An optimization system and method for calculating intraocular lens power
By generating postoperative anterior segment images and calculating ELP values using generative AI technology, the calculation of intraocular lens power is optimized, solving the problem of inaccurate intraocular lens calculation in existing technologies and improving postoperative visual quality.
Patent Information
- Application Number
- CN202611309255.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-27
- Publication Date
- 2026-11-10
AI Technical Summary
The accuracy of existing intraocular lens calculation formulas is insufficient, leading to increased refractive errors after cataract surgery and affecting patients' visual quality.
Using generative AI technology, postoperative anterior segment images are generated based on preoperative anterior segment optical correlation tomography images. The formula for calculating intraocular lens power is optimized by calculating ELP values. The model is trained using generative adversarial networks and loss functions to accurately predict the position of the intraocular lens in the eye.
It improves the accuracy of intraocular lens power calculation, reduces postoperative refractive errors in cataract surgery, and improves postoperative visual quality for patients.
Smart Images

Figure CN123068754A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of medical image processing and pattern recognition technology, and in particular to a system and method for optimizing the calculation of intraocular lens power. Background Technology
[0002] Cataract extraction combined with intraocular lens implantation is currently the only effective treatment for cataracts. The choice of intraocular lens (IOL) is a crucial factor affecting the visual quality of patients after cataract surgery. Inaccurate measurement and calculation of the IOL can lead to increased refractive errors post-cataract surgery. Generally, measurements of ocular bioparameters such as corneal curvature, axial length, anterior chamber depth, and the choice of IOL calculation formula can all influence the accuracy of IOL power selection. The effective lens position (ELP) refers to the position of the IOL implanted in the eye after cataract surgery. Preoperatively, the IOL power and predicted IOL position can be calculated using the selected IOL calculation formula.
[0003] The accuracy of IOL calculation formulas after the third generation has been significantly improved. These formulas introduce the concept of Effective Lens Position (FELP), enhancing the accuracy and personalized applicability of IOL power calculations by incorporating more parameters and more precise biometric data. For example, in the Hoffer Q formula, the ELP value is calculated based on the axial length (A) constant. The FELP calculation formula is as follows: [FELP = 0.58357 × A_constant - 63.896]. (A_constant) is the constant of axial length (A). This formula is used to derive personalized anterior chamber depth (pACD, i.e., ELP) from the A constant when an A-constant-based calculation is selected. The Haigis formula predicts the effective lens position (ELP) of the IOL using preoperatively measured anterior chamber depth (ACD) and axial length (AL). The Haigis formula is expressed as: [ELP = a0 + a1 × ACD + a2 × AL]. The three constants (a0), (a1), and (a2) in the formula are derived from clinical data on different IOL types and populations. The applicability of various intraocular lens (IOL) calculation formulas differs, and their accuracy varies in practical applications. Currently, there is no perfect IOL calculation formula that can completely eliminate postoperative refractive errors.
[0004] The increased prediction error of ELP after cataract surgery is a major cause of increased refractive errors. Whether pursuing the accuracy of ocular bioparameter measurements or optimizing the IOL calculation formula, the goal is to accurately predict the effective intraocular lens position (ELP) after cataract surgery. Therefore, developing a method and system that can accurately and quickly calculate the effective intraocular lens position (ELP) after cataract surgery from preoperative anterior segment images of patients, using generative artificial intelligence technology to generate postoperative cataract images from preoperative images, obtaining predicted ELP values, and optimizing the FELP calculation formula for the intraocular lens, is of significant importance and application value in reducing postoperative refractive errors and improving postoperative visual quality for patients. Summary of the Invention
[0005] The purpose of this invention is to propose an optimized system and method for calculating the power of an intraocular lens, in order to solve the problems existing in the prior art.
[0006] To achieve the above objectives, the present invention provides the following solution: An intraocular lens power calculation and optimization system includes: an acquisition module, a measurement module, a calculation module, and a selection module; The acquisition module is used to acquire preoperative anterior segment optical coherence tomography images of the subject; The measurement module is used to acquire postoperative anterior segment images of cataracts based on the preoperative anterior segment optical correlation tomography images using generative AI technology. The calculation module is used to calculate the ELP value in the postoperative anterior segment image; The selection module is used to substitute the ELP value into the intraocular lens calculation formula to obtain the intraocular lens power result.
[0007] Optionally, the measurement module includes: The dataset construction unit is used to collect preoperative anterior segment optical coherence tomography images and corresponding postoperative anterior segment images to construct the training dataset. A data preprocessing unit is used to preprocess the image data in the training dataset; The model training unit is used to train the generative adversarial network using the preprocessed dataset to obtain a generative model. The generation unit is used to input the preoperative anterior segment optical correlation tomography image into the generation model to generate the postoperative anterior segment image.
[0008] Optionally, the model training unit is further provided with a loss function for training the generative adversarial network; The loss function includes: mean absolute error loss and generative adversarial loss.
[0009] Optionally, the computing module includes: A contour filtering unit is used to perform contour filtering on the postoperative anterior segment image, retaining the contour of the anterior chamber region; The ELP calculation unit is used to obtain the centroid coordinates of the contour of the anterior chamber region and calculate the ELP value based on the centroid coordinates.
[0010] Optionally, in the contour filtering unit: The postoperative anterior segment image was converted into a grayscale image and binarized. At the same time, the image was Gaussian blurred to obtain a binary image. Contour detection is performed on the binary image. Contours with an area smaller than a preset area threshold and contours with an average gray value of pixels greater than a preset gray value threshold are removed. The remaining contours are the contours of the anterior chamber region.
[0011] Optionally, in the ELP calculation unit: Traverse all points on the outline of the anterior chamber region that are below the centroid level. Calculate the distance between all points on the centroid level and the centroid in the horizontal direction. The nearest outline point to the left of the centroid is the left contact point A between the iris and the anterior surface of the intraocular lens, and the nearest outline point to the right of the centroid is the right contact point B between the iris and the anterior surface of the intraocular lens. Connect the left contact point A and the right contact point B with a straight line. Move the AB line segment downward pixel by pixel in the vertical direction of the image. At the same time, count the sum of the gray values of all pixels on the line segment after the movement. When the sum of the gray values is greater than a preset threshold, it indicates that the line segment has contacted the anterior surface of the intraocular lens, so stop moving. Record the two endpoints of the line segment as C and D. First, obtain the midpoint O of line segment CD, then obtain the perpendicular bisector L of line segment CD, and calculate the intersection point E of the perpendicular bisector L and the contour of the corneal region. The straight line length of OE is the distance value of ELP.
[0012] An optimization method for calculating intraocular lens power, using the system described above, the method comprising: Preoperative optical coherence tomography (OCT) images of the anterior segment of the patient were acquired. Based on the preoperative anterior segment optical coherence tomography images, generative AI technology is used to obtain postoperative anterior segment images for cataract surgery. Calculate the ELP value in the postoperative anterior segment image; The ELP value is substituted into the intraocular lens calculation formula to obtain the intraocular lens power result.
[0013] Optionally, based on the preoperative anterior segment optical coherence tomography (OCT) images, generative AI technology is used to obtain postoperative anterior segment images for cataract surgery, including: Preoperative optical coherence tomography (OCT) images of the anterior segment and corresponding postoperative anterior segment images were collected to construct a training dataset. The image data in the training dataset is preprocessed; Using the preprocessed dataset, a generative adversarial network is trained based on a preset loss function to obtain a generative model; wherein, the loss function includes: mean absolute error loss and generative adversarial loss; The preoperative anterior segment optical coherence tomography (OCT) image is input into the generation model to generate the postoperative anterior segment image.
[0014] Optionally, calculating the ELP value in the postoperative anterior segment image includes: Contour filtering was performed on the postoperative anterior segment images to preserve the contour of the anterior chamber region; Obtain the centroid coordinates of the contour of the anterior chamber region, traverse all points on the contour of the anterior chamber region that are below the centroid horizontal line, calculate the distance between all points below the centroid horizontal line and the centroid in the horizontal direction, the contour point closest to the left of the centroid is the left contact point A of the iris and the anterior surface of the intraocular lens, and the contour point closest to the right of the centroid is the right contact point B of the iris and the anterior surface of the intraocular lens. Connect the left contact point A and the right contact point B with a straight line. Move the AB line segment downward pixel by pixel in the vertical direction of the image. At the same time, count the sum of the gray values of all pixels on the line segment after the movement. When the sum of the gray values is greater than a preset threshold, it indicates that the line segment has contacted the anterior surface of the intraocular lens, so stop moving. Record the two endpoints of the line segment as C and D. First, obtain the midpoint O of line segment CD, then obtain the perpendicular bisector L of line segment CD, and calculate the intersection point E of the perpendicular bisector L and the contour of the corneal region. The straight line length of OE is the distance value of ELP.
[0015] Optionally, contour filtering of the postoperative anterior segment image includes: The postoperative anterior segment image was converted into a grayscale image and binarized. At the same time, the image was Gaussian blurred to obtain a binary image. Contour detection is performed on the binary image. Contours with an area smaller than a preset area threshold and contours with an average gray value of pixels greater than a preset gray value threshold are removed. The remaining contours are the contours of the anterior chamber region.
[0016] The beneficial effects of this invention are as follows: This invention is based on generative AI technology. It predicts the accurate position of the intraocular lens (IOL) in the eye after surgery based on the anterior segment parameters before surgery. Furthermore, it uses the predicted IOL position to optimize the calculation of the IOL power for cataract surgery before surgery, thereby improving the accuracy of the calculation. Attached Figure Description
[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments 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.
[0018] Figure 1 This is a schematic diagram of a system for calculating and optimizing the power of an intraocular lens according to an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating how generative AI technology is used to construct postoperative anterior segment images for cataract surgery based on preoperative anterior segment optical coherence tomography images. Figure 3 This is a schematic diagram illustrating the automatic calculation of predicted ELP values in a measured and predicted image according to an embodiment of the present invention. Figure 4 This is a schematic diagram of a method for optimizing the calculation of intraocular lens power according to an embodiment of the present invention. Detailed Implementation
[0019] 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.
[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0021] like Figure 1 As shown in the figure, this embodiment proposes an optimization system for calculating the power of an intraocular lens, including: an acquisition module, a measurement module, a calculation module, and a selection module; The acquisition module is used to acquire preoperative anterior segment optical coherence tomography images of the patient; The measurement module is used to acquire postoperative anterior segment images of cataracts based on preoperative anterior segment optical correlation tomography images using generative AI technology. The calculation module is used to calculate the ELP value in the anterior segment image after surgery, which is the vertical distance from the posterior surface of the cornea to the anterior surface of the intraocular lens; The selection module is used to substitute the ELP value into the intraocular lens calculation formula to obtain the intraocular lens power result.
[0022] Furthermore, the measurement module includes: The dataset construction unit is used to collect preoperative anterior segment optical coherence tomography images and corresponding postoperative anterior segment images to construct the training dataset. The data preprocessing unit is used to preprocess the image data in the training dataset; The model training unit is used to train the generative adversarial network using the preprocessed dataset to obtain a generative model. The generation unit is used to input preoperative anterior segment optical coherence tomography images into the generation model to generate postoperative anterior segment images.
[0023] Furthermore, a loss function is also set in the model training unit to train the generative adversarial network; The loss functions include: mean absolute error loss and generative adversarial loss.
[0024] Specifically, in this embodiment, a generative adversarial network (GAN) based on pix2pix is used to construct a generative model from preoperative anterior segment optical coherence tomography (OCT) images to postoperative anterior segment images after cataract surgery. The specific technical solution is as follows: First, 500 pairs of preoperative anterior segment OCT images and their corresponding postoperative anterior segment images are collected to construct a high-quality training dataset. During data preprocessing, all images are uniformly adjusted to a resolution of 256×256 pixels to match the model's input. Regarding the model architecture, a pix2pix generative adversarial network is used as the base model based on the PyTorch deep learning framework. The generator network uses a ResNet residual network with excellent feature extraction capabilities as a convolutional structure to improve the detail representation of the generated images. The model's loss function consists of two parts: (1) Mean absolute error (L1) loss: ; in The source image is input to the generator. For a real "target" image, The fake images generated by the generator The L1 norm is calculated by summing the absolute values of the differences between the two values. This represents the expected value (average) of all data in the dataset.
[0025] (2) Generate adversarial loss: ; in, For generator networks. For discriminator networks. The discriminator is used to determine the probability that a "real image" is real. The discriminator is used to determine the probability that the "generated image" is true.
[0026] During training, the weight factor for the Mean Absolute Error (MAO) L1 loss was set to 100 to better maintain pixel-level consistency between the generated and real images. Model training was performed on a computing platform with two NVIDIA 3090 GPUs and converged after 200 iterations. In the prediction phase, a preoperative anterior segment optical coherence tomography (OCT) image was input, the system invoked the trained generative model for inference, and then output the corresponding postoperative anterior segment synthetic image. Figure 2 This demonstrates postoperative anterior segment images of cataract surgery constructed using a measurement model based on generative AI technology and acquired preoperative anterior segment optical correlation tomography images.
[0027] Furthermore, the computing module includes: The contour filtering unit is used to filter the contours of the anterior segment images after surgery, preserving the contours of the anterior chamber region. The ELP calculation unit is used to obtain the centroid coordinates of the anterior chamber region's contour and calculate the ELP value based on the centroid coordinates.
[0028] Furthermore, in the contour filtering unit: The anterior segment image after surgery was converted to grayscale and binarized. At the same time, Gaussian blur was applied to the image to obtain a binary image. Contour detection is performed on the binary image. Contours with an area smaller than a preset area threshold and contours with an average gray value of pixels greater than a preset gray value threshold are removed. The remaining contours are the contours of the anterior chamber region.
[0029] Furthermore, such as Figure 3 As shown, in the ELP calculation unit: Traverse all points on the outline of the anterior chamber region that are below the centroid level. Calculate the distance between all points on the centroid level and the centroid in the horizontal direction. The nearest outline point to the left of the centroid is the left contact point A between the iris and the anterior surface of the intraocular lens, and the nearest outline point to the right of the centroid is the right contact point B between the iris and the anterior surface of the intraocular lens. Connect the left contact point A and the right contact point B with a straight line. Move the AB line segment downward pixel by pixel in the vertical direction of the image. At the same time, count the sum of the gray values of all pixels on the line segment after the movement. When the sum of the gray values is greater than the preset threshold (set to 2000 in this system), it indicates that the line segment has contacted the anterior surface of the intraocular lens, so stop moving. Record the two endpoints of the line segment as C and D. The shapely library in Python is used to perform geometric operations. First, the midpoint O of line segment CD is obtained, then the perpendicular bisector L of line segment CD is obtained, and the intersection point E of the perpendicular bisector L and the contour of the corneal region is calculated. The straight line length OE is the distance value of ELP.
[0030] In this embodiment, the selection module is used to optimize the intraocular lens calculation formula; the ELP value obtained by the method of this system is substituted into the intraocular lens calculation formula for calculation to obtain the intraocular lens power result.
[0031] The following uses the Haigis intraocular lens calculation formula as an example: The Haigis formula is a fourth-generation intraocular lens (IOL) power calculation formula. It predicts the effective lens position (ELP) of the IOL by measuring the anterior chamber depth (ACD) and axial length (AL) before surgery.
[0032] 1. The ELP in the Haigis formula is: ELP = a0 + a1 × ACD + a2 × AL; Where: ELP is the effective lens position, ACD is the anterior chamber depth, and AL is the axial length.
[0033] The three constants (a0), (a1), and (a2) in the Haigis formula are derived from clinical data of different IOL types and populations. They are not fixed values but need to be adjusted and optimized according to specific circumstances. Therefore, the Haigis formula does not directly provide a specific pACD value, but uses these constants to predict ELP and then calculate the IOL degree. In this embodiment, generative AI technology is used to construct postoperative anterior segment images of cataract surgery from preoperative anterior segment optical coherence tomography (OCT) images and automatically measure the predicted ELP (d).
[0034] 2. Substitute ELP(d) into the formula to calculate IOL diopter (P): ; That is, the calculation obtained in this embodiment like Figure 4 As shown in the figure, this embodiment also proposes an optimization method for calculating the power of an intraocular lens, the method including: Preoperative optical coherence tomography (OCT) images of the anterior segment of the patient were acquired. Based on preoperative anterior segment optical correlation tomography images, generative AI technology is used to obtain postoperative anterior segment images of cataracts. Calculate the ELP value in the anterior segment image after surgery; The ELP value is substituted into the intraocular lens calculation formula to obtain the intraocular lens power result.
[0035] Furthermore, based on preoperative anterior segment optical coherence tomography (OCT) images, generative AI technology was used to obtain postoperative anterior segment images of cataracts, including: Preoperative optical coherence tomography (OCT) images of the anterior segment and corresponding postoperative anterior segment images were collected to construct a training dataset. Preprocess the image data in the training dataset; Using the preprocessed dataset, a generative adversarial network is trained based on a preset loss function to obtain a generative model; the loss function includes: mean absolute error loss and generative adversarial loss; The preoperative anterior segment optical coherence tomography (OCT) images are input into the generative model to generate postoperative anterior segment images.
[0036] Furthermore, calculating the ELP value in the postoperative anterior segment images includes: First, the original image is converted to grayscale and binarized. Simultaneously, a Gaussian blur is applied to smooth the image and reduce noise. Contour detection is then performed on the preprocessed binary image. Contours with an area smaller than a preset threshold (10,000 pixels in this embodiment) and contours with an average grayscale value greater than a preset threshold (100 in this embodiment) are removed. The remaining contours are the contours of the anterior chamber region. The centroid coordinates of this contour are obtained. Then, all points on the contour that lie below the centroid are traversed, and their distances from the centroid are calculated horizontally. The contour point closest to the left of the centroid is the left contact point A between the iris and the anterior surface of the intraocular lens, and the contour point closest to the right of the centroid is the right contact point B between the iris and the anterior surface of the intraocular lens. The line segment AB is moved downwards pixel by pixel in the vertical direction of the image. Simultaneously, the sum of the grayscale values of all pixels on the line segment after the movement is calculated. When the sum of the grayscale values exceeds a preset threshold (2000 in this embodiment), it indicates that the line segment has contacted the anterior surface of the intraocular lens, and the movement stops. The two endpoints of this line segment are denoted as C and D. Finally, the shapely library in Python is used for geometric operations. First, the midpoint O of line segment CD is obtained, then the perpendicular bisector L of CD is obtained, and the intersection point E of L and the anterior surface of the cornea is calculated. The length of OE is the distance value of the ELP (Elastic Perpendicular Plexus).
[0037] In this embodiment, the ELP value obtained by this method is finally substituted into the intraocular lens calculation formula to obtain the intraocular lens power result.
[0038] Existing technologies all calculate intraocular lens (IOL) power based on preoperative ocular biological parameters. The technical solution of this embodiment uses preoperative parameter prediction to obtain the postoperative IOL position, thereby acquiring postoperative anterior segment biological parameters such as the effective postoperative IOL position. Based on the parameters of the effective postoperative IOL position, and combined with preoperative biological parameters, the IOL power for cataract surgery is optimized and calculated. Especially in this application, the postoperative ELP (Elastic Lens Position) can be generated and measured using AI; the optimization of IOL calculation, based on each patient's preoperative data, uses equations to obtain the parameters of the effective postoperative IOL position, achieving truly individualized optimization of the IOL power selection.
[0039] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.
Claims
1. A system for optimizing the calculation of intraocular lens power, characterized in that, include: The module includes a data acquisition module, a measurement module, a calculation module, and a selection module. The acquisition module is used to acquire preoperative anterior segment optical coherence tomography images of the subject; The measurement module is used to acquire postoperative anterior segment images of cataracts based on the preoperative anterior segment optical correlation tomography images using generative AI technology. The calculation module is used to calculate the ELP value in the postoperative anterior segment image; The selection module is used to substitute the ELP value into the intraocular lens calculation formula to obtain the intraocular lens power result.
2. The intraocular lens power calculation and optimization system according to claim 1, characterized in that, The measurement module includes: The dataset construction unit is used to collect preoperative anterior segment optical coherence tomography images and corresponding postoperative anterior segment images to construct the training dataset. A data preprocessing unit is used to preprocess the image data in the training dataset; The model training unit is used to train the generative adversarial network using the preprocessed dataset to obtain a generative model. The generation unit is used to input the preoperative anterior segment optical correlation tomography image into the generation model to generate the postoperative anterior segment image.
3. The intraocular lens power calculation and optimization system according to claim 2, characterized in that, The model training unit also includes a loss function for training the generative adversarial network. The loss function includes: mean absolute error loss and generative adversarial loss.
4. The intraocular lens power calculation and optimization system according to claim 1, characterized in that, The computing module includes: A contour filtering unit is used to perform contour filtering on the postoperative anterior segment image, retaining the contour of the anterior chamber region; The ELP calculation unit is used to obtain the centroid coordinates of the contour of the anterior chamber region and calculate the ELP value based on the centroid coordinates.
5. The intraocular lens power calculation and optimization system according to claim 4, characterized in that, In the contour filtering unit: The postoperative anterior segment image was converted into a grayscale image and binarized. At the same time, the image was Gaussian blurred to obtain a binary image. Contour detection is performed on the binary image. Contours with an area smaller than a preset area threshold and contours with an average gray value of pixels greater than a preset gray value threshold are removed. The remaining contours are the contours of the anterior chamber region.
6. The intraocular lens power calculation and optimization system according to claim 4, characterized in that, In the ELP calculation unit: Traverse all points on the outline of the anterior chamber region that are below the centroid level. Calculate the distance between all points on the centroid level and the centroid in the horizontal direction. The nearest outline point to the left of the centroid is the left contact point A between the iris and the anterior surface of the intraocular lens, and the nearest outline point to the right of the centroid is the right contact point B between the iris and the anterior surface of the intraocular lens. Connect the left contact point A and the right contact point B with a straight line. Move the AB line segment downward pixel by pixel in the vertical direction of the image. At the same time, count the sum of the gray values of all pixels on the line segment after the movement. When the sum of the gray values is greater than a preset threshold, it indicates that the line segment has contacted the anterior surface of the intraocular lens, so stop moving. Record the two endpoints of the line segment as C and D. First, obtain the midpoint O of line segment CD, then obtain the perpendicular bisector L of line segment CD, and calculate the intersection point E of the perpendicular bisector L and the contour of the corneal region. The straight line length of OE is the distance value of ELP.
7. An optimized method for calculating the power of an intraocular lens, characterized in that, The method, using the system as described in any one of claims 1-6, comprises: Preoperative optical coherence tomography (OCT) images of the anterior segment of the patient were acquired. Based on the preoperative anterior segment optical coherence tomography images, generative AI technology is used to obtain postoperative anterior segment images for cataract surgery. Calculate the ELP value in the postoperative anterior segment image; The ELP value is substituted into the intraocular lens calculation formula to obtain the intraocular lens power result.
8. The method for optimizing the calculation of intraocular lens power according to claim 7, characterized in that, Based on the preoperative anterior segment optical coherence tomography (OCT) images, generative AI technology is used to obtain postoperative anterior segment images for cataract surgery, including: Preoperative optical coherence tomography (OCT) images of the anterior segment and corresponding postoperative anterior segment images were collected to construct a training dataset. The image data in the training dataset is preprocessed; Using the preprocessed dataset, a generative adversarial network is trained based on a preset loss function to obtain a generative model; wherein, the loss function includes: mean absolute error loss and generative adversarial loss; The preoperative anterior segment optical coherence tomography (OCT) image is input into the generation model to generate the postoperative anterior segment image.
9. The method for optimizing the calculation of intraocular lens power according to claim 7, characterized in that, Calculating the ELP value in the postoperative anterior segment image includes: Contour filtering was performed on the postoperative anterior segment images to preserve the contour of the anterior chamber region; Obtain the centroid coordinates of the contour of the anterior chamber region, traverse all points on the contour of the anterior chamber region that are below the centroid horizontal line, calculate the distance between all points below the centroid horizontal line and the centroid in the horizontal direction, the contour point closest to the left of the centroid is the left contact point A of the iris and the anterior surface of the intraocular lens, and the contour point closest to the right of the centroid is the right contact point B of the iris and the anterior surface of the intraocular lens. Connect the left contact point A and the right contact point B with a straight line. Move the AB line segment downward pixel by pixel in the vertical direction of the image. At the same time, count the sum of the gray values of all pixels on the line segment after the movement. When the sum of the gray values is greater than a preset threshold, it indicates that the line segment has contacted the anterior surface of the intraocular lens, so stop moving. Record the two endpoints of the line segment as C and D. First, obtain the midpoint O of line segment CD, then obtain the perpendicular bisector L of line segment CD, and calculate the intersection point E of the perpendicular bisector L and the contour of the corneal region. The straight line length of OE is the distance value of ELP.
10. The method for optimizing the calculation of intraocular lens power according to claim 7, characterized in that, Contour filtering of the postoperative anterior segment images includes: The postoperative anterior segment image was converted into a grayscale image and binarized. At the same time, the image was Gaussian blurred to obtain a binary image. Contour detection is performed on the binary image. Contours with an area smaller than a preset area threshold and contours with an average gray value of pixels greater than a preset gray value threshold are removed. The remaining contours are the contours of the anterior chamber region.
