and accuracy; it also realizes the comprehensive measurement of the anterior segment parameters, and provides comprehensive and accurate data support for ophthalmic clinical diagnosis and treatment; it has a high degree of automation and can quickly process a large number of panoramic AS-OCT images, which is suitable for large-scale clinical screening and diagnosis, and improves the work efficiency of ophthalmic medical treatment. At the same time, accurate measurement of the anterior segment parameters helps to detect and diagnose ophthalmic diseases at an early stage, improves the treatment effect of patients, and has significant clinical application value and social benefits.

[0072] Exemplarily, step S110 includes: using specific texture features for the central axis defects in the panoramic AS-OCT image, generating a corresponding mask to make preliminary preparations for image restoration. The mask and its corresponding mask image are used as input, and the panoramic AS-OCT image is roughly restored using a generative model based on a generative adversarial network consisting of a convolution head, a Transformer block, and a convolution tail. The roughly restored panoramic AS-OCT image is refined using a Conv-U-Net neural network to optimize high-frequency details and improve the overall image quality.

[0073] Specifically, combined with Figure 2 In step S110, the area to be repaired in the panoramic AS-OCT image can be masked using a mask generated according to the specific texture features of the central axis defect, and then the corresponding unobstructed AS-OCT image is generated using the GAN-based generation model, thereby further effectively eliminating the central axis and artifacts in the panoramic AS-OCT image, improving the clarity and quality of the image, and reducing the interference of noise on subsequent segmentation and measurement.

[0074] Exemplarily, step S120 includes: using SAM and prompt points to segment the iris structure and corneal structure in the restored panoramic AS-OCT image. Using SAM and prompt points, initial segmentation annotations corresponding to the anterior chamber structure and lens structure in the restored panoramic AS-OCT image are obtained, the initial segmentation annotations are used as supervision data of SAM, SAM is fine-tuned, and the anterior chamber structure and lens structure in the restored panoramic AS-OCT image are segmented using the fine-tuned SAM. When fine-tuning SAM, a serial adapter method is used to adjust the image encoder of SAM.

[0075] Specifically, due to the variability of the iris and the fuzziness of the corneal boundary, step S120 may use SAM and inherent cue points to perform preliminary segmentation of the iris and cornea.

[0076] For the anterior chamber and lens, initial segmentation annotations are obtained through SAM and cue points and used as supervision data for SAM. SAM is fine-tuned using the serial adapter method to adjust the image encoder of SAM, thereby using the fine-tuned SAM to achieve autonomous segmentation of the anterior chamber and lens without cueing.

[0077] Combined Figure 3In step S120, the unobstructed AS-OCT image is segmented by using the fine-tuned SAM model to obtain the corresponding key structure segmentation information, thereby further realizing the accurate segmentation of key structures such as the iris structure, corneal structure, anterior chamber structure, and lens structure in the unobstructed AS-OCT image, improving the image segmentation accuracy, and solving the shortcomings of the traditional method in image segmentation accuracy.

[0078] Exemplarily, step S130 includes: using the YOLOv8 model to detect and locate the scleral protrusion in the repaired panoramic AS-OCT image to obtain the coordinates corresponding to the scleral protrusion.

[0079] Specifically, combined with Figure 4 In order to efficiently and accurately detect and locate the scleral protrusion, step S130 can adopt the latest version of the YOLO model, namely the YOLOv8 model, and use the YOLOv8 positioning model to efficiently and accurately locate the scleral protrusion, and identify the scleral protrusion position information in the unobstructed AS-OCT image, thereby further improving the accuracy and speed of key structure recognition and providing a reliable basis for subsequent biological parameter measurement.

[0080] Exemplarily, step S140 includes: using the coordinates corresponding to the segmented image and the scleral protrusion, applying the Canny edge detection algorithm to obtain the edge information of each segmented area in the segmented image. Based on the edge information of each segmented area and the definition of the anterior segment parameter measurement, identify the key point coordinates. The key point coordinates include the scleral protrusion, the highest point of the anterior chamber structure, the highest point of the corneal structure, the end point of the iris structure, and the highest point of the lens structure. Based on the key point coordinates, determine the biological parameter measurement results corresponding to the anterior segment parameters involved in the panoramic AS-OCT image.

[0081] This implementation can automatically identify key points and measure various anterior segment parameters, avoiding the tedious process of manual measurement, reducing human errors, and significantly improving measurement efficiency and accuracy.

[0082] Exemplary, combined Figure 5 In step S140, based on the edge information of each segmented area and the definition of the anterior segment parameter measurement, the key point coordinates are identified, including: connecting the scleral protrusions located on both sides of the image and belonging to the same eyeball to obtain the corresponding scleral protrusion connection line, i.e., the SS line. Obtain the perpendicular bisector of the scleral protrusion connection line, i.e., the SS line, as the central midline of the anterior segment. Using the central midline and the segmented areas corresponding to the corneal structure, the anterior chamber structure, and the lens structure, respectively, determine the highest point and the lowest point of the corneal structure, the highest point of the anterior chamber structure, and the highest point and the lowest point of the lens structure, respectively.

[0083] This embodiment can further automatically identify the coordinates of key points, avoid the tedious process of manual measurement, reduce human errors, and significantly improve the efficiency and accuracy of biological parameter measurement.

[0084] Exemplary, combined Figure 5 The anterior segment parameters involved in the panoramic AS-OCT image include central corneal thickness (CCT), anterior chamber depth (ACD), anterior chamber width (ACW), lens thickness (LT), anterior chamber area (ACArea), iris area (I-Area), and iris endpoint width (IEW).

[0085] In step S140, based on the key point coordinates, the biological parameter measurement results corresponding to the anterior segment parameters involved in the panoramic AS-OCT image are determined, including:

[0086] The central corneal thickness was calculated by calculating the height difference between the highest point of the corneal structure and the lowest point of the corneal structure.

[0087] The anterior chamber depth was calculated by calculating the length of the central midline between the lowest point of the corneal structure and the highest point of the lens structure.

[0088] The length of the scleral process connection line was calculated to obtain the anterior chamber width.

[0089] The height difference between the highest point of the lens structure and the lowest point of the lens structure is calculated to obtain the lens thickness.

[0090] The area of ​​the segmented region corresponding to the anterior chamber structure is calculated to obtain the anterior chamber area.

[0091] The area of ​​the segmented region corresponding to the iris structure is calculated to obtain the iris area.

[0092] The endpoints on both sides of the iris structure are determined using a line parallel to the scleral process connection line, and the length of the line connecting the endpoints on both sides of the iris structure is calculated to obtain the iris endpoint width.

[0093] This implementation can further automatically measure a variety of key anterior segment parameters, providing comprehensive and accurate data support for ophthalmic clinical diagnosis and treatment, facilitating early detection and diagnosis of ophthalmic diseases, and improving treatment outcomes for patients.

[0094] Another embodiment of the present disclosure relates to an anterior segment parameter measurement device, including an image restoration module, a key structure segmentation module, a key structure positioning module, and a biological parameter measurement module.

[0095] The image restoration module is used to restore the panoramic AS-OCT image based on the generative adversarial network and eliminate the central axis and artifacts in the panoramic AS-OCT image.

[0096] The key structure segmentation module is used to segment the anterior segment structures in the restored panoramic AS-OCT image based on the image segmentation model to obtain the corresponding segmented image.

[0097] The key structure positioning module is used to use the preset positioning model to locate the key structures of the restored panoramic AS-OCT image and obtain the corresponding positioning results.

[0098] The biological parameter measurement module is used to measure the anterior segment parameters involved in the panoramic AS-OCT image based on the segmented image and the positioning result to obtain the corresponding biological parameter measurement results.

[0099] Specifically, Figure 6 As shown, the anterior segment parameter measurement device first uses the image restoration module, i.e., the central axis vertical line restoration module, to restore the panoramic AS-OCT image, and then uses the key structure segmentation module and the key structure positioning module to segment and position the restored panoramic AS-OCT image, and finally uses the biological parameter measurement module to measure the anterior segment parameters involved in the panoramic AS-OCT image based on the image segmentation result and the positioning result, to obtain the final biological parameter measurement result.

[0100] The specific implementation method of the anterior segment parameter measurement device provided in the embodiment of the present disclosure can refer to the anterior segment parameter measurement method provided in the embodiment of the present disclosure, and will not be repeated here.

[0101] Compared with the prior art, the anterior segment parameter measurement device provided in the embodiment of the present invention realizes the automatic measurement of the anterior segment parameters, avoids the tedious process of manual measurement, reduces human errors, and significantly improves the measurement efficiency and accuracy; it also realizes the comprehensive measurement of the anterior segment parameters, and provides comprehensive and accurate data support for ophthalmic clinical diagnosis and treatment; it has a high degree of automation and can quickly process a large number of panoramic AS-OCT images, which is suitable for large-scale clinical screening and diagnosis, and improves the work efficiency of ophthalmic medical treatment. At the same time, accurate measurement of the anterior segment parameters helps to detect and diagnose ophthalmic diseases at an early stage, improve the treatment effect of patients, and has significant clinical application value and social benefits.

[0102] Another embodiment of the present disclosure relates to an electronic device, such as Figure 7 As shown, including:

[0103] at least one processor 701; and,

[0104] A memory 702 is communicatively connected to at least one processor 701; wherein,

[0105] The memory 702 stores instructions that can be executed by at least one processor 701 . The instructions are executed by at least one processor 701 so that the at least one processor 701 can execute the anterior segment parameter measurement method described in the above embodiment.

[0106] Among them, the memory and the processor are connected in a bus manner, and the bus may include any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors and memories together. The bus can also connect various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be one element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices on a transmission medium. The data processed by the processor is transmitted on a wireless medium via an antenna, and further, the antenna also receives data and transmits the data to the processor.

[0107] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory can be used to store data used by the processor when performing operations.

[0108] Another embodiment of the present disclosure relates to a computer-readable storage medium storing a computer program, which implements the anterior segment parameter measurement method described in the above embodiment when executed by a processor.

[0109] That is, those skilled in the art can understand that all or part of the steps in the method described in the above embodiments can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including several instructions for making a device (which can be a single-chip microcomputer, chip, etc.) or a processor (processor) perform all or part of the steps of the method described in each embodiment of the present disclosure. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0110] Those skilled in the art will appreciate that the above-mentioned embodiments are specific embodiments for implementing the present disclosure, and in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present disclosure.

Claims

1. A method for measuring anterior segment parameters, characterized in that: The measuring method comprises: Based on a generative adversarial network, the panoramic AS-OCT image is repaired to eliminate the central axis and artifacts in the panoramic AS-OCT image; Based on the image segmentation model, the structures of the anterior segment of the eye in the restored panoramic AS-OCT image are segmented to obtain a corresponding segmented image; Using a preset positioning model, key structure positioning is performed on the restored panoramic AS-OCT image to obtain a corresponding positioning result; Based on the segmented image and the positioning result, measuring the anterior segment parameters involved in the panoramic AS-OCT image to obtain corresponding biological parameter measurement results; The image segmentation model is used to segment the anterior segment structures in the restored panoramic AS-OCT image to obtain a corresponding segmented image, including: Using SAM and cue points, segmenting the iris structure and corneal structure in the restored panoramic AS-OCT image; Obtaining initial segmentation annotations corresponding to the anterior chamber structure and the lens structure in the restored panoramic AS-OCT image through the SAM and the prompt points, using the initial segmentation annotations as supervision data of the SAM, fine-tuning the SAM, and using the fine-tuned SAM to segment the anterior chamber structure and the lens structure in the restored panoramic AS-OCT image; Wherein, when fine-tuning the SAM, a serial adapter method is used to adjust the image encoder of the SAM.

2. The measuring method according to claim 1, characterized in that: The method of repairing the panoramic AS-OCT image based on the generative adversarial network to eliminate the central axis and artifacts in the panoramic AS-OCT image includes: Using specific texture features of the mid-axis defect in the panoramic AS-OCT image, generating a corresponding mask; Taking the mask and its corresponding mask image as input, roughly repairing the panoramic AS-OCT image using a generative model based on the generative adversarial network consisting of a convolution head, a Transformer block, and a convolution tail; The Conv-U-Net neural network is used to refine and repair the roughly repaired panoramic AS-OCT image.

3. The measuring method according to claim 1, characterized in that: The method of using the preset positioning model to locate key structures of the restored panoramic AS-OCT image to obtain corresponding positioning results includes: The YOLOv8 model is used to detect and locate the scleral protrusion in the repaired panoramic AS-OCT image to obtain the coordinates corresponding to the scleral protrusion.

4. The measuring method according to claim 3, characterized in that: The measuring of the anterior segment parameters involved in the panoramic AS-OCT image based on the segmented image and the positioning result to obtain corresponding biological parameter measurement results includes: Using the segmented image and the coordinates corresponding to the scleral protrusion, applying the Canny edge detection algorithm to obtain edge information of each segmented area in the segmented image; Based on the edge information of each segmented area and the definition of the anterior segment parameter measurement, the key point coordinates are identified; the key point coordinates include the scleral protrusion, the highest point of the anterior chamber structure, the highest point of the corneal structure, the end point of the iris structure, and the highest point of the lens structure; Based on the key point coordinates, biological parameter measurement results corresponding to the anterior segment parameters involved in the panoramic AS-OCT image are determined.

5. The measuring method according to claim 4, characterized in that: The identifying key point coordinates based on the edge information of each segmented area and the definition of the anterior segment parameter measurement includes: Connecting the scleral protrusions located on both sides of the image and belonging to the same eyeball to obtain corresponding scleral protrusion connection lines; Obtaining the perpendicular bisector of the scleral process connection line as the central midline of the anterior segment; The highest point and the lowest point of the corneal structure, the highest point of the anterior chamber structure, and the highest point and the lowest point of the lens structure are determined respectively by using the central midline and the segmented areas corresponding to the corneal structure, the anterior chamber structure, and the lens structure.

6. The measuring method according to claim 5, characterized in that: The anterior segment parameters involved in the panoramic AS-OCT image include central corneal thickness, anterior chamber depth, anterior chamber width, lens thickness, anterior chamber area, iris area, and iris endpoint width; Determining the biological parameter measurement result corresponding to the anterior segment parameter involved in the panoramic AS-OCT image based on the key point coordinates includes: Calculating the height difference between the highest point of the corneal structure and the lowest point of the corneal structure to obtain the central corneal thickness; Calculating the length of the central midline between the lowest point of the corneal structure and the highest point of the lens structure to obtain the anterior chamber depth; Calculate the length of the scleral protrusion connection line to obtain the anterior chamber width; Calculating the height difference between the highest point of the lens structure and the lowest point of the lens structure to obtain the lens thickness; Calculating the area of ​​the segmented region corresponding to the anterior chamber structure to obtain the anterior chamber area; Calculating the area of ​​the segmented region corresponding to the iris structure to obtain the iris area; The endpoints on both sides of the iris structure are determined using a line parallel to the scleral prominence connection line, and the length of a line connecting the endpoints on both sides of the iris structure is calculated to obtain the iris endpoint width.

7. An anterior segment parameter measuring device, characterized in that: The measuring device comprises: An image restoration module, used to restore the panoramic AS-OCT image based on a generative adversarial network, and eliminate the central axis and artifacts in the panoramic AS-OCT image; A key structure segmentation module, used to segment the anterior segment structures in the restored panoramic AS-OCT image based on an image segmentation model to obtain a corresponding segmented image; A key structure positioning module is used to use a preset positioning model to perform key structure positioning on the restored panoramic AS-OCT image to obtain a corresponding positioning result; A biological parameter measurement module, used to measure the anterior segment parameters involved in the panoramic AS-OCT image based on the segmented image and the positioning result, and obtain corresponding biological parameter measurement results; The image segmentation model is used to segment the anterior segment structures in the restored panoramic AS-OCT image to obtain a corresponding segmented image, including: Using SAM and cue points, segmenting the iris structure and corneal structure in the restored panoramic AS-OCT image; Obtaining initial segmentation annotations corresponding to the anterior chamber structure and the lens structure in the restored panoramic AS-OCT image through the SAM and the prompt points, using the initial segmentation annotations as supervision data of the SAM, fine-tuning the SAM, and using the fine-tuned SAM to segment the anterior chamber structure and the lens structure in the restored panoramic AS-OCT image; Wherein, when fine-tuning the SAM, a serial adapter method is used to adjust the image encoder of the SAM.

8. An electronic device, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the anterior segment parameter measurement method according to any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the anterior segment parameter measurement method according to any one of claims 1 to 6 is implemented.

Images