Optical biometric measurement device and method for automatically determining the state of an eye

Abstract

The application discloses an optical biology measuring device and method for automatically judging eyeball state, and relates to the field of medical image processing. The device comprises a pre-scanning image acquisition module, an eyeball state discrimination module and a measuring module. The pre-scanning image acquisition module is used for acquiring a standard scanning image. The eyeball state discrimination module is used for discriminating based on the scanning image and acquiring the eyeball state. The measuring module is used for completing biology measurement based on the eyeball state. According to the scanning image, the application adopts deep learning technology to extract features of a lens position pattern, and compares the lens position pattern with a preset database. Meanwhile, the application accurately discriminates the lens state according to a discrimination criterion. Furthermore, biology parameter scanning and / or measurement are performed according to the correct eyeball state, so that the accuracy and efficiency of biology parameter measurement can be effectively improved.

Description

Technical Field

[0001] This invention relates to the field of medical image processing, and more specifically to an optical biological measurement device and method for automatically determining the state of the eyeball. Background Technology

[0002] Ocular biometry refers to the measurement of various parameters of the eye structure using various examination methods, providing a reference for the diagnosis and treatment of eye diseases. Accurate measurement of these parameters is particularly crucial for assessing myopia progression and calculating the intraocular lens power during cataract surgery. Generally, the condition of the eye undergoing ocular biometry varies depending on the intraocular lens, affecting the measurement of ocular biological parameters. These lens conditions include: aphakia, intraocular lens (IOL), complete lens dislocation, and IOL with an intraocular lens. Because ultrasound and light waves travel at different speeds in different media (lens, IOL, silicone oil, aqueous humor), different eye conditions can significantly impact the accuracy of ocular biometry measurements.

[0003] Current optical ocular biometry techniques require operators to assess the patient's medical history (e.g., whether they have undergone cataract surgery and had an intraocular lens implanted, or whether they have had silicone oil implantation) and to manually determine the eye's condition based on the specific circumstances of the examination. This allows for adjustments to the measurement to ensure accurate biometry. However, some patients provide incomplete or incorrect medical histories, leading to inaccurate assessments of the eye's condition and affecting the measurement of ocular biometric parameters. Furthermore, since axial length measurement is performed along the visual axis, some patients have conditions such as lens subluxation or complete dislocation, meaning the light or ultrasound waves do not pass through the lens medium along the visual axis (in which case the ocular condition should be selected as aphakic). However, clinicians often perform measurements as if the patient had a phakic eye, resulting in errors in ocular biometry measurements. Summary of the Invention

[0004] To address the problems mentioned above, this invention aims to provide an ocular ultrasound biological measurement device and method that can automatically and accurately determine different eye conditions, for use in clinical practice to efficiently and accurately measure ocular biological parameters.

[0005] To achieve the above objectives, the present invention provides an optical biological measurement device for automatically determining the state of the eyeball, comprising: a pre-scan image acquisition module, an eyeball state discrimination module, and a measurement module;

[0006] The pre-scan image acquisition module is used to acquire standard scan images;

[0007] The eye state discrimination module is used to discriminate and obtain the eye state based on the scanned image; the eye state refers to the state of the lens; the eye state discrimination module uses different discrimination criteria for different eyes to discriminate the eye state, and the discrimination criteria are as follows:

[0008] When the ratio of lens area to pupil area is 1, it is determined that there is no lens dislocation.

[0009] When the ratio of lens area to pupil area is 0, it is determined to be a complete lens dislocation;

[0010] When the ratio of lens area to pupil area is >0 but <1, it is determined that there is incomplete lens dislocation.

[0011] The measurement module is used to perform biological measurements based on the state of the eyeball.

[0012] Preferably, the lens state includes: aphakic eye, phakic eye, intraocular lens eye, ICL eye, and incomplete lens dislocation.

[0013] Preferably, the eye state discrimination module includes: an automatic pupil region recognition unit, a pupil region image preprocessing unit, and a pupil region image classification unit;

[0014] The automatic pupil region recognition unit is used to recognize the pupil region based on the scanned image and obtain the circular outline of the pupil region;

[0015] The pupil area image preprocessing unit is used to preprocess the circular outline to obtain the processing result;

[0016] The pupil area image classification unit is used to classify and distinguish based on the processing results to obtain the eyeball state.

[0017] Preferably, the workflow of the automatic pupil region recognition unit includes:

[0018] The acquired color image is converted to grayscale and then Gaussian blurred using a 5x5 Gaussian kernel.

[0019] Edge features in the blurred image are captured by performing Canny edge detection and setting a low threshold of 50 and a high threshold of 150.

[0020] Based on edge detection, the Hough transform method based on image gradient information is used to locate the circular contour and complete the recognition.

[0021] Preferably, the workflow of the pupil area image preprocessing unit includes:

[0022] The identified circular outline is segmented to obtain an image of the pupil region;

[0023] The resolution of the pupil area image was adjusted to 112×112 and normalized to complete the preprocessing.

[0024] Preferably, the workflow of the pupil area image classification unit includes: training an image classification model using a convolutional neural network with a ResNet50 structure, wherein the initial learning rate is set to 1×10⁻⁶ during training. -3 Simultaneously, a learning rate decay strategy is employed for dynamic adjustment; Adam is selected as the optimizer to update model weights; cross-entropy loss is used as the loss function; and the batch size is set to 64.

[0025] The image classification model uses the trained model weights to perform forward propagation on the pupil area image, and outputs the probability distribution of each category through the softmax layer, thereby completing the automatic classification and recognition of eye state.

[0026] Preferably, when the determination result indicates incomplete lens dislocation, the determination is based on whether the lens position is detected at the center of the pupil, according to the following criteria:

[0027] The lens was detected at the center of the pupil; the phakic eye measurement mode was selected.

[0028] No lens was detected at the center of the pupil; select aphakic eye measurement mode.

[0029] This invention also provides an optical biological measurement method for automatically determining the state of the eyeball. The method is applied to the above-mentioned system, and the steps include:

[0030] Acquire standard scan images;

[0031] Based on the scanned image, the eye state is determined; the eye state refers to the lens state; the eye state determination module uses different determination criteria for different eyeballs to determine the eye state, and the determination criteria are as follows:

[0032] When the ratio of lens area to pupil area is 1, it is determined that there is no lens dislocation.

[0033] When the ratio of lens area to pupil area is 0, it is determined to be a complete lens dislocation;

[0034] When the ratio of lens area to pupil area is >0 but <1, it is determined that there is incomplete lens dislocation.

[0035] Biological measurements were performed based on the described eye condition.

[0036] Preferably, the lens state includes: aphakic eye, phakic eye, intraocular lens eye, ICL eye, and incomplete lens dislocation.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0038] This invention uses deep learning technology to extract features from the scanned image of the lens location and compares them with a pre-set database. Simultaneously, it accurately determines the lens state according to the discrimination criteria. Furthermore, it scans and / or measures biological parameters based on the correct eye condition, which can effectively improve the accuracy and efficiency of biological parameter measurement. Attached Figure Description

[0039] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are 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.

[0040] Figure 1 This is a schematic diagram of the discrimination process according to an embodiment of the present invention;

[0041] Figure 2 This is a schematic diagram of the system structure according to an embodiment of the present invention;

[0042] Figure 3 This is a schematic diagram of the lens state according to an embodiment of the present invention; wherein (a) indicates that the lens is not completely dislocated and the visual axis area does not pass through the lens medium; (b) indicates that the lens is not completely dislocated, but the visual axis area can pass through the lens medium. Detailed Implementation

[0043] 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.

[0044] 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.

[0045] Example 1:

[0046] This embodiment provides an optical biological measurement device for automatically determining the state of the eyeball, including: a pre-scan image acquisition module, an eyeball state discrimination module, and a measurement module; the pre-scan image acquisition module is used to acquire standard scan waveforms and scan images; the eyeball state discrimination module is used to discriminate based on the scan waveforms and scan images to obtain the eyeball state; the measurement module is used to complete biological measurements based on the eyeball state.

[0047] The following will, in conjunction with this embodiment, explain in detail how the present invention solves technical problems in real life.

[0048] First, the pre-scan image acquisition module uses existing ocular biological measurement scanning technology to perform a standard pre-scan of the eyeball based on the default eyeball state (phakic eye) to acquire standard scan images and / or scan waveforms.

[0049] The eye state discrimination module then uses convolutional neural networks from deep learning to extract features from the scanned waveforms and images at the location of the lens image. Based on a pre-set database and predetermined judgment criteria, it performs lens state discrimination on the waveforms and images. Different discrimination criteria are used for different ocular biological measurement techniques. Its main processes are database construction and pupil area recognition.

[0050] Database construction: Obtained ocular biological measurement scan images from 50 cases each of aphakic eyes, phakic eyes, intraocular lens eyes, phakic eyes after intraocular lens implantation (ICL eyes), and lens dislocation, and constructed a database.

[0051] Pupil region identification: Feature extraction is performed on the pupil region image, and a deep learning algorithm is used to determine the eye condition. The eye condition includes: aphakic, phakic, intraocular lens, ICL, and incomplete lens dislocation.

[0052] The specific process is as follows:

[0053] First, the automatic pupil region recognition unit implements an image processing workflow using the OpenCV library, aiming to accurately detect the circular outline of the pupil region from a given image. The processing begins by converting the original color image to grayscale using the `cv2.cvtColor` function, followed by calling the `cv2.GaussianBlur` function and applying a 5x5 Gaussian kernel to Gaussian blur the image, effectively smoothing it and reducing noise. Next, the `cv2.Canny` function performs Canny edge detection, capturing edge features in the image by setting a low threshold of 50 and a high threshold of 150. Building on edge detection, the module further calls the `cv2.HoughCircles` function to identify circles, using the `cv2.HOUGH_GRADIENT` method based on image gradient information to locate the circles. To improve detection accuracy, the minimum distance between circle centers, `minDist`, is set to 100 to reduce redundant detection. `param1` is set to 100 as the high threshold in Canny edge detection, while `param2` is set to 20 to determine the accumulator threshold for circle center detection. Furthermore, by setting minRadius=35 and maxRadius=40, the module limits the radius range of the detected circles, ensuring that only circles meeting specific dimensions are recognized, thereby accurately locating the pupil area. The circle with the largest radius that is ultimately recognized is the circular outline of the pupil area, and the center of this circle is the center of the pupil.

[0054] After recognizing the circular outline of the pupil region, the pupil region image preprocessing unit segments out the smallest square covering this circular outline. The coordinates of the top-left corner of this square are set to [(center x-coordinate - radius * 1.10), (center y-coordinate - radius * 1.10)], and the coordinates of the bottom-right corner are set to [(center x-coordinate + radius * 1.10), (center y-coordinate + radius * 1.10)]. Multiplying the radius by 1.10 ensures that the entire pupil region is covered. After segmenting the square image containing the pupil region, the resolution of this image is adjusted to 112×112 and normalized to adapt to the input of subsequent models.

[0055] Finally, automatic classification and recognition were performed using a pupil region image classification unit. This was a multi-classification task, aiming to automatically classify pupil region images into 7 categories. First, 1500 original images were collected and pupil region images were extracted. Then, each image was labeled with 7 categories, with approximately 200 samples per category. Next, a ResNet50 convolutional neural network was used to train an image classification model. During training, the initial learning rate was set to 1×10⁻⁶. -3The model employs a learning rate decay strategy for dynamic adjustment; Adam is selected as the optimizer to update model weights; cross-entropy loss is used as the loss function; and the batch size is set to 64. Training is completed in 100 rounds using the PyTorch programming framework on an Ubuntu environment. During the prediction phase, the model uses the trained model weights to perform forward propagation on new pupil region images, outputting the probability distribution of each category through a softmax layer, thereby completing the automatic classification and recognition task of eye state.

[0056] like Figure 1 As shown, the judgment criteria for incomplete lens dislocation are used to determine the state of the eyeball. The judgment criteria are as follows:

[0057] The lens / pupil area ratio is analyzed. When the lens / pupil area ratio is 1, it is automatically determined that there is no lens dislocation, and the lens-containing eye measurement mode is automatically selected.

[0058] When the ratio of lens area to pupil area is 0, it is automatically identified as complete lens dislocation, and the aphakic eye measurement mode is automatically selected.

[0059] When the ratio of lens area to pupil area is greater than 0 but less than 1, it is determined that there is incomplete lens dislocation. The determination is automatically made based on whether the lens is detected at the center of the pupil; if the lens is detected at the center of the pupil, the measurement mode for phakic eyes is automatically selected; if the lens is not detected at the center of the pupil, the measurement mode for aphakic eyes is automatically selected.

[0060] The above-mentioned different eye conditions include lens conditions such as aphakia, intraocular lens, complete lens dislocation, and intraocular lens with an intraocular lens.

[0061] Finally, based on the discrimination results from the aforementioned eye condition discrimination module, the measurement module performs correct biological measurements or, after rescanning with the correct eye condition, performs correct biological parameter measurements. A schematic diagram of the system structure in this embodiment is shown below. Figure 2 As shown.

[0062] Example 2:

[0063] This embodiment also provides an optical biological measurement method for automatically determining the state of the eyeball, the steps of which include:

[0064] S1. Obtain a standard scan image.

[0065] Using existing ocular biological measurement scanning technology, a standard pre-scan of the eyeball is performed based on the default eyeball state (phakic eye) to obtain standard scan waveforms and / or scan images.

[0066] S2. Based on the scanned image, determine the state of the eyeball.

[0067] The scanning waveforms and images are processed using convolutional neural networks in deep learning to extract features at the location of the lens image. Based on a pre-set database and predetermined criteria, the state of the lens is determined from the waveforms and images. Different criteria are used for different ocular biological measurement techniques to determine the eye state. The main steps include database construction and pupil area identification.

[0068] Database construction: Obtained ocular biological measurement scan images from 50 cases each of aphakic eyes, phakic eyes, intraocular lens eyes, phakic eyes after intraocular lens implantation (ICL eyes), and lens dislocation, and constructed a database.

[0069] Pupil region identification: Feature extraction is performed on the pupil region image, and a deep learning algorithm is used to determine the eye condition. The eye condition includes: aphakic, phakic, intraocular lens, ICL, and incomplete lens dislocation.

[0070] The aforementioned deep learning algorithm comprises three parts. The automatic pupil region recognition part implements an image processing workflow using the OpenCV library, aiming to accurately detect the circular outline of the pupil region from a given image. The processing first converts the original color image to grayscale using the `cv2.cvtColor` function, then calls the `cv2.GaussianBlur` function and applies a 5x5 Gaussian kernel to Gaussian blur the image, effectively smoothing the image and reducing noise interference. Next, the `cv2.Canny` function performs Canny edge detection, capturing edge features in the image by setting a low threshold of 50 and a high threshold of 150. Based on edge detection, the module further calls the `cv2.HoughCircles` function to identify circles, using the `cv2.HOUGH_GRADIENT` method based on image gradient information to locate the circles. To improve detection accuracy, the minimum distance between circle centers, `minDist`, is set to 100 to reduce redundant detection. `param1` is set to 100 as the high threshold in Canny edge detection, while `param2` is set to 20 to determine the accumulator threshold for circle center detection. Furthermore, by setting minRadius=35 and maxRadius=40, the module limits the radius range of the detected circles, ensuring that only circles meeting specific dimensions are recognized, thereby accurately locating the pupil area. The circle with the largest radius that is ultimately recognized is the circular outline of the pupil area, and the center of this circle is the center of the pupil.

[0071] After identifying the circular outline of the pupil region, the preprocessing part of the pupil area image needs to segment out the smallest square covering the circular outline. The coordinates of the top left corner of this square are set to [(center x-coordinate - radius * 1.10), (center y-coordinate - radius * 1.10)], and the coordinates of the bottom right corner are set to [(center x-coordinate + radius * 1.10), (center y-coordinate + radius * 1.10)]. Multiplying the radius by 1.10 ensures that the pupil region is completely covered. After segmenting the square image containing the pupil region, the resolution of this image is adjusted to 112×112 and normalized to adapt to the input of the subsequent model.

[0072] Finally, automatic classification and recognition are performed using the pupil region image classification component. This is a multi-classification task, aiming to automatically classify pupil region images into 7 categories. First, 1500 original images were collected and pupil region images were extracted. Then, each image was labeled with 7 categories, with approximately 200 samples per category. Next, a ResNet50 convolutional neural network was used to train an image classification model. During training, the initial learning rate was set to 1×10⁻⁶. -3 The model employs a learning rate decay strategy for dynamic adjustment; Adam is selected as the optimizer to update model weights; cross-entropy loss is used as the loss function; and the batch size is set to 64. Training is completed in 100 rounds using the PyTorch programming framework on an Ubuntu environment. During the prediction phase, the model uses the trained model weights to perform forward propagation on new pupil region images, outputting the probability distribution of each category through a softmax layer, thereby completing the automatic classification and recognition task of eye state.

[0073] like Figure 1 As shown, the judgment criteria for incomplete lens dislocation are used to determine the state of the eyeball. The judgment criteria are as follows:

[0074] The lens / pupil area ratio is analyzed. When the lens / pupil area ratio is 1, it is automatically determined that there is no lens dislocation, and the lens-containing eye measurement mode is automatically selected.

[0075] When the ratio of lens area to pupil area is 0, it is automatically identified as complete lens dislocation, and the aphakic eye measurement mode is automatically selected.

[0076] When the ratio of lens area to pupil area is greater than 0 but less than 1, it is determined that there is incomplete lens dislocation. The determination is automatically made based on whether the lens is detected at the center of the pupil; if the lens is detected at the center of the pupil, the measurement mode for phakic eyes is automatically selected; if the lens is not detected at the center of the pupil, the measurement mode for aphakic eyes is automatically selected.

[0077] The above-mentioned different eye conditions include lens conditions such as aphakia, intraocular lens, complete lens dislocation, and intraocular lens with an intraocular lens.

[0078] S3. Based on the stated eye condition, complete the biological measurements.

[0079] Based on the discrimination results of the above-mentioned eye condition discrimination module, the correct biological measurements are performed, or the correct biological parameters are measured after rescanning with the correct eye condition.

[0080] Example 3:

[0081] To illustrate the superiority of this invention compared to existing technologies, this embodiment is provided for verification.

[0082] An analysis of data from 21 clinical patients (23 eyes) revealed significant differences in ocular biological parameters measured using different ocular status measurement modes for the same eye (especially in axial length, with an average difference of 0.21 mm). This resulted in significant differences in the reserved power of intraocular lenses (IOLs) used clinically (an average difference as high as 0.55 D). Axial length measured in aphakic mode was larger than that measured in phakic mode; significant differences also existed in IOL power measurements. (See Table 1.)

[0083] Table 1

[0084]

[0085] In the table, P < 0.05 indicates a statistically significant difference, and the intraocular lens power is calculated based on a margin close to 0D.

[0086] like Figure 3 As shown, the degree and extent of incomplete lens dislocation vary. Specifically, Figure 3 (a) indicates incomplete lens dislocation, with the visual axis area not passing through the lens medium; Figure 3 (b) indicates incomplete lens dislocation, but the visual axis can pass through the lens medium. Incorrect selection of the measurement mode can lead to measurement errors and increase postoperative refractive errors in patients. Therefore, the selection of the optical ocular biology measurement mode should be determined based on whether the visual axis passes through the lens medium.

[0087] 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. An optical biological measurement device for automatically determining the state of the eyeball, characterized in that, include: Pre-scan image acquisition module, eye state discrimination module, and measurement module; The pre-scan image acquisition module is used to acquire standard scan images; The eye state discrimination module is used to discriminate and obtain the eye state based on the scanned image; the eye state refers to the state of the lens; the eye state discrimination module uses different discrimination criteria for different eyes to discriminate the eye state, and the discrimination criteria are as follows: When the ratio of lens area to pupil area is 1, it is determined that there is no lens dislocation. When the ratio of lens area to pupil area is 0, it is determined to be a complete lens dislocation; When the ratio of lens area to pupil area is >0 but <1, it is determined that there is incomplete lens dislocation. When the determination result indicates incomplete lens dislocation, the determination is based on whether the lens is detected at the center of the pupil, according to the following criteria: If the lens is detected at the center of the pupil, select the phakic eye measurement mode; if the lens is not detected at the center of the pupil, select the aphakic eye measurement mode, as detailed below: Analyze the ratio of lens area to pupil area. When the ratio of lens area to pupil area is 1, it is automatically determined that there is no lens dislocation and the lens-containing eye measurement mode is automatically selected. When the ratio of lens area to pupil area is 0, it is automatically identified as complete lens dislocation, and the aphakic eye measurement mode is automatically selected. When the ratio of lens area to pupil area is >0 but <1, it is determined that there is incomplete lens dislocation; the determination is automatically made based on whether the lens position is detected at the center of the pupil; if the lens is detected at the center of the pupil, the measurement mode for phakic eyes is automatically selected. No lens was detected at the center of the pupil, so the aphakic eye measurement mode was automatically selected. The eye state discrimination module includes: an automatic pupil region recognition unit, a pupil region image preprocessing unit, and a pupil region image classification unit; the automatic pupil region recognition unit is used to recognize the pupil region based on the scanned image and obtain the circular outline of the pupil region; the pupil region image preprocessing unit is used to preprocess the circular outline to obtain a processing result; the pupil region image classification unit is used to classify and discriminate based on the processing result to obtain the eye state; The workflow of the automatic pupil region recognition unit includes: converting the acquired color image into a grayscale image and performing Gaussian blurring on the image using a 5x5 Gaussian kernel; capturing edge features in the blurred image by performing Canny edge detection and setting a low threshold of 50 and a high threshold of 150; and locating the circular contour using the Hough transform method based on image gradient information based on edge detection to complete the recognition. The workflow of the pupil region image classification unit includes: training an image classification model using a convolutional neural network with a ResNet50 structure, wherein the initial learning rate is set to 1×10⁻⁶ during training. -3 Simultaneously, a learning rate decay strategy is employed for dynamic adjustment; Adam is selected as the optimizer to update model weights; cross-entropy loss is used as the loss function; and the batch size is set to 64. The image classification model uses the trained model weights to perform forward propagation on the pupil area image, and outputs the probability distribution of each category through the softmax layer, thereby completing the automatic classification and recognition of eye state. The measurement module is used to perform biological measurements based on the state of the eyeball.

2. The optical biological measurement device for automatically determining the state of the eyeball according to claim 1, characterized in that, The lens states include: aphakic, phakic, intraocular lens, ICL, and incomplete lens dislocation.

3. The optical biological measurement device for automatically determining the state of the eyeball according to claim 1, characterized in that, The workflow of the pupil area image preprocessing unit includes: The identified circular outline is segmented to obtain an image of the pupil region; The resolution of the pupil area image was adjusted to 112×112 and normalized to complete the preprocessing.

4. An automatic optical biological measurement method for determining the state of the eyeball, the method being applied to the device according to any one of claims 1-3, characterized in that the steps... include: Acquire standard scan images; Based on the scanned image, the state of the eyeball is determined; the state of the eyeball refers to the state of the lens. The eye state discrimination module uses different discrimination criteria for different eyes to determine the eye state. The discrimination criteria are as follows: When the ratio of lens area to pupil area is 1, it is determined that there is no lens dislocation. When the ratio of lens area to pupil area is 0, it is determined to be a complete lens dislocation; When the ratio of lens area to pupil area is >0 but <1, it is determined that there is incomplete lens dislocation. Biological measurements were performed based on the described eye condition.

5. The optical biological measurement method for automatically determining the state of the eyeball according to claim 4, characterized in that, The lens states include: aphakic, phakic, intraocular lens, ICL, and incomplete lens dislocation.

Images