Eye vaulting prediction method, apparatus, device, and medium

Abstract

Embodiments of the present application disclose an eye vault prediction method, device, equipment and medium, the method comprises: obtaining attribute data of an anterior segment image to be predicted and a to-be-implanted lens, the anterior segment image to be predicted is an image based on optical coherence tomography technology; inputting the anterior segment image to be predicted and size data of the to-be-implanted lens into a preset prediction model to obtain prediction data of eye vault, the preset prediction model is a model constructed based on a convolutional neural network architecture, and the preset prediction model comprises a feature extraction algorithm, a feature fusion algorithm and a feature regression algorithm. By constructing a prediction model based on a convolutional neural network architecture, image features can be automatically extracted, and postoperative eye vault prediction data can be predicted by regression, thereby improving prediction accuracy and simplifying the calculation process of selecting a to-be-implanted lens before a lens implantation surgery.

Description

Technical Field

[0001] This invention relates to the field of medical technology, and in particular to a method, device, equipment and medium for predicting the arch height of the eye. Background Technology

[0002] ICL (Implantable Collamer Lens) surgery is a procedure that involves implanting a phakic intraocular lens into the posterior chamber of the eye. It is considered a cutting-edge technology that can replace other refractive surgeries for refractive correction and is one of the newest and safest products for myopia correction internationally, widely used today. It can correct a wide range of myopia, hyperopia, and astigmatism without removing or damaging corneal tissue or requiring postoperative sutures, and is particularly effective for treating high myopia.

[0003] Currently, the mainstream postoperative arch height prediction models for implantable lens (ICL) surgery are based on traditional machine learning algorithms. Traditional machine learning methods require doctors to manually extract image features, such as the location of key points in the patient's anterior segment OCT image. This process is time-consuming and cumbersome. Furthermore, manually extracted features can vary due to differences in the surgeon's technique, significantly impacting the prediction results. Choosing an ICL lens size that is too large or too small based on the prediction can cause discomfort or even inflammation in the patient. Summary of the Invention

[0004] In view of this, the present invention provides a method, apparatus, device and medium for predicting eye arch height, which solves the problems of time-consuming and cumbersome manual extraction of image features in the prior art, and the easy occurrence of errors that affect the prediction results.

[0005] To achieve one or more of the above objectives or other objectives, the present invention proposes a method for predicting the arch height of the eye, comprising: acquiring an anterior segment image to be predicted and attribute data of the lens to be implanted, wherein the anterior segment image to be predicted is an image based on optical coherence tomography (OCT) technology;

[0006] The anterior segment image to be predicted and the attribute data are input into a preset prediction model to obtain the predicted data of eye arch height. The preset prediction model is a model built on a convolutional neural network architecture and includes a feature extraction algorithm, a feature fusion algorithm and a feature regression algorithm.

[0007] Optionally, before the step of acquiring the anterior segment image to be predicted and the attribute data of the lens to be implanted, the method further includes:

[0008] Acquire an initial anterior segment image and identify the format data of the initial anterior segment image;

[0009] The format data is compared with a preset format to determine whether the format data of the initial anterior segment image is consistent with the preset format;

[0010] If the format data of the initial anterior segment image is consistent with the preset format, the initial anterior segment image is scaled to obtain a first image of a preset size, which is used as the anterior segment image to be predicted.

[0011] If the format data of the initial anterior segment image is inconsistent with the preset format, the initial anterior segment image is converted into a target anterior segment image of the preset format, and the target anterior segment image is scaled to obtain a second image of a preset size, which is used as the anterior segment image to be predicted.

[0012] Optionally, before the step of inputting the anterior segment image to be predicted and the attribute data into a preset prediction model to obtain predicted data for the eye arch height, wherein the preset prediction model is a model built based on a convolutional neural network architecture and the preset prediction model includes a feature extraction algorithm, a feature fusion algorithm, and a feature regression algorithm, the method further includes:

[0013] An initial prediction model is constructed based on a convolutional neural network architecture, which includes a feature extraction algorithm, a feature fusion algorithm, and a feature regression algorithm.

[0014] The initial prediction model is trained based on the training set data to obtain the preset prediction model. The training set data includes historical anterior segment images before the lens implantation surgery and the lens size corresponding to the historical anterior segment images.

[0015] Optionally, the step of training the initial prediction model based on the training set data to obtain the preset prediction model includes:

[0016] The training set data is input into the initial prediction model to obtain the initial predicted arch height data;

[0017] The historical arch height data corresponding to the historical anterior segment image is used as training labels. The loss value of the initial prediction model is calculated based on the initial predicted arch height data and the historical arch height data. When the change in the loss value is less than or equal to a threshold, the training of the initial prediction model is completed, and the preset prediction model is obtained. The historical arch height data is the eye arch height data after lens implantation surgery.

[0018] Optionally, the step of inputting the anterior segment image to be predicted and the attribute data into a preset prediction model to obtain predicted data for the eye arch height, wherein the preset prediction model is a model built based on a convolutional neural network architecture, and the preset prediction model includes the steps of feature extraction algorithm, feature fusion algorithm, and feature regression algorithm, including:

[0019] The anterior segment image to be predicted is input into the feature extraction algorithm in the preset prediction model to obtain the feature map of the anterior segment image to be predicted.

[0020] The feature map is input into the feature fusion algorithm in the preset prediction model to obtain feature maps of different sizes, which are used as target feature maps. All the target feature maps are then spliced ​​and fused to obtain the feature vector of the anterior segment image to be predicted.

[0021] The feature vector and the attribute data are input into the feature regression algorithm of the preset prediction model to obtain the predicted data of eye arch height. The feature regression algorithm includes a first sub-algorithm and a second sub-algorithm.

[0022] Optionally, the step of inputting the feature map into the feature fusion algorithm of the preset prediction model to obtain feature maps of different sizes as target feature maps, and concatenating and fusing all the target feature maps to obtain the feature vector of the anterior segment image to be predicted, includes:

[0023] The feature map is input into the feature fusion algorithm in the preset prediction model so that the feature fusion algorithm can obtain target feature maps of different sizes for the feature map through downsampling methods of different magnification.

[0024] The target feature maps of different sizes are transformed to obtain one-dimensional feature maps of each target feature map. The feature fusion algorithm then splices and fuses the one-dimensional feature maps of each target feature map to obtain the feature vector of the anterior segment image to be predicted.

[0025] Optionally, the step of inputting the feature vector and the attribute data into a feature regression algorithm of a preset prediction model to obtain predicted data on eye arch height, wherein the feature regression algorithm includes the steps of a first sub-algorithm and a second sub-algorithm, including:

[0026] The feature vector is input into the first sub-algorithm of the feature regression algorithm to obtain the first data, wherein the number of input nodes of the first sub-algorithm is equal to the length of the feature vector;

[0027] The first data and the attribute data of the lens to be implanted are spliced ​​together to obtain spliced ​​data. The spliced ​​data is then input into the second sub-algorithm of the feature regression algorithm to obtain the predicted data of the eye arch height. The attribute data includes size and degree.

[0028] On the other hand, embodiments of this application provide an eye arch height prediction device, the device comprising:

[0029] The data acquisition module is used to acquire the anterior segment image to be predicted and the attribute data of the lens to be implanted. The anterior segment image to be predicted is an image based on optical coherence tomography (OCT) technology.

[0030] The prediction module is used to input the anterior segment image to be predicted and the attribute data into a preset prediction model to obtain the predicted data of the eye arch height. The preset prediction model is a model built on a convolutional neural network architecture and includes a feature extraction algorithm, a feature fusion algorithm and a feature regression algorithm.

[0031] On the other hand, embodiments of this application provide an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory through the bus. When the machine-readable instructions are executed by the processor, the steps of the eye arch height prediction method described above are performed.

[0032] On the other hand, embodiments of this application provide a computer-readable storage medium, characterized in that a computer program is stored on the computer-readable storage medium, and the computer program is executed by a processor to perform the steps of the eye arch height prediction method described above.

[0033] Implementing the embodiments of the present invention will have the following beneficial effects:

[0034] By acquiring an anterior segment image to be predicted and attribute data of the lens to be implanted, wherein the anterior segment image to be predicted is an image based on optical coherence tomography (OCT), and inputting the anterior segment image to be predicted and the attribute data into a preset prediction model, the predicted arch height of the eye is obtained. The preset prediction model is a model built on a convolutional neural network architecture, and includes feature extraction algorithm, feature fusion algorithm, and feature regression algorithm. By building a prediction model using a convolutional neural network architecture, image features can be automatically extracted, and the predicted data of the postoperative arch height of the eye can be predicted through regression, thereby improving prediction accuracy and simplifying the calculation process of selecting the lens to be implanted before lens implantation surgery. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] in:

[0037] Figure 1 This is a flowchart of a method for predicting eye arch height provided in an embodiment of this application;

[0038] Figure 2This is a flowchart of another eye arch height prediction method provided in the embodiments of this application;

[0039] Figure 3 This is a schematic diagram of the structure of an eye arch height prediction device provided in an embodiment of this application;

[0040] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application;

[0041] Figure 5 This is a schematic diagram of the structure of a computer-readable storage medium provided in an embodiment of this application. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] like Figure 1 As shown in the embodiment of this application, a method for predicting eye arch height is provided, including:

[0044] S101. Obtain the anterior segment image to be predicted and the attribute data of the lens to be implanted, wherein the anterior segment image to be predicted is an image based on optical coherence tomography (OCT) technology.

[0045] For example, Optical Coherence Tomography (OCT) technology is based on the principle of low-coherence interference of light to acquire the backscattered light of a sample. By demodulating and processing the interference signal, a two-dimensional tomographic or three-dimensional structural image of the sample can be reconstructed. OCT technology combines multiple techniques such as low-coherence interference, confocal microscopy, heterodyne detection, and digital image processing. Compared with commonly used biomedical imaging technologies such as X-ray imaging, nuclear magnetic resonance, and ultrasound imaging, it has advantages such as high resolution at the micrometer level, non-contact, non-invasiveness, and lower cost, and has high application value in clinical and commercial fields.

[0046] S102. Input the anterior segment image to be predicted and the attribute data into a preset prediction model to obtain the predicted data of eye arch height. The preset prediction model is a model built on a convolutional neural network architecture and includes a feature extraction algorithm, a feature fusion algorithm and a feature regression algorithm.

[0047] For example, by inputting the patient's preoperative anterior segment OCT image and different lens size values ​​into a preset prediction model, the preset prediction model will predict the postoperative arch height value after different implanted lenses.

[0048] By acquiring an anterior segment image to be predicted and attribute data of the lens to be implanted, wherein the anterior segment image to be predicted is an image based on optical coherence tomography (OCT), and inputting the anterior segment image to be predicted and the attribute data into a preset prediction model, the predicted arch height of the eye is obtained. The preset prediction model is a model built on a convolutional neural network architecture, and includes feature extraction algorithm, feature fusion algorithm, and feature regression algorithm. By building a prediction model using a convolutional neural network architecture, image features can be automatically extracted, and the predicted data of the postoperative arch height of the eye can be predicted through regression, thereby improving prediction accuracy and simplifying the calculation process of selecting the lens to be implanted before lens implantation surgery.

[0049] In one possible implementation, prior to the step of acquiring the anterior segment image to be predicted and the attribute data of the lens to be implanted, the method further includes:

[0050] Acquire an initial anterior segment image and identify the format data of the initial anterior segment image;

[0051] The format data is compared with a preset format to determine whether the format data of the initial anterior segment image is consistent with the preset format;

[0052] If the format data of the initial anterior segment image is consistent with the preset format, the initial anterior segment image is scaled to obtain a first image of a preset size, which is used as the anterior segment image to be predicted.

[0053] If the format data of the initial anterior segment image is inconsistent with the preset format, the initial anterior segment image is converted into a target anterior segment image of the preset format, and the target anterior segment image is scaled to obtain a second image of a preset size, which is used as the anterior segment image to be predicted.

[0054] For example, the standard image format (JPG) is a very popular image format. JPG format can be used on all modern cameras, and most computers can open JPG format images. Users can also set the compression level at will to preserve image quality. It is a very convenient photo format. Therefore, using JPG format as the default format can ensure the applicability of this solution.

[0055] For example, if the format data of the initial anterior segment image is consistent with the preset format, it is adjusted to a 1024*1024 image (the first image of the preset size) by cropping and scaling.

[0056] For example, if the format data of the initial anterior segment image is inconsistent with the preset format, for example, the preset format is JPG format and the format data of the initial anterior segment image is in another format, then the initial anterior segment image in the other format is converted to JPG format to obtain the target anterior segment image corresponding to the initial anterior segment image, and then adjusted to a size of 1024*1024 (the second image of the preset size) by cropping and scaling.

[0057] For example, if the format data of the initial anterior segment image is inconsistent with the preset format, the image information (the initial anterior segment image) requiring format conversion is subjected to quality detection to obtain the original encoding quality parameters. The image information is then decoded based on these original encoding quality parameters to obtain the original data. Next, the corresponding weighted value is calculated based on the original encoding quality parameters and a preset weight table, and this weighted value is calculated using a preset function to obtain the target encoding quality parameters. Finally, the original data is encoded based on these target encoding quality parameters to obtain the target format image. Specifically, as shown... If the image information includes multiple parameters, the corresponding weighted values ​​of these parameters can be calculated separately in this way. For example, if the image information includes brightness and / or chromaticity, the specific steps are as follows: Multiply the elements in the brightness quantization table with the corresponding elements in the preset brightness weight table to obtain the corresponding products, and calculate the sum of all products to obtain the corresponding brightness weighted value; and / or, multiply the elements in the chromaticity quantization table with the corresponding elements in the preset chromaticity weight table to obtain the corresponding products, and calculate the sum of all products to obtain the corresponding chromaticity weighted value.

[0058] For example, the method for setting up a weight table can be as follows:

[0059] Obtain all coding quality parameters within the preset range, calculate the weight for each coding quality parameter, establish a one-to-one mapping relationship between the weight and each coding quality parameter, and save the mapping relationship as a weight table.

[0060] The step "calculate weights for each coding quality parameter and establish a one-to-one mapping relationship between the weights and each coding quality parameter" may include:

[0061] Calculate the quantization table corresponding to each quality based on the coding quality parameters, calculate the weight for each quantization table corresponding to each quality, and establish a one-to-one mapping relationship between the weight and the quantization table corresponding to each quality.

[0062] In one possible implementation, before the step of inputting the anterior segment image to be predicted and the attribute data into a preset prediction model to obtain predicted data for the eye arch height, wherein the preset prediction model is a model built based on a convolutional neural network architecture and includes a feature extraction algorithm, a feature fusion algorithm, and a feature regression algorithm, the following method is further included:

[0063] An initial prediction model is constructed based on a convolutional neural network architecture, which includes a feature extraction algorithm, a feature fusion algorithm, and a feature regression algorithm.

[0064] The initial prediction model is trained based on the training set data to obtain the preset prediction model. The training set data includes historical anterior segment images before the lens implantation surgery and the lens size corresponding to the historical anterior segment images.

[0065] For example, the dataset is divided into training and test sets in a 5:1 ratio. The training set is used to train the model. Training is complete once the loss function stabilizes. During model training, weights at different training iterations are saved. If the training accuracy is significantly higher after the loss function stabilizes, the learning rate can be adjusted to improve training accuracy. Training is then complete once the loss function stabilizes again. The test set images are then input into the model for testing. If the accuracy on the test set is significantly lower than the accuracy on the training set, it indicates that the model is overfitting. Weight values ​​with fewer training iterations can be used for testing. When the prediction accuracy of the training and test sets is approximately equal, these model weights are considered appropriate, thus yielding the preset prediction model.

[0066] In one possible implementation, the step of training the initial prediction model based on training set data to obtain the preset prediction model includes:

[0067] The training set data is input into the initial prediction model to obtain the initial predicted arch height data;

[0068] The historical arch height data corresponding to the historical anterior segment image is used as training labels. The loss value of the initial prediction model is calculated based on the initial predicted arch height data and the historical arch height data. When the change in the loss value is less than or equal to a threshold, the training of the initial prediction model is completed, and the preset prediction model is obtained. The historical arch height data is the eye arch height data after lens implantation surgery.

[0069] For example, the difference between the initial predicted arch height data and the historical arch height data is calculated as the loss value. When the change in the loss value is less than or equal to a threshold, the initial prediction model is considered to be stable, the training of the initial prediction model is completed, and the preset prediction model is obtained.

[0070] In one possible implementation, the step of inputting the anterior segment image to be predicted and the attribute data into a preset prediction model to obtain predicted data for the eye arch height, wherein the preset prediction model is a model built based on a convolutional neural network architecture, and the preset prediction model includes the steps of feature extraction algorithm, feature fusion algorithm, and feature regression algorithm, including:

[0071] The anterior segment image to be predicted is input into the feature extraction algorithm in the preset prediction model to obtain the feature map of the anterior segment image to be predicted.

[0072] The feature map is input into the feature fusion algorithm in the preset prediction model to obtain feature maps of different sizes, which are used as target feature maps. All the target feature maps are then spliced ​​and fused to obtain the feature vector of the anterior segment image to be predicted.

[0073] The feature vector and the attribute data are input into the feature regression algorithm of the preset prediction model to obtain the predicted data of eye arch height. The feature regression algorithm includes a first sub-algorithm and a second sub-algorithm.

[0074] For example, the anterior segment image to be predicted is input into a feature extraction module corresponding to the feature extraction algorithm. This feature extraction module is built on a deep learning framework (PyTorch) architecture. It processes the image using a 7*7 convolutional layer, a batch normalization (BN) layer, a corrected linear activation function (ReLU), and a 2*2 max pooling layer to reduce the size of the feature map and output a feature map of size 256*256. The feature extraction module also includes four residual blocks: a first residual block, a second residual block, a third residual block, and a fourth residual block. The first residual block receives the 256*256 feature map, reduces its size by half to 128*128, and then processes it through the second residual block to obtain a 64*64 feature map. Finally, the 64*64 feature map is processed by the third residual block, resulting in a feature map size of 32*32. After the 32*32 feature map is processed by the fourth residual block, the feature map size becomes 16*16. The feature maps with size of 128*128, 64*64, 32*32, and 16*16 are all used as feature maps of the anterior segment image to be predicted.

[0075] In one possible implementation, the step of inputting the feature map into the feature fusion algorithm of the preset prediction model to obtain feature maps of different sizes as target feature maps, and concatenating and fusing all the target feature maps to obtain the feature vector of the anterior segment image to be predicted includes:

[0076] The feature map is input into the feature fusion algorithm in the preset prediction model so that the feature fusion algorithm can obtain target feature maps of different sizes for the feature map through downsampling methods of different magnification.

[0077] The target feature maps of different sizes are transformed to obtain one-dimensional feature maps of each target feature map. The feature fusion algorithm then splices and fuses the one-dimensional feature maps of each target feature map to obtain the feature vector of the anterior segment image to be predicted.

[0078] For example, the feature fusion module in the preset prediction model, corresponding to the feature fusion algorithm, can fuse features from feature maps of different scales, improving the model's prediction accuracy and stability. The feature fusion module obtains feature maps of different scales through downsampling at different magnifications, such as... Figure 2 As shown, the backbone network (feature fusion algorithm) used downsampling rates of 0, 2, 4, and 8 to obtain feature maps at four scales, with sizes of 16*16, 8*8, 4*4, and 2*2, respectively. These two-dimensional feature maps were then unfolded into one-dimensional feature vectors. The unfolded vector lengths were 256, 64, 16, and 4, respectively. Finally, the vectors formed by unfolding the four feature maps were concatenated, resulting in a concatenated vector with a length of 340.

[0079] For example, the downsampling process is the process of reducing the size of the image. In order to make the image fit the size of the display area, a thumbnail of the corresponding image (feature maps of four scales) is generated. For example, the working process of pooling layers or convolutional layers in neural networks is a downsampling process. However, the image reduction caused by the working process of pooling layers or convolutional layers is for feature extraction, while pooling downsampling is to reduce the dimensionality of the features. Therefore, the role of downsampling is to reduce the amount of computation, prevent overfitting, and increase the receptive field so that more global information can be learned.

[0080] In one possible implementation, the step of inputting the feature vector and the attribute data into a feature regression algorithm of a preset prediction model to obtain predicted data on eye arch height, wherein the feature regression algorithm includes the steps of a first sub-algorithm and a second sub-algorithm, including:

[0081] The feature vector is input into the first sub-algorithm of the feature regression algorithm to obtain the first data, wherein the number of input nodes of the first sub-algorithm is equal to the length of the feature vector;

[0082] The first data and the attribute data of the lens to be implanted are spliced ​​together to obtain spliced ​​data. The spliced ​​data is then input into the second sub-algorithm of the feature regression algorithm to obtain the predicted data of the eye arch height. The attribute data includes size and degree.

[0083] For example, the feature regression module corresponding to the feature regression algorithm includes a first fully connected layer corresponding to the first sub-algorithm and a second fully connected layer corresponding to the second sub-algorithm. The number of input nodes of the first fully connected layer is the length of the output vector of the feature fusion module, i.e., 340, and the number of output nodes is set to 100. Since the postoperative arch height value is also affected by the size and power (attribute data) of the implanted ICL lens, the size and power values ​​are concatenated with the output nodes of the first fully connected layer. That is, the number of input nodes of the second fully connected layer is 102, and the number of output nodes is 1, to obtain the predicted data of the eye arch height.

[0084] For example, each node in a fully connected layer is connected to all nodes in the previous layer to synthesize the extracted features. Due to the fully connected nature of the layer, it typically has the most parameters. For instance, in VGG16, the second fully connected layer has 4096 input nodes, and the first fully connected layer has 7*7*512=25088 output nodes, so this transmission requires 4096*25088 weights.

[0085] For example, the first fully connected layer corresponding to the first sub-algorithm is used to fully connect each neuron in the first fully connected layer with the output vector of the feature fusion module. The first sub-algorithm integrates local information with class discrimination in the output vector of the feature fusion module, that is, feature information. The activation function of each neuron in the first fully connected layer and the second fully connected layer adopts the ReLU function.

[0086] For example, each neuron in the second fully connected layer is connected to a neuron in the first fully connected layer. The first sub-algorithm is a classification algorithm, which can use softmax logistic regression for classification. The second fully connected layer can also be called a softmax layer.

[0087] On the other hand, embodiments of this application provide an eye arch height prediction device, the device comprising:

[0088] The data acquisition module 201 is used to acquire the anterior segment image to be predicted and the attribute data of the lens to be implanted. The anterior segment image to be predicted is an image based on optical coherence tomography (OCT) technology.

[0089] The prediction module 202 is used to input the anterior segment image to be predicted and the attribute data into a preset prediction model to obtain the predicted data of the eye arch height. The preset prediction model is a model built on a convolutional neural network architecture and includes a feature extraction algorithm, a feature fusion algorithm and a feature regression algorithm.

[0090] By acquiring an anterior segment image to be predicted and attribute data of the lens to be implanted, wherein the anterior segment image to be predicted is an image based on optical coherence tomography (OCT), and inputting the anterior segment image to be predicted and the attribute data into a preset prediction model, the predicted arch height of the eye is obtained. The preset prediction model is a model built on a convolutional neural network architecture, and includes feature extraction algorithm, feature fusion algorithm, and feature regression algorithm. By building a prediction model using a convolutional neural network architecture, image features can be automatically extracted, and the predicted data of the postoperative arch height of the eye can be predicted through regression, thereby improving prediction accuracy and simplifying the calculation process of selecting the lens to be implanted before lens implantation surgery.

[0091] One possible implementation, such as Figure 4 As shown, this application embodiment provides an electronic device 300, including: a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor 320. When the processor 320 executes the computer program 311, it performs the following: acquiring an anterior segment image to be predicted and attribute data of a lens to be implanted, wherein the anterior segment image to be predicted is an image based on optical coherence tomography (OCT); inputting the anterior segment image to be predicted and the attribute data into a preset prediction model to obtain predicted data of the eye arch height, wherein the preset prediction model is a model built based on a convolutional neural network architecture, and the preset prediction model includes a feature extraction algorithm, a feature fusion algorithm, and a feature regression algorithm.

[0092] By acquiring an anterior segment image to be predicted and attribute data of the lens to be implanted, wherein the anterior segment image to be predicted is an image based on optical coherence tomography (OCT), and inputting the anterior segment image to be predicted and the attribute data into a preset prediction model, the predicted arch height of the eye is obtained. The preset prediction model is a model built on a convolutional neural network architecture, and includes feature extraction algorithm, feature fusion algorithm, and feature regression algorithm. By building a prediction model using a convolutional neural network architecture, image features can be automatically extracted, and the predicted data of the postoperative arch height of the eye can be predicted through regression, thereby improving prediction accuracy and simplifying the calculation process of selecting the lens to be implanted before lens implantation surgery.

[0093] In one possible implementation, such as Figure 5As shown, this application embodiment provides a computer-readable storage medium 400, on which a computer program 411 is stored. When the computer program 411 is executed by a processor, it performs the following: acquiring an anterior segment image to be predicted and attribute data of a lens to be implanted, wherein the anterior segment image to be predicted is an image based on optical coherence tomography (OCT); inputting the anterior segment image to be predicted and the attribute data into a preset prediction model to obtain predicted data of the eye arch height, wherein the preset prediction model is a model constructed based on a convolutional neural network architecture, and the preset prediction model includes a feature extraction algorithm, a feature fusion algorithm, and a feature regression algorithm.

[0094] By acquiring an anterior segment image to be predicted and attribute data of the lens to be implanted, wherein the anterior segment image to be predicted is an image based on optical coherence tomography (OCT), and inputting the anterior segment image to be predicted and the attribute data into a preset prediction model, the predicted arch height of the eye is obtained. The preset prediction model is a model built on a convolutional neural network architecture, and includes feature extraction algorithm, feature fusion algorithm, and feature regression algorithm. By building a prediction model using a convolutional neural network architecture, image features can be automatically extracted, and the predicted data of the postoperative arch height of the eye can be predicted through regression, thereby improving prediction accuracy and simplifying the calculation process of selecting the lens to be implanted before lens implantation surgery.

[0095] The computer storage medium of this invention can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0096] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0097] Program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, etc., or any suitable combination thereof.

[0098] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0099] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computing device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0100] Note that the above description is merely a preferred embodiment of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, the scope of which is determined by the scope of the appended claims.

[0101] The above description discloses only preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for predicting eye arch height, characterized in that, include: Acquire the anterior segment image to be predicted and the attribute data of the lens to be implanted, wherein the anterior segment image to be predicted is an image based on optical coherence tomography (OCT) technology; The anterior segment image to be predicted and the attribute data are input into a preset prediction model to obtain predicted data for the eye arch height. The preset prediction model is a model built on a convolutional neural network architecture and includes a feature extraction algorithm, a feature fusion algorithm, and a feature regression algorithm. Specifically, the anterior segment image to be predicted is input into the feature extraction algorithm in the preset prediction model to obtain a feature map of the anterior segment image; the feature map is input into the feature fusion algorithm in the preset prediction model so that the feature fusion algorithm obtains target feature maps of different sizes for the feature map through downsampling methods of different magnifications; different The target feature map of the size is transformed to obtain a one-dimensional feature map of each target feature map. The feature fusion algorithm then splices and fuses the one-dimensional feature maps of each target feature map to obtain the feature vector of the anterior segment image to be predicted. The feature vector is input into the first sub-algorithm of the feature regression algorithm to obtain first data, wherein the number of input nodes of the first sub-algorithm is equal to the length of the feature vector. The first data and the attribute data of the lens to be implanted are spliced ​​to obtain spliced ​​data. The spliced ​​data is input into the second sub-algorithm of the feature regression algorithm to obtain the predicted data of the eye arch height, wherein the attribute data includes size and degree.

2. The method for predicting eye arch height as described in claim 1, characterized in that, Prior to the step of acquiring the anterior segment image to be predicted and the attribute data of the lens to be implanted, the method further includes: Acquire an initial anterior segment image and identify the format data of the initial anterior segment image; The format data is compared with a preset format to determine whether the format data of the initial anterior segment image is consistent with the preset format; If the format data of the initial anterior segment image is consistent with the preset format, the initial anterior segment image is scaled to obtain a first image of a preset size, which is used as the anterior segment image to be predicted. If the format data of the initial anterior segment image is inconsistent with the preset format, the initial anterior segment image is converted into a target anterior segment image of the preset format, and the target anterior segment image is scaled to obtain a second image of a preset size, which is used as the anterior segment image to be predicted.

3. The method for predicting eye arch height as described in claim 1, characterized in that, Before the step of inputting the anterior segment image to be predicted and the attribute data into a preset prediction model to obtain the predicted data of the eye arch height, wherein the preset prediction model is a model built based on a convolutional neural network architecture and includes a feature extraction algorithm, a feature fusion algorithm, and a feature regression algorithm, the following steps are also included: An initial prediction model is constructed based on a convolutional neural network architecture, which includes a feature extraction algorithm, a feature fusion algorithm, and a feature regression algorithm. The initial prediction model is trained based on the training set data to obtain the preset prediction model. The training set data includes historical anterior segment images before the lens implantation surgery and the lens size corresponding to the historical anterior segment images.

4. The method for predicting eye arch height as described in claim 3, characterized in that, The step of training the initial prediction model based on the training set data to obtain the preset prediction model includes: The training set data is input into the initial prediction model to obtain the initial predicted arch height data; The historical arch height data corresponding to the historical anterior segment image is used as training labels. The loss value of the initial prediction model is calculated based on the initial predicted arch height data and the historical arch height data. When the change in the loss value is less than or equal to a threshold, the training of the initial prediction model is completed, and the preset prediction model is obtained. The historical arch height data is the eye arch height data after lens implantation surgery.

5. A device for predicting the height of an eye arch, characterized in that, The device includes: The data acquisition module is used to acquire the anterior segment image to be predicted and the attribute data of the lens to be implanted. The anterior segment image to be predicted is an image based on optical coherence tomography (OCT) technology. The prediction module is used to input the anterior segment image to be predicted and the attribute data into a preset prediction model to obtain predicted data of eye arch height. The preset prediction model is a model built based on a convolutional neural network architecture. The preset prediction model includes a feature extraction algorithm, a feature fusion algorithm, and a feature regression algorithm. Specifically, the anterior segment image to be predicted is input into the feature extraction algorithm in the preset prediction model to obtain a feature map of the anterior segment image to be predicted; the feature map is input into the feature fusion algorithm in the preset prediction model so that the feature fusion algorithm obtains target feature maps of different sizes for the feature map through downsampling methods of different magnification. The target feature maps of different sizes are dimensionally transformed to obtain one-dimensional feature maps of each target feature map. The feature fusion algorithm then splices and fuses these one-dimensional feature maps to obtain the feature vector of the anterior segment image to be predicted. The feature vector is input into the first sub-algorithm of the feature regression algorithm to obtain first data, wherein the number of input nodes of the first sub-algorithm is equal to the length of the feature vector. The first data and the attribute data of the lens to be implanted are spliced ​​together to obtain spliced ​​data. The spliced ​​data is then input into the second sub-algorithm of the feature regression algorithm to obtain predicted data for the arch height of the eye, wherein the attribute data includes size and degree.

6. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the steps of the eye arch height prediction method as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the eye arch height prediction method as described in any one of claims 1 to 4.

Images