OCT (Optical Coherence Tomography) imaging method and cornea position prediction model training method

Abstract

The invention provides an OCT imaging method and a cornea position prediction model training method, and the method comprises the steps: inputting an obtained target B-Scan image into a pre-trained cornea position prediction model, and outputting the position information of the front surface of a target cornea in the target B-Scan image; performing curve fitting according to the position information to obtain a fitted curve; determining a target cornea vertex of the target cornea based on the curve vertex of the fitted curve; and taking the vertex of the target cornea as a scanning center, and carrying out OCT imaging on the target eye. According to the method, the position information of the front surface of the target cornea in the target B-Scan image can be quickly and accurately predicted through the cornea position prediction model, the vertex of the target cornea can be accurately determined through fitting processing, and the accuracy of OCT imaging can be improved by taking the vertex of the target cornea as a scanning center.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to an OCT imaging method and a training method for a corneal position prediction model. Background Art

[0002] Corneal thickness topography can be used for the preliminary evaluation of corneal refractive surgery and to monitor the progression of corneal diseases. In order to achieve high-precision measurement of corneal thickness, a star scanning protocol is often used. This protocol uses the corneal vertex on the anterior surface of the cornea as the center and scans several corneal slice images of the anterior segment at equal angles. The collected images are then comprehensively analyzed and calculated to generate a corneal thickness topography. Before the scan is performed, in order to calculate accurate thickness data, the corneal vertex needs to be aligned with the center of the entire scanning area. In related technologies, the position of the corneal vertex can be determined by analyzing and calculating corneal slice images based on traditional image processing algorithms. However, these methods have a long processing time and are not well adapted to the rapid and subtle movements of the eye during the acquisition process and various scanning interferences, making it difficult to accurately identify the position of the corneal vertex, and thus difficult to ensure that the corneal vertex is aligned with the center of the entire scanning area, reducing the accuracy of OCT imaging. Summary of the Invention

[0003] The purpose of the present invention is to provide an OCT imaging method and a training method for a corneal position prediction model to accurately determine the target corneal vertex, thereby improving the accuracy of OCT imaging.

[0004] The present invention provides an OCT imaging method, comprising: acquiring a target B-Scan image of a target cornea; wherein the target B-Scan image includes a first B-Scan image and a second B-Scan image respectively acquired in mutually orthogonal scanning directions; inputting the target B-Scan image into a pre-trained corneal position prediction model, so as to output position information of the front surface of the target cornea in the target B-Scan image through the corneal position prediction model; performing curve fitting based on the position information to obtain a fitting curve; determining a target corneal vertex of the target cornea based on the vertex of the fitting curve; and performing OCT imaging on the target eye using the target corneal vertex as the scanning center.

[0005] Furthermore, the step of performing OCT imaging on the target eye with the target corneal vertex as the scanning center includes: controlling the operation of a motor in the OCT device to move the position of the scanning galvanometer so that the scanning center is aligned with the target corneal vertex, and performing OCT imaging on the target eye. Furthermore, before the step of performing OCT imaging on the target eye, the method further includes: if the alignment time between the scanning center and the target corneal vertex exceeds a preset time threshold, repeatedly performing the step of acquiring a target B-Scan image of the target cornea, so as to update the target corneal vertex of the target cornea according to the reacquired target B-Scan image.

[0006] Furthermore, the reacquired target B-Scan image is a target B-Scan image with a reduced sampling frequency.

[0007] Furthermore, the method also includes: displaying the collected eye image of the target eye through the terminal device, and displaying the target corneal vertex on the eye image; wherein, as the relative position between the target corneal vertex and the scanning center is different, the display color of the target corneal vertex is different.

[0008] Furthermore, the method also includes: displaying the collected eye image of the target eye through a terminal device, and displaying a scan line corresponding to a preset scanning protocol on the eye image; wherein, the display color of the scan line is different as the relative position between the target corneal vertex and the scanning center is different.

[0009] The present invention provides a method for training a corneal position prediction model, which includes: obtaining a sample B-Scan image of a sample cornea; wherein the sample B-Scan image is annotated with corneal classification information and corneal anterior surface position information; inputting the sample B-Scan image into an initial model to output a corneal classification prediction result and a corneal anterior surface position prediction result of the sample B-Scan image through the initial model; wherein the corneal classification prediction result is used to: indicate the probability that each pixel of the sample B-Scan image belongs to the cornea; based on a preset first loss function, the corneal classification information and the corneal classification prediction result, a first loss value is calculated; based on a preset second loss function, the corneal anterior surface position prediction result and the corneal anterior surface position information, a second loss value is calculated; and the initial model is trained based on the first loss value and the second loss value to obtain a trained corneal position prediction model.

[0010] Furthermore, the sample B-Scan image includes: a first sample image and a second sample image respectively acquired in mutually orthogonal scanning directions.

[0011] The present invention provides an electronic device comprising a processor and a memory, wherein the memory stores machine-executable instructions that can be executed by the processor, and the processor executes the machine-executable instructions to implement any of the above-mentioned OCT imaging methods or any of the above-mentioned corneal position prediction model training methods.

[0012] The present invention provides a machine-readable storage medium, which stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement any of the above-mentioned OCT imaging methods, or any of the above-mentioned corneal position prediction model training methods.

[0013] The OCT imaging method and corneal position prediction model training method provided by the present invention input a captured target B-scan image into a pre-trained corneal position prediction model, outputting the position information of the anterior surface of the target cornea in the target B-scan image; performing curve fitting based on the position information to obtain a fitting curve; determining the target corneal vertex of the target cornea based on the vertex of the fitting curve; and performing OCT imaging of the target eye using the target corneal vertex as the scanning center. This method, using the corneal position prediction model, can quickly and accurately predict the position information of the anterior surface of the target cornea in the target B-scan image. The target corneal vertex can be accurately determined through fitting, and using the target corneal vertex as the scanning center can improve the accuracy of OCT imaging. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0015] Figure 1 A flowchart of an OCT imaging method provided by an embodiment of the present invention; Figure 2 A schematic diagram of a display effect of a scan line provided by an embodiment of the present invention; Figure 3 A flowchart of a method for training a corneal position prediction model provided by an embodiment of the present invention; Figure 4 A schematic structural diagram of a corneal position prediction model provided by an embodiment of the present invention; Figure 5 A schematic structural diagram of an OCT imaging device provided in an embodiment of the present invention; Figure 6 A schematic structural diagram of a corneal position prediction model training device provided by an embodiment of the present invention; Figure 7 A schematic structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0016] The following will clearly and completely describe the technical solutions of the present invention in conjunction with the embodiments. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0017] Optical coherence tomography (OCT) technology has been widely used for corneal thickness measurement and other applications. OCT technology can produce high-resolution images at the micron level, which is crucial for accurate corneal thickness measurement. Corneal thickness topography can be used for preliminary assessment of corneal refractive surgery and to monitor the progression of corneal disease. To achieve high-precision corneal thickness measurement, a star scanning protocol is often used. In this protocol, each scan line passes through the scan center. Before scanning, the center of the entire scan area must be aligned with the corneal vertex, which is crucial for accurate thickness calculation. The adjustment process typically requires the operator to manually adjust the corresponding motor while observing the eye image captured by the pupil camera and the OCT image captured by the OCT device on the LiveView interface. This process requires high operator experience and precise operation skills. To address these issues, a series of related technologies based on traditional image processing algorithms have emerged, aiming to determine the corneal vertex position by analyzing and calculating the optical coherence tomography image (B-scan) containing the cornea. However, these methods take a long time to process and are poorly adaptable to rapid, minute eye movements and various scanning interferences during the acquisition process. This makes it difficult to accurately identify the position of the corneal vertex and, consequently, to ensure that the corneal vertex is aligned with the center of the entire scanning area. This reduces the accuracy of OCT imaging and limits its practicality. Therefore, embodiments of the present invention provide an OCT imaging method and a method for training a corneal position prediction model. This technology can be applied to applications requiring accurate determination of the corneal vertex.

[0018] To facilitate understanding of this embodiment, an OCT imaging method disclosed in an embodiment of the present invention is first introduced. Figure 1 As shown, the method includes the following steps: Step S102, acquiring a target B-Scan image of the target cornea; wherein the target B-Scan image includes a first B-Scan image and a second B-Scan image acquired in mutually orthogonal scanning directions; The above-mentioned target cornea can be the cornea of ​​the patient's target eye, and the cornea is a transparent thin film at the front of the eye; the above-mentioned mutually orthogonal scanning directions can be understood as two scanning directions perpendicular to each other, that is, the angle between the two scanning directions is 90°, for example, it can be a horizontal scanning direction and a vertical scanning direction, etc.; in actual implementation, when OCT imaging of the target eye is required, the first B-Scan image and the second B-Scan image of the target cornea can be obtained first, and the first B-Scan image and the second B-Scan image are two images collected in mutually orthogonal scanning directions.

[0019] Step S104, inputting the target B-Scan image into a pre-trained corneal position prediction model, so as to output position information of the front surface of the target cornea in the target B-Scan image through the corneal position prediction model; The above-mentioned corneal position prediction model can be a neural network in different forms, such as a fully convolutional network with a U-shaped structure of an encoder and a decoder, and of course it can also be a fully connected network, etc. The specific network structure can be selected according to actual needs, and is not limited here; the above-mentioned front surface of the cornea can be understood as the surface of the target cornea facing the outside of the eye; the above-mentioned position information can specifically be position coordinates, etc.; in actual implementation, the above-mentioned first B-Scan image and the second B-Scan image can be respectively input into a pre-trained corneal position prediction model, and the corneal position prediction model outputs the position information of the front surface of the target cornea in the first B-Scan image, and outputs the position information of the front surface of the target cornea in the second B-Scan image.

[0020] Step S106, performing curve fitting according to the position information to obtain a fitting curve; A quadratic curve can be fitted based on the position information of the front surface of the target cornea in the first B-Scan image. For example, the least squares method can be used. The general formula of the quadratic curve is: , the coefficients a, b, and c are obtained by quadratic curve fitting, and then a quadratic curve, namely the fitting curve mentioned above, can be obtained; similarly, a quadratic curve fitting can be performed based on the position information of the front surface of the target cornea in the second B-Scan image to obtain the fitted quadratic curve.

[0021] Step S108, determining a target corneal vertex of the target cornea based on the curve vertex of the fitting curve; After obtaining the fitting curve corresponding to the first B-Scan image, the vertex of the fitting curve can be calculated according to the preset vertex calculation formula. Similarly, after obtaining the fitting curve corresponding to the second B-Scan image, the vertex of the fitting curve can be calculated according to the preset vertex calculation formula. According to the vertices of the two fitting curves, the target corneal vertex of the target cornea can be determined on the projection plane. Specifically, the first B-Scan image and the second B-Scan image respectively obtain the positions of the curve vertices in the x-axis and y-axis directions. According to the vertex coordinates in the x-axis and y-axis directions, the position of the target corneal vertex can be uniquely determined on the plane as ( , ).

[0022] Step S110 , taking the target corneal vertex as the scanning center, performing OCT imaging on the target eye.

[0023] In actual implementation, the scanning center can be aligned to the target corneal apex, and then a suitable scanning protocol can be used to scan the target eye for OCT imaging.

[0024] The OCT imaging method described above inputs the acquired target B-scan image into a pre-trained corneal position prediction model, outputting the position information of the anterior surface of the target cornea in the target B-scan image. Curve fitting is performed based on this position information to obtain a fitted curve. The target corneal vertex of the target cornea is determined based on the vertex of the fitted curve. The target corneal vertex serves as the scanning center for OCT imaging of the target eye. This method uses the corneal position prediction model to quickly and accurately predict the position information of the anterior surface of the target cornea in the target B-scan image. The fitting process accurately determines the target corneal vertex, and using the target corneal vertex as the scanning center improves OCT imaging accuracy.

[0025] The present invention also provides another OCT imaging method, which is implemented based on the method of the above embodiment and includes the following steps: Step 1: Acquire a target B-Scan image of the target cornea; wherein the target B-Scan image includes a first B-Scan image and a second B-Scan image acquired in mutually orthogonal scanning directions; Step 2: Input the target B-Scan image into a pre-trained corneal position prediction model to output the position information of the front surface of the target cornea in the target B-Scan image through the corneal position prediction model; Step 3: Perform curve fitting based on the position information to obtain a fitting curve; Step 4, determining a target corneal vertex of the target cornea based on the curve vertex of the fitting curve; Step 5: Control the motor in the OCT device to move the scanning galvanometer so that the scanning center is aligned with the target corneal vertex to perform OCT imaging on the target eye.

[0026] The scanning galvanometer is a high-precision optical scanning device primarily used to quickly and precisely control the deflection direction of a laser or other light beam. In practical implementation, the motor in the OCT device can be controlled to drive the scanning galvanometer, thereby adjusting its position. Specifically, the position of the scanning galvanometer can be adjusted by adjusting the position voltage signal input to the scanning galvanometer. After aligning the scanning center with the target corneal vertex, the target eye is scanned using an appropriate scanning protocol for OCT imaging.

[0027] In one embodiment, before performing OCT imaging on the target eye, the method further includes: if the alignment time between the scan center and the target corneal vertex exceeds a preset time threshold, repeatedly performing the step of acquiring a target B-Scan image of the target cornea, thereby updating the target corneal vertex of the target cornea based on the newly acquired target B-Scan image. In actual implementation, when the alignment time exceeds the preset time threshold, it is generally considered that there is a high possibility of eye movement. In this case, the target B-Scan image of the target cornea can be reacquired, that is, the first B-Scan image and the second B-Scan image are respectively acquired in mutually orthogonal scanning directions, and the target corneal vertex is recalculated based on the two newly acquired B-Scan images. It can also be understood that if the scan center and the target corneal vertex are misaligned for a long time, the target corneal vertex can be updated to guide the user to perform precise alignment in the current situation.

[0028] In one implementation, the reacquired target B-scan image is a target B-scan image obtained after the sampling frequency has been reduced. During the corneal position prediction model training process, training data from various scanning conditions and low-quality scan images is typically incorporated. Therefore, even with the relatively low-quality target B-scan image rapidly input in this embodiment, the corneal position prediction model can effectively calculate the target corneal vertex and provide rapid real-time guidance.

[0029] Step six, displaying the collected eye image of the target eye through the terminal device, and displaying the target corneal vertex on the eye image; wherein, the display color of the target corneal vertex is different as the relative position between the target corneal vertex and the scanning center is different.

[0030] The terminal device may be a computer, a mobile terminal, etc., and the eye image of the target eye displayed on the terminal device may be an eye image captured by a pupil camera, such as Figure 2A schematic diagram of an eye image is shown; in actual implementation, the target corneal vertex and the scanning center can be displayed on the eye image through the terminal device. In the process of aligning the scanning center with the target corneal vertex, the display color of the target corneal vertex can be changed according to the relative position relationship between the scanning center and the target corneal vertex. For example, if the scanning center is aligned with the target corneal vertex, the target corneal vertex can be displayed in green. If the scanning center is not aligned with the target corneal vertex, the target corneal vertex can be displayed in red. By displaying different colors, the user can intuitively judge whether the target corneal vertex is aligned with the scanning center, thereby improving the user's operating experience.

[0031] Step seven: display the collected eye image of the target eye through the terminal device, and display the scan line corresponding to the preset scanning protocol on the eye image; wherein, the display color of the scan line is different as the relative position between the target corneal vertex and the scanning center is different.

[0032] The eye image of the target eye displayed on the terminal device may be an eye image captured by a pupil camera, such as Figure 2 The eye image shown; the above-mentioned preset scanning protocol can be a star scanning protocol, a cross scanning protocol, etc.; the above-mentioned scanning line can be understood as the path or trajectory of the actual data collection during the scanning process; in actual implementation, the target corneal vertex and the scanning center can be displayed by the terminal device. In the process of aligning the scanning center with the target corneal vertex, the display color of the scanning line can be changed according to the relative position relationship between the scanning center and the target corneal vertex. For example, if the scanning center is aligned with the target corneal vertex, the scanning line can be displayed in green. If the scanning center is not aligned with the target corneal vertex, the scanning line can be displayed in red. By displaying different colors, the user can intuitively judge whether the target corneal vertex and the scanning center are aligned, thereby improving the user's operating experience.

[0033] from Figure 2 It can be seen that when the scan center is not aligned with the target corneal vertex, the scan line may be displayed in black, and when the scan center is aligned with the target corneal vertex, the scan line may be displayed in gray.

[0034] The aforementioned OCT imaging method utilizes a deep learning-based corneal position prediction model that can quickly and accurately predict the position of the anterior surface of the target cornea in the target B-scan image. This model then performs quadratic curve fitting based on this anterior surface position information, with the vertex of the resulting fitted curve representing the target corneal vertex. Furthermore, to obtain the spatial location of the target corneal vertex, this method rapidly and continuously scans two B-scan images in mutually orthogonal directions (typically horizontal and vertical scanning directions) as input to the corneal position prediction model. The model then predicts the corneal vertex in these two orthogonal directions, thereby guiding the operator or automatically adjusting the program to move it to the center of the scan.

[0035] This method uses a corneal position prediction model to quickly and accurately predict the position information of the front surface of the target cornea in the target B-Scan image. The target corneal vertex can be accurately determined through fitting processing. Only the position of the scanning galvanometer needs to be moved to align the scanning center with the target corneal vertex, thereby improving the efficiency of the alignment operation. In addition, during the alignment process, as the relative position between the target corneal vertex and the scanning center changes, the target corneal vertex and the scanning line can display corresponding colors, allowing the user to intuitively judge whether the target corneal vertex and the scanning center are aligned based on the displayed color, thereby improving the user's operating experience.

[0036] The embodiment of the present invention provides a method for training a corneal position prediction model, such as Figure 3 As shown, the method includes the following steps: Step S302, obtaining a sample B-Scan image of the sample cornea; wherein the sample B-Scan image is annotated with corneal classification information and corneal anterior surface position information; The above-mentioned corneal classification information is typically pixel-level annotation of the sample cornea in the sample B-Scan image, i.e., the true label of each pixel in the sample B-Scan image (also referred to as the classification information of each pixel) is pre-annotated to indicate whether each pixel belongs to the sample cornea. For example, each pixel can be labeled 0 or 1, where 0 indicates that the pixel does not belong to the sample cornea and 1 indicates that the pixel does belong to the sample cornea. The above-mentioned corneal anterior surface position information is typically pre-annotated position coordinates of the corneal anterior surface in the sample B-Scan image. The sample B-Scan image includes: a first sample image and a second sample image acquired in mutually orthogonal scanning directions, for example, the first sample image and the second sample image can be acquired in the horizontal scanning direction and the vertical scanning direction, respectively.

[0037] Step S304: Input the sample B-Scan image into the initial model, so that the initial model outputs a classification prediction result of the cornea and a prediction result of the anterior corneal surface position of the sample B-Scan image; wherein the classification prediction result of the cornea is used to indicate the probability that each pixel of the sample B-Scan image belongs to the cornea; The initial model can be a neural network of various forms, such as a fully convolutional network with a U-shaped encoder and decoder, or a fully connected network. The appropriate network structure can be selected based on actual needs and is not limited here. The cornea classification prediction result can represent the classification prediction result for each pixel in the sample B-Scan image, for example, 0.3, 0.8, 1, etc. The corneal anterior surface position prediction result refers to the position coordinates of the corneal anterior surface in the sample B-Scan image predicted by the initial model. In actual implementation, the sample B-Scan image of the sample cornea can be input into the initial model, and the initial model can be used to predict the cornea classification prediction result and the corneal anterior surface position prediction result of the sample B-Scan image.

[0038] Step S306, calculating a first loss value based on a preset first loss function, the cornea classification information, and the cornea classification prediction result; The first loss function mentioned above can be expressed as: ; in, L represents the first loss function; N Indicates the number of pixels in the sample B-Scan image; Indicates the The classification information of pixels (i.e. pixels of the true label); Indicates the The classification prediction results of pixels.

[0039] In actual implementation, the first loss value can be calculated according to the above-mentioned first loss function. The first loss value is used to represent the gap between the classification information of the cornea and the classification prediction result of the cornea. Generally, the larger the first loss value, the greater the gap between the two, and the smaller the first loss value, the smaller the gap between the two.

[0040] Step S308, calculating a second loss value based on a preset second loss function, the corneal anterior surface position prediction result, and the corneal anterior surface position information; The above second loss function can be expressed as: ; in, represents the second loss function; Indicates the pixel width of the sample B-Scan image; The first The position information of the anterior corneal surface corresponding to the column (i.e. The actual label of the position of the anterior corneal surface corresponding to the column); The first The columns correspond to the predicted results of the anterior corneal surface position.

[0041] In actual implementation, the second loss value can be calculated according to the above-mentioned second loss function. The second loss value is used to represent the gap between the corneal anterior surface position prediction result and the corneal anterior surface position information. Generally, the larger the second loss value, the greater the gap between the two, and the smaller the second loss value, the smaller the gap between the two.

[0042] Step S310: training an initial model based on the first loss value and the second loss value to obtain a trained corneal position prediction model.

[0043] In actual implementation, the weight parameters of the initial model can be updated according to the first loss value and the second loss value until the first loss value converges and the second loss value converges, thereby obtaining a trained corneal position prediction model.

[0044] For easier understanding, see Figure 4 The structure diagram of a corneal position prediction model shown in the figure includes a corneal B-Scan image, a feature extraction network, a class prediction header, a surface prediction header, a pixel-wise loss function, an L1 loss function, pixel-level annotation of the corneal area, and surface position annotation. Table 1 below describes each part of the model. Table 1

[0045] During the training phase of the corneal position prediction model, the model uses the Class Prediction Header and the Surface Prediction Header to simultaneously predict the pixel-level classification and surface position coordinates of corneal tissue. The model then uses the corresponding loss function to calculate the loss value and supervise network training. During the prediction phase, only the portion predicting the anterior corneal surface position, as shown in the dashed box, is typically retained to improve inference speed. The dashed box represents the process during normal use of the trained corneal position prediction model.

[0046] The above-mentioned corneal position prediction model training method obtains a sample B-Scan image of a sample cornea; wherein the sample B-Scan image is annotated with corneal classification information and corneal anterior surface position information; the sample B-Scan image is input into an initial model, so that the initial model outputs a corneal classification prediction result and a corneal anterior surface position prediction result of the sample B-Scan image; wherein the corneal classification prediction result is used to indicate the probability that each pixel in the sample B-Scan image belongs to the cornea; a first loss value is calculated based on a preset first loss function, the corneal classification information, and the corneal classification prediction result; a second loss value is calculated based on a preset second loss function, the corneal anterior surface position prediction result, and the corneal anterior surface position information; and the initial model is trained based on the first loss value and the second loss value to obtain a trained corneal position prediction model. This method can improve the training reliability and accuracy of the corneal position prediction model, thereby improving the recognition accuracy of the corneal anterior surface position information.

[0047] In summary, the above OCT imaging method has the following beneficial effects: 1. High accuracy Through the powerful learning ability of the deep learning model, the corneal position in the OCT image can be accurately identified, improving the accuracy of tracking.

[0048] 2. Strong robustness By adding training data of various scanning conditions and low-quality scan images during the training phase, the trained corneal position prediction model has strong robustness and can accurately track the position of the anterior corneal surface even in the case of poor image quality or interference.

[0049] An embodiment of the present invention provides an OCT imaging device, such as Figure 5 As shown, the device includes: a first acquisition module 50, used to acquire a target B-Scan image of the target cornea; wherein the target B-Scan image includes a first B-Scan image and a second B-Scan image respectively acquired in mutually orthogonal scanning directions; a first output module 51, used to input the target B-Scan image into a pre-trained corneal position prediction model, so as to output the position information of the front surface of the target cornea in the target B-Scan image through the corneal position prediction model; a fitting module 52, used to perform curve fitting according to the position information to obtain a fitting curve; a determination module 53, used to determine the target corneal vertex of the target cornea based on the curve vertex of the fitting curve; and an imaging module 54, used to perform OCT imaging of the target eye with the target corneal vertex as the scanning center.

[0050] The above-mentioned OCT imaging device can quickly and accurately predict the position information of the front surface of the target cornea in the target B-Scan image through the corneal position prediction model. The target corneal vertex can be accurately determined through fitting processing. The target corneal vertex is used as the scanning center, which can improve the accuracy of OCT imaging.

[0051] Furthermore, the imaging module is also used to: control the operation of the motor in the OCT device to move the position of the scanning galvanometer so that the scanning center is aligned with the target corneal vertex to perform OCT imaging on the target eye. Furthermore, the imaging module is also used to: if the alignment time between the scanning center and the target corneal vertex exceeds a preset time threshold, repeat the step of acquiring a target B-Scan image of the target cornea, so as to update the target corneal vertex of the target cornea according to the re-acquired target B-Scan image.

[0052] Furthermore, the reacquired target B-Scan image is a target B-Scan image with a reduced sampling frequency.

[0053] Furthermore, the device is also used to: display the collected eye image of the target eye through the terminal device, and display the target corneal vertex on the eye image; wherein, the display color of the target corneal vertex is different as the relative position between the target corneal vertex and the scanning center is different.

[0054] Furthermore, the device is also used to: display the collected eye image of the target eye through a terminal device, and display the scanning line corresponding to the preset scanning protocol on the eye image; wherein, the display color of the scanning line is different as the relative position between the target corneal vertex and the scanning center is different.

[0055] The OCT imaging device provided in the embodiment of the present invention has the same implementation principle and technical effects as those in the aforementioned OCT imaging method embodiment. For the sake of brief description, any matters not mentioned in the OCT imaging device embodiment can be referred to the corresponding content in the aforementioned OCT imaging method embodiment.

[0056] The embodiment of the present invention provides a training device for a corneal position prediction model, such as Figure 6As shown, the device includes: a second acquisition module 60, used to acquire a sample B-Scan image of a sample cornea; wherein the sample B-Scan image is annotated with the classification information of the cornea and the position information of the anterior surface of the cornea; a second output module 61, used to input the sample B-Scan image into the initial model to output the classification prediction result of the cornea and the corneal anterior surface position prediction result of the sample B-Scan image through the initial model; wherein the classification prediction result of the cornea is used to: indicate the probability that each pixel of the sample B-Scan image belongs to the cornea; a first calculation module 62, used to calculate a first loss value based on a preset first loss function, the classification information of the cornea and the classification prediction result of the cornea; a second calculation module 63, used to calculate a second loss value based on a preset second loss function, the corneal anterior surface position prediction result and the corneal anterior surface position information; a training module 64, used to train the initial model based on the first loss value and the second loss value to obtain a trained corneal position prediction model.

[0057] The above-mentioned training device for the corneal position prediction model can improve the training reliability and accuracy of the corneal position prediction model, thereby improving the recognition accuracy of the position information of the anterior surface of the cornea.

[0058] Furthermore, the sample B-Scan image includes: a first sample image and a second sample image respectively acquired in mutually orthogonal scanning directions.

[0059] The training device for the corneal position prediction model provided in the embodiment of the present invention has the same implementation principle and technical effects as those in the aforementioned embodiment of the training method for the corneal position prediction model. For the sake of brief description, for matters not mentioned in the embodiment of the training device for the corneal position prediction model, reference may be made to the corresponding content in the aforementioned embodiment of the training method for the corneal position prediction model.

[0060] The embodiment of the present invention further provides an electronic device, see Figure 7 As shown, the electronic device includes a processor 130 and a memory 131, wherein the memory 131 stores machine executable instructions that can be executed by the processor 130, and the processor 130 executes the machine executable instructions to implement the above-mentioned OCT imaging method or the training method of the corneal position prediction model.

[0061] Furthermore, Figure 7 The electronic device shown further includes a bus 132 and a communication interface 133 , and the processor 130 , the communication interface 133 and the memory 131 are connected via the bus 132 .

[0062] The memory 131 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage. The communication connection between the system network element and at least one other network element is achieved through at least one communication interface 133 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used. The bus 132 may be an ISA bus, a PCI bus, or an EISA bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 7 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0063] The processor 130 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by hardware integrated logic circuits in the processor 130 or by software instructions. The above processor 130 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in conjunction with the embodiments of the present invention can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium well-known in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in memory 131, and processor 130 reads information in memory 131 and, in conjunction with its hardware, completes the steps of the method of the aforementioned embodiment.

[0064] An embodiment of the present invention also provides a machine-readable storage medium, which stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the above-mentioned OCT imaging method or the training method of the corneal position prediction model. The specific implementation can be found in the method embodiment and will not be repeated here.

[0065] The computer program products of the OCT imaging method and the corneal position prediction model training method provided in the embodiments of the present invention include a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the previous method embodiments. For specific implementation, please refer to the method embodiments and will not be repeated here.

[0066] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0067] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An OCT imaging method, characterized in that: The method comprises: Acquire a target B-Scan image of a target cornea; wherein the target B-Scan image includes a first B-Scan image and a second B-Scan image acquired in mutually orthogonal scanning directions; Inputting the target B-Scan image into a pre-trained corneal position prediction model to output position information of the front surface of the target cornea in the target B-Scan image through the corneal position prediction model; Performing curve fitting according to the position information to obtain a fitting curve; determining a target corneal vertex of the target cornea based on a curve vertex of the fitting curve; The target corneal vertex is used as the scanning center to perform OCT imaging on the target eye.

2. The method according to claim 1, characterized in that The steps of performing OCT imaging on the target eye with the target corneal vertex as the scanning center include: The motor in the OCT device is controlled to move the position of the scanning galvanometer so that the scanning center is aligned with the target corneal vertex to perform OCT imaging on the target eye.

3. The method according to claim 2, characterized in that Before the step of performing OCT imaging on the target eye, the method further includes: If the alignment time between the scan center and the target corneal vertex exceeds a preset time threshold, the step of acquiring a target B-Scan image of the target cornea is repeated to update the target corneal vertex of the target cornea according to the reacquired target B-Scan image.

4. The method according to claim 3, characterized in that The reacquired target B-scan image is a target B-scan image after the sampling frequency is reduced.

5. The method according to claim 1, characterized in that The method further comprises: The collected eye image of the target eye is displayed by a terminal device, and the target corneal vertex is displayed on the eye image; wherein, as the relative position between the target corneal vertex and the scanning center is different, the display color of the target corneal vertex is different.

6. The method according to claim 1, characterized in that The method further comprises: The collected eye image of the target eye is displayed by a terminal device, and a scan line corresponding to a preset scanning protocol is displayed on the eye image; wherein, the display color of the scan line is different as the relative position between the target corneal vertex and the scanning center is different.

7. A method for training a corneal position prediction model, characterized in that: The method comprises: Acquire a sample B-Scan image of the sample cornea; wherein the sample B-Scan image is annotated with corneal classification information and corneal anterior surface position information; Inputting the sample B-Scan image into an initial model, so as to output a classification prediction result of the cornea and a prediction result of the anterior corneal surface position of the sample B-Scan image through the initial model; wherein the classification prediction result of the cornea is used to indicate the probability that each pixel of the sample B-Scan image belongs to the cornea; Calculating a first loss value based on a preset first loss function, the classification information of the cornea, and the classification prediction result of the cornea; Calculating a second loss value based on a preset second loss function, the corneal anterior surface position prediction result, and the corneal anterior surface position information; The initial model is trained based on the first loss value and the second loss value to obtain a trained corneal position prediction model.

8. The method according to claim 7, characterized in that The sample B-Scan image includes: a first sample image and a second sample image respectively acquired in mutually orthogonal scanning directions.

9. An electronic device, characterized in that: It includes a processor and a memory, the memory storing machine-executable instructions that can be executed by the processor, and the processor executing the machine-executable instructions to implement the OCT imaging method described in any one of claims 1 to 6, or the training method of the corneal position prediction model described in any one of claims 7 to 8.

10. A machine-readable storage medium, characterized in that The machine-readable storage medium stores machine-executable instructions. When the machine-executable instructions are called and executed by the processor, the machine-executable instructions prompt the processor to implement the OCT imaging method described in any one of claims 1-6, or the training method of the corneal position prediction model described in any one of claims 7-8.

Images