Optical path refraction automatic adjustment method and device based on fundus coherence tomography scanner

Abstract

The application discloses a kind of optical path diopter automatic adjustment method and device based on fundus coherence tomography, the method includes according to preset initial diopter value fixed diopter position, traverse each optical path position, until can be positioned from retinal image retinal position, and calculate corresponding initial best optical path value;With the initial best optical path value and initial diopter value as starting point, with the motion range of diopter module and optical path module as traversal range, cooperatively traverse optical path position and diopter position, find out the optical path value and diopter value corresponding to the retinal image with the maximum signal intensity as the best optical path value and best diopter value;Drive optical path module and diopter module respectively move to best optical path value and best diopter value.The application focuses on the central region of retina more finely, and higher imaging effect can be obtained.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of retinal image processing, and particularly relates to a light path diopter automatic adjustment method and device based on a fundus coherence tomograph. BACKGROUND

[0002] Optical coherence tomography (OCT) is an imaging technology for forming fundus images. Since it can reflect the reflection and scattering characteristics of different physiological structures of the fundus to incident weak coherent light, the three-dimensional image formed has depth information, and has unique advantages compared with fundus color images.

[0003] In the process of collecting fundus images using OCT, fundus focusing needs to be performed so that it can be focused on key parts of the fundus to achieve the best imaging effect. The current OCT products use a structured light method to perform fundus focusing. The fundus focusing function module contains a structured light module, which is inside the OCT instrument, emits infrared light and splits the light, a part of the infrared light directly enters the image sensor for direct collection, and the other part of the infrared light is first collected into the user's fundus and then collected by the image sensor. At this time, two light spots will be collected in the image collected by the image sensor, and the focusing data is derived by calculating the relative position of the pixels and cooperating with the hardware parameters to achieve the focusing purpose. It can be seen that the existing scheme mainly has the following two shortcomings: (1) the addition of the structured light module will cause problems of heat dissipation and maintainability in system design, and there are precision calibration problems, and the cost will also increase; (2) the final purpose of focusing is to obtain the best imaging effect, but this technical solution relies too much on the calculation of structured light and is difficult to stably obtain the optimal effect. SUMMARY

[0004] In view of the above problems, the application provides a light path diopter automatic adjustment method and device based on a fundus coherence tomograph, which directly obtains image signal intensity by segmenting the foreground of the retinal image, and aligns the focusing position to the retinal area more finely, so that higher imaging effect can be obtained.

[0005] In order to achieve the above technical purposes and achieve the above technical effects, the application realizes the following technical solutions:

[0006] In the first aspect, the application provides a light path diopter automatic adjustment method based on a fundus coherence tomograph, which comprises the following steps:

[0007] Fix the diopter position according to a preset initial diopter value, traverse each light path position, until the retinal position can be located from the retinal image, and the corresponding initial best light path value is calculated;

[0008] Taking the initial optimal optical path value and initial dioptric value as a starting point, and taking the movement range of the dioptric module and the optical path module as a traversal range, the optical path value and the dioptric value corresponding to the retinal image with the maximum signal strength are found out as the optimal optical path value and the optimal dioptric value by traversing the optical path position and the dioptric position.

[0009] The dioptric module and the optical path module are driven to move to the optimal dioptric value and the optimal optical path value, respectively.

[0010] Optionally, the method for obtaining the initial optical path value comprises:

[0011] The dioptric position of the dioptric module is fixed as Q0;

[0012] According to a preset movement trajectory, the optical path module is moved to traverse, and in the movement process, the retinal image is collected in real time; when it is detected that the retinal image contains the retina, the current optical path position d is recorded, and the pixel height h and the concave direction u of the retina are calculated synchronously;

[0013] The optical path module is adjusted to move the retina from the lowermost end to the uppermost end of the retinal image in the preview mode, and the actual movement distance D of the optical path module and the pixel row number H of the retinal image in the preview mode are recorded, and the ratio r of the pixel to the actual movement distance of the optical path module is calculated, r=D / H;

[0014] The initial optimal optical path value is calculated, and the calculation formula of the initial optimal optical path value is:

[0015] G0=u*r*(β*H-h)+d.

[0016] Wherein, β∈[40%,90%], which indicates the position of the retina in the retinal image.

[0017] Optionally, the Q0 is the center of the total dioptric scale; β=75%.

[0018] Optionally, the method for calculating the pixel height h of the retina comprises:

[0019] The retinal image is sent into a trained segmentation model, and the retinal image is segmented into a foreground image and a background image by the segmentation model;

[0020] The longitudinal pixel height of all pixel points in the foreground image is obtained, and the minimum value in all longitudinal pixel heights is calculated as the pixel height h of the retina.

[0021] Optionally, the method for calculating the concave direction u of the retina comprises:

[0022] The retinal image is sent into a trained segmentation model, and the retinal image is segmented into a foreground image and a background image by the segmentation model;

[0023] Traverse the pixel height of the lowest pixel in each column of the foreground image, use the pixel height of each lowest pixel as the vertical coordinate, and perform binomial fitting y = a*x^2 + b*x + c. If a ≥ 0, the direction of the retinal concavity is positive, otherwise it is negative.

[0024] Optionally, the method for finding the optimal optical path value and the optimal refractive index value includes:

[0025] The optical path module and the refractive module are moved in coordination so that their positions are moved from the initial optimal optical path value Q0 and the initial refractive value G0 to the initial positions Q1 and G1, wherein when Q0>G0, Q1=0, G1=G0-Q0; when Q0≤G0, the refractive index is Q1=Q0-G0, G1=0;

[0026] The optical path position and refractive position are traversed collaboratively, and the retinal image is collected in real time. The optical path module and the refractive module move the same distance at each moment, which is recorded as Δq, and the corresponding optical path value G is obtained by the optical path position sensor, refractive position sensor, and image acquisition device at the same frequency. i , refractive value Q i and retinal image I i ;

[0027] Based on each retinal image I i The signal strength of the retinal image with the largest signal strength is found. i The corresponding optical path value and refractive index are taken as the optimal optical path value and optimal refractive index.

[0028] Optionally, the signal strength calculation method includes:

[0029] The retinal image is fed into a trained segmentation model, and the segmentation model segments the retinal image into a foreground image and a background image;

[0030] The pixel mean of the foreground image was taken as the signal intensity of the retinal image.

[0031] Optionally, the retinal image I with the maximum signal intensity i Methods for finding include:

[0032] Record the signal intensity of the first retinal image I1, and use its corresponding optical path value and refractive index as the local peak point;

[0033] Traverse the subsequent retinal images I i If the signal intensity is higher than 1, the local peak point is updated until no new local peak appears. The local peak at this time is defined as the global signal peak, and the optical path value and refractive value at this time are recorded as the optimal optical path value Q. max and the best refractive value Gmax .

[0034] In a second aspect, the present invention provides an optical path refraction automatic adjustment device based on a fundus coherence tomography scanner, comprising: an adjuster, an optical path module, an optical path module driver, a refraction module, a refraction module driver, and an image acquisition device;

[0035] The regulator controls the refractive module driver to fix the refractive position according to a preset initial refractive value, and controls the optical path module driver to traverse each optical path position until the retinal position is located based on the retinal image captured by the image acquisition device, and the corresponding initial optimal optical path value is calculated;

[0036] Taking the initial optimal optical path value and initial refractive index as a starting point and the motion range of the refractive index module and the optical path module as a traversal range, the regulator controls the optical path module driver and the refractive index module driver to cooperatively drive the optical path module and the refractive index module to traverse the optical path positions and refractive index positions, and finds the optical path value and refractive index corresponding to the retinal image with the maximum signal intensity, which are used as the optimal optical path value and optimal refractive index;

[0037] The regulator controls the optical path module driving member and the refractive module driving member to drive the optical path module and the refractive module to move to the optimal optical path value and the optimal refractive value respectively.

[0038] Optionally, the optical path refraction automatic adjustment device also includes an optical path position sensor and a refraction position sensor, which are respectively connected to the adjuster. The optical path position sensor is used to measure the optical path position of the optical path module in real time; the refraction position sensor is used to measure the refraction position of the refraction module in real time.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] The present invention directly evaluates the imaging quality of the retinal image, and uses the image signal intensity obtained at each moment as an evaluation index to traverse the optical path refraction, directly obtaining the best image effect in the entire working range, which has a more stable effect than the indirect acquisition method of structured light.

[0041] The technical solution of the present invention does not rely on a structured light device, but only relies on the imaging equipment of a fundus coherence tomography scanner, thereby reducing hardware costs.

[0042] The present invention does not require a structured light device, thereby reducing power consumption and increasing the life of the entire device.

[0043] Equipment movement may cause the structured light device to loosen. In the present invention, there is no need to set up a structured light device, so calibration work can be avoided and maintainability is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to make the content of the present application more easily understood, the present application is further described in detail below according to specific embodiments and in conjunction with the accompanying drawings, in which:

[0045] Figure 1 Fig. 1 is a flowchart of an optical path refraction automatic adjustment method according to an embodiment of the present application;

[0046] Fig. 2(a) is a schematic diagram of a low signal intensity retinal image according to an embodiment of the present application;

[0047] Fig. 2(b) is a schematic diagram of a medium signal intensity retinal image according to an embodiment of the present application;

[0048] Fig. 2(c) is a schematic diagram of a high signal intensity retinal image according to an embodiment of the present application;

[0049] Fig. 3(a) is a schematic diagram of a collected retinal image according to an embodiment of the present application;

[0050] Fig. 3(b) is a label image corresponding to the retinal image in Fig. 3(a);

[0051] Figure 4 Fig. 4 is a diagram of the walking routes of the optical path module and the refraction module in the focusing process according to an embodiment of the present application. DETAILED DESCRIPTION

[0052] In order to make the content of the present application more easily understood, the present application is further described in detail below according to specific embodiments and in conjunction with the accompanying drawings, in which:

[0053] The application principle of the present application is described in detail below in conjunction with the accompanying drawings.

[0054] Embodiment 1

[0055] An optical path refraction automatic adjustment method based on a fundus coherence tomography instrument is provided in the embodiments of the present application, as shown in the following steps: Figure 1

[0056] (1) Fix the refraction position according to a preset initial refraction value, traverse each optical path position until the retinal position is located from the retinal image, and calculate the corresponding initial best optical path value;

[0057] (2) Take the initial best optical path value and the initial refraction value as the starting point, take the movement range of the refraction module and the optical path module as the traversal range, and cooperatively traverse the optical path position and the refraction position to find the optical path value and the refraction value corresponding to the retinal image with the maximum signal intensity as the best optical path value and the best refraction value;

[0058] ​(2) Drive the optical path module and the refractive module to move to the optimal optical path value and the optimal refractive value respectively.

[0059] In a specific implementation of the embodiment of the present invention, the method for obtaining the initial optical path value includes:

[0060] Fix the refractive position of the refractive module to Q0; in a specific implementation, Q0 is generally selected as the center of the total refractive scale;

[0061] According to the preset moving trajectory, the moving optical path module is traversed, and the retinal image is collected in real time during the movement. When the presence of the retina is detected in the retinal image, the current optical path position d is recorded, and the pixel height h and the concave direction u of the retina are synchronously calculated. In the specific implementation process, specifically: the movement mode of the optical path module is linear motion, its movement range is 0-100mm, and the measurement accuracy is 1mm. The initial position of the moving trajectory of the optical path module is 0mm, and the corresponding retinal image is a full-noise image. Then, the moving traversal is started. During the traversal process, it is detected whether there is a retina in each retinal image. If there is a retina, the pixel height h and the concave direction u of the retina are calculated, and the current optical path position d is recorded;

[0062] Adjust the optical path module to move the retina from the bottom to the top of the retinal image in the preview mode, record the actual movement distance D of the optical path module and the number of pixel rows H of the retinal image in the preview mode, and calculate the ratio r of pixels to the actual movement distance of the optical path module, r = D / H;

[0063] Calculate the initial optimal optical path value, the calculation formula of the initial optimal optical path value is:

[0064] G0=u*r*(β*Hh)+d.

[0065] Wherein, β∈[40%, 90%] refers to the position of the retina in the retinal image, and when it is close to 100%, the retina is below the image. Preferably, β=75%.

[0066] In a specific implementation of the embodiment of the present invention, the method for calculating the retinal pixel height h includes the following steps:

[0067] The retinal image is fed into a trained segmentation model, and the segmentation model segments the retinal image into a foreground image and a background image;

[0068] Get the vertical pixel heights of all pixels in the foreground image, and calculate the minimum value of all heights as the pixel height h of the retina.

[0069] The method for calculating the retinal concave direction u includes:

[0070] The retinal image is fed into a trained segmentation model, and the segmentation model segments the retinal image into a foreground image and a background image;

[0071] Traverse the pixel height of the lowest pixel in each column of the foreground image, use the pixel height of each lowest pixel as the vertical coordinate, and perform binomial fitting y = a*x^2 + b*x + c. If a ≥ 0, the direction of the retinal concavity is positive, otherwise it is negative.

[0072] In a specific implementation of the embodiment of the present invention, the method for finding the best refractive value includes:

[0073] The optical path module and the refractive module are moved in coordination so that their positions move from the initial optimal optical path value Q0 and the initial refractive value G0 to the initial positions Q1 and G1, wherein when Q0>G0, Q1=0, G1=G0-Q0; when Q0≤G0, the refractive index is Q1=Q0-G0, G1=0;

[0074] The optical path position and refractive position are traversed collaboratively, and the retinal image is collected in real time. The optical path module and the refractive module move the same distance at each moment, which is recorded as Δq, and the corresponding optical path value G is obtained by the optical path position sensor, refractive position sensor, and image acquisition device at the same frequency. i , refractive value Q i and retinal image I i ;

[0075] Based on each retinal image I i The signal strength of the retinal image with the largest signal strength is found. i The corresponding optical path value and refractive index are taken as the optimal optical path value and optimal refractive index.

[0076] In a specific implementation of the embodiment of the present invention, the retinal image I with the largest signal intensity i Methods for finding include:

[0077] Record the signal intensity of the first retinal image I1, and use its corresponding optical path value and refractive index as the local peak point;

[0078] Traverse the subsequent retinal images I i If a higher signal intensity appears, the local peak point is updated until no new local peak appears. The local peak at this time is defined as the global signal peak, and the optical path value and refractive value at this time are recorded as the optimal optical path value Q. max and the best refractive value G max .

[0079] The signal strength calculation method includes:

[0080] The retinal image is fed into a trained segmentation model, and the segmentation model segments the retinal image into a foreground image and a background image;

[0081] The pixel mean of the foreground image was taken as the signal intensity of the retinal image.

[0082] The method in the embodiment of the present invention is described in detail below with reference to a specific implementation manner.

[0083] The optical path refraction automatic adjustment method in the embodiment of the present invention mainly includes two stages, specifically:

[0084] Stage 1: locating the retinal position;

[0085] Stage 2: Enhance retinal signal stage.

[0086] The retinal location phase includes the following steps:

[0087] 1. Data collection.

[0088] Acquire retinal images from multiple individuals to achieve the goal of covering actual usage scenarios and the goal of retinal image diversity. The acquired retinal images should at least include retinal images with low signal intensity, retinal images with medium signal intensity, and retinal images with high signal intensity. The retinal position can be anywhere from the top to the bottom of the retinal image, the retina can be folded near the edge, and various combinations thereof. Among them, the intensity range of retinal images with low signal intensity is [0%, 10%), the intensity range of retinal images with medium signal intensity is [10%, 60%), and the intensity range of retinal images with high signal intensity is [60%, 99%). For details, see Figure 2(a)-Figure 2(c) .

[0089] 2. Data labeling.

[0090] The retinal image is divided into foreground and background using image segmentation. The retinal part (including the optic disc area) present in the retinal image is marked as the foreground image, and the part outside the retina and the full noise image without retina are marked as the background image. For details, see Figure 3(a)-Figure 3(b) .

[0091] 3. Model construction and training.

[0092] Considering the speed and functional requirements in actual scenarios, we use Tiny-unet as the basic network. The original retinal images and the annotated retinal images (i.e., labels) are downsampled to 256*256 size and sent to Tiny-unet for training to obtain the segmentation model.

[0093] 4. Fix the position of the refractive module at Q0 (usually selected as the center of the total refractive scale), traverse each optical path position until the retinal position can be located from the retinal image, and calculate the retinal pixel position h, the retinal depression direction u, and record the optical path position d when the retina appears, and then calculate the optimal optical path position G0 = u*r*(β*Hh)+d in the first stage.

[0094] Wherein, obtaining the pixel position h of the retina in the retinal image includes the following steps:

[0095] Feed the retinal image into the segmentation model to obtain the foreground image;

[0096] Get the vertical pixel heights of all pixels in the foreground image, and calculate the minimum value of all heights as the pixel position h of the retina.

[0097] Obtaining the concave direction u of the retina in the retinal image includes the following steps:

[0098] Feed the retinal image into the segmentation model to obtain the foreground image;

[0099] Traverse the pixel height of the lowest pixel in each column of the foreground image, use the pixel height of each lowest pixel as the vertical coordinate, and perform binomial fitting y = a*x^2 + b*x + c. If a ≥ 0, the retinal concavity direction is positive, otherwise it is negative;

[0100] Obtaining the pixel-to-optical path length ratio scale r includes the following steps:

[0101] Adjust the optical path module to move the retina from the bottom to the top of the previewed retinal image, record the actual movement distance D of the optical path module and the number of pixel rows H of the previewed retinal image, and calculate the ratio r of pixels to the actual movement distance of the optical path module, r = D / H.

[0102] During the movement of the optical path module, it is generally chosen to move the optical path module to place the retina position in the middle and lower part of the retinal image, that is, 75% of the retinal image from top to bottom. Positioning the retinal area in the lower part of the retinal image can obtain better imaging effects, and the 75% position is at a certain distance from the bottom of the retinal image to ensure that the retina does not fold, so β ​​is generally set to 75%.

[0103] The retinal signal enhancement stage includes the following steps:

[0104] Move the optical path module and the refractive module to the initial positions Q1 and G1. Moving the refractive module with a fixed optical path module (or moving the optical path module with a fixed refractive module) will destroy the retina that has been positioned in the first stage. Therefore, the optical path module and the refractive module need to be moved synchronously to allow the retina in the fixed position to traverse the signal intensity. That is, the refractive module, like the optical path module, moves in a linear motion. Its position scale range is 0-100mm, and its positioning accuracy is 1mm. When the optical path module and the refractive module move forward by the same distance at the same time, the retinal position will not change, only the signal intensity will change. There are two situations for moving the initial position: when Q0>G0, Q1=0, G1=G0-Q0; when Q0≤G0, the refractive index is Q1=Q0-G0, G1=0; see for details. Figure 4 ;

[0105] The same distance that the optical path and refraction move at each moment is recorded as Δq. The optical path value G at this moment is obtained by the optical path position sensor, refraction position sensor, and image acquisition device. i , refractive value Q i and retinal image I i ;

[0106] Calculate the signal intensity of the retinal image at each moment, use the hill climbing method to find the signal peak, record the signal intensity of the first retinal image I1, and use its corresponding optical path value and refractive index as the local peak point. Continuously record the subsequent 20mm range. If a higher signal intensity appears during the recording process, update the local peak point and refresh it cyclically until no new local peak appears. The local peak at this time is defined as the global signal peak and the optical path and refractive index at this time are recorded as G respectively. max and Q max ;

[0107] Move the diopter module and optical path module to G max and Q max , collect retinal images.

[0108] Example 2

[0109] An embodiment of the present invention provides an optical path refraction automatic adjustment device, comprising: an adjuster, an optical path module, an optical path module driver, a refraction module, a refraction module driver, and an image acquisition device;

[0110] The regulator controls the refractive module driver to fix the refractive position according to a preset initial refractive value, and controls the optical path module driver to traverse each optical path position until the retinal position is initially located based on the retinal image captured by the image acquisition device, and the corresponding initial optimal optical path value is calculated;

[0111] Taking the initial optimal optical path value and initial refractive index as a starting point and the motion range of the refractive index module and the optical path module as a traversal range, the regulator controls the optical path module driver and the refractive index module driver to cooperatively drive the optical path module and the refractive index module to traverse the optical path position and the refractive index position, and finds the optical path value and the refractive index corresponding to the retinal image with the maximum signal intensity, and uses them as the optimal optical path value and the optimal refractive index;

[0112] The regulator controls the optical path module driving member and the refractive module driving member to drive the optical path module and the refractive module to move to the optimal optical path value and the optimal refractive value respectively.

[0113] In a specific implementation of an embodiment of the present invention, the optical path refraction automatic adjustment device also includes an optical path position sensor and a refraction position sensor, which are respectively connected to the adjuster. The optical path position sensor is used to measure the optical path position of the optical path module in real time; the refraction position sensor is used to measure the refraction position of the refraction module in real time.

[0114] The optical path refraction automatic adjustment device in the embodiment of the present invention mainly includes two working stages, specifically:

[0115] Stage 1: locating the retinal position;

[0116] Stage 2: Enhance retinal signal stage.

[0117] The retinal location phase includes the following steps:

[0118] 1. Data collection.

[0119] Acquire retinal images from multiple individuals to achieve the goal of covering actual usage scenarios and the goal of retinal image diversity. The acquired retinal images should at least include retinal images with low signal intensity, retinal images with medium signal intensity, and retinal images with high signal intensity. The retinal position can be anywhere from the top to the bottom of the retinal image, the retina can be folded near the edge, and various combinations thereof. Among them, the intensity range of retinal images with low signal intensity is [0%, 10%), the intensity range of retinal images with medium signal intensity is [10%, 60%), and the intensity range of retinal images with high signal intensity is [60%, 99%). For details, see Figure 2(a)-Figure 2(c) .

[0120] 2. Data labeling.

[0121] The retinal image is divided into foreground and background using image segmentation. The retinal part (including the optic disc area) present in the retinal image is marked as the foreground image, and the part outside the retina and the full noise image without retina are marked as the background image. For details, see Figure 3(a)-Figure 3(b) .

[0122] 3. Model construction and training.

[0123] Considering the speed and functional requirements in actual scenarios, we use Tiny-unet as the basic network. The original retinal images and the annotated retinal images (i.e., labels) are downsampled to 256*256 size and sent to Tiny-unet for training to obtain the segmentation model.

[0124] 4. Fix the position of the refractive module at Q0 (usually selected as the center of the total refractive scale), traverse each optical path position until the retinal position can be located from the retinal image, and calculate the retinal pixel position h, the retinal depression direction u, and record the optical path position d when the retina appears, and then calculate the optimal optical path position G0 = u*r*(β*Hh)+d in the first stage.

[0125] Wherein, obtaining the pixel position h of the retina in the retinal image includes the following steps:

[0126] Feed the retinal image into the segmentation model to obtain the foreground image;

[0127] Get the vertical pixel heights of all pixels in the foreground image, and calculate the minimum value of all heights as the pixel position h of the retina.

[0128] Obtaining the concave direction u of the retina in the retinal image includes the following steps:

[0129] Feed the retinal image into the segmentation model to obtain the foreground image;

[0130] Traverse the pixel height of the lowest pixel in each column of the foreground image, use the pixel height of each lowest pixel as the vertical coordinate, and perform binomial fitting y = a*x^2 + b*x + c. If a ≥ 0, the retinal concavity direction is positive, otherwise it is negative;

[0131] Obtaining the pixel-to-optical path length ratio scale r includes the following steps:

[0132] Adjust the optical path module to move the retina from the bottom to the top of the previewed retinal image, record the actual movement distance D of the optical path module and the number of pixel rows H of the previewed retinal image, and calculate the ratio r of pixels to the actual movement distance of the optical path module, r = D / H.

[0133] During the movement of the optical path module, it is generally chosen to move the optical path module to place the retina position in the middle and lower part of the retinal image, that is, 75% of the retinal image from top to bottom. Positioning the retinal area in the lower part of the retinal image can obtain better imaging effects, and the 75% position is at a certain distance from the bottom of the retinal image to ensure that the retina does not fold, so β ​​is generally set to 75%.

[0134] The retinal signal enhancement stage includes the following steps:

[0135] Move the optical path module and the refractive module to the initial positions Q1 and G1. Moving the refractive module with a fixed optical path module (or moving the optical path module with a fixed refractive module) will destroy the retina that has been positioned in the first stage. Therefore, the optical path module and the refractive module need to be moved synchronously to allow the retina in the fixed position to traverse the signal intensity. That is, the refractive module, like the optical path module, moves in a linear motion. Its position scale range is 0-100mm, and its positioning accuracy is 1mm. When the optical path module and the refractive module move forward by the same distance at the same time, the retinal position will not change, only the signal intensity will change. There are two situations for moving the initial position: when Q0>G0, Q1=0, G1=G0-Q0; when Q0≤G0, the refractive index is Q1=Q0-G0, G1=0; see for details. Figure 4 ;

[0136] The same distance that the optical path and refraction move at each moment is recorded as Δq. The optical path value G at this moment is obtained by the optical path position sensor, refraction position sensor, and image acquisition device. i , refractive value Q i and retinal image I i ;

[0137] Calculate the signal intensity of the retinal image at each moment, use the hill climbing method to find the signal peak, record the signal intensity of the first retinal image I1, and use its corresponding optical path value and refractive index as the local peak point. Continuously record the subsequent 20mm range. If a higher signal intensity appears during the recording process, update the local peak point and refresh it cyclically until no new local peak appears. The local peak at this time is defined as the global signal peak and the optical path and refractive index at this time are recorded as G respectively. max and Q max ;

[0138] Move the diopter module and optical path module to G max and Q max , collect retinal images.

[0139] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for automatic adjustment of optical path refraction based on fundus coherence tomography scanner, characterized in that: include: The refractive position is fixed according to the preset initial refractive value, and each optical path position is traversed until the retinal position can be located from the retinal image and the corresponding initial optimal optical path value is calculated; Taking the initial optimal optical path value and the initial refractive value as the starting point, and taking the movement range of the refractive module and the optical path module as the traversal range, collaboratively traversing the optical path position and the refractive position, and finding the optical path value and the refractive value corresponding to the retinal image with the maximum signal intensity as the optimal optical path value and the optimal refractive value, including: collaboratively moving the optical path module and the refractive module so that the positions of the two are moved from the initial optimal optical path value Q0 and the initial refractive value G0 to the initial positions Q1 and G1, wherein when Q0>G0, Q1=0, G1=G0-Q0; when Q0≤G0, Q1=Q0-G0, G1=0; The optical path position and refractive position are traversed collaboratively, and the retinal image is collected in real time. The optical path module and the refractive module move the same distance at each moment, which is recorded as Δq, and the corresponding optical path value G is obtained by the optical path position sensor, refractive position sensor, and image acquisition device at the same frequency. i , refractive value Q i and retinal image I i ; Based on each retinal image I i The signal strength of the retinal image with the largest signal strength is found. i The corresponding optical path value and refractive index are taken as the optimal optical path value and optimal refractive index; The optical path module and the refractive module are driven to move to the optimal optical path value and the optimal refractive value respectively.

2. The method for automatically adjusting optical path refraction based on a fundus coherence tomography scanner according to claim 1, characterized in that: The method for obtaining the initial optimal optical path value includes: Fix the refractive position of the refractive module to Q0; According to the preset movement trajectory, the optical path module is traversed, and the retinal image is collected in real time during the movement. When the retina is detected in the retinal image, the current optical path position d is recorded, and the pixel height h and concave direction u of the retina are calculated synchronously; Adjust the optical path module to move the retina from the bottom to the top of the retinal image in the preview mode, record the actual movement distance D of the optical path module and the number of pixel rows H of the retinal image in the preview mode, and calculate the ratio r of pixels to the actual movement distance of the optical path module, r = D / H; Calculate the initial optimal optical path value, the calculation formula of the initial optimal optical path value is: G0=u*r*(β*Hh)+d; Among them, β∈[40%,90%] refers to the position of the retina in the retinal image.

3. The method for automatically adjusting optical path refraction based on a fundus coherence tomography scanner according to claim 2, characterized in that: Q0 is the center of the total refractive scale; β=75%.

4. The method for automatically adjusting optical path refraction based on a fundus coherence tomography scanner according to claim 2, characterized in that: The method for calculating the retinal pixel height h includes: The retinal image is fed into a trained segmentation model, and the segmentation model segments the retinal image into a foreground image and a background image; The vertical pixel heights of all pixels in the foreground image are obtained, and the minimum value of all vertical pixel heights is calculated as the pixel height h of the retina.

5. The method for automatically adjusting optical path refraction based on a fundus coherence tomography scanner according to claim 2, characterized in that: The method for calculating the retinal concave direction u includes: The retinal image is fed into a trained segmentation model, and the segmentation model segments the retinal image into a foreground image and a background image; Traverse the pixel height of the lowest pixel in each column of the foreground image, use the pixel height of each lowest pixel as the vertical coordinate, and perform binomial fitting y = a*x^2 + b*x + c. If a ≥ 0, the direction of the retinal concavity is positive, otherwise it is negative.

6. The method for automatically adjusting optical path refraction based on a fundus coherence tomography scanner according to claim 1, characterized in that: The signal strength calculation method includes: The retinal image is fed into a trained segmentation model, and the segmentation model segments the retinal image into a foreground image and a background image; The pixel mean of the foreground image was taken as the signal intensity of the retinal image.

7. The method for automatically adjusting optical path refraction based on a fundus coherence tomography scanner according to claim 1 or 6, characterized in that: The retinal image I with the maximum signal intensity i Methods for finding include: Record the signal intensity of the first retinal image I1, and use its corresponding optical path value and refractive value as the local peak point; traverse the subsequent retinal images I i If the signal intensity is higher than 1, the local peak point is updated until no new local peak appears. The local peak at this time is defined as the global signal peak, and the optical path value and refractive value at this time are recorded as the optimal optical path value Q. max and the best refractive value G max .

8. An automatic optical path refraction adjustment device based on a fundus coherence tomography scanner, characterized in that: include: Regulator, optical path module, optical path module driver, refractive module, refractive module driver and image acquisition equipment; The regulator controls the refractive module driver to fix the refractive position according to a preset initial refractive value, and controls the optical path module driver to traverse each optical path position until the retinal position is located based on the retinal image captured by the image acquisition device and the corresponding initial optimal optical path value is calculated; Taking the initial optimal optical path value and the initial refractive value as the starting point, and the movement range of the refractive module and the optical path module as the traversal range, the regulator controls the optical path module driver and the refractive module driver to collaboratively drive the optical path module and the refractive module to traverse the optical path position and the refractive position, and finds the optical path value and the refractive value corresponding to the retinal image with the maximum signal intensity as the optimal optical path value and the optimal refractive value, including: collaboratively moving the optical path module and the refractive module from the initial optimal optical path value Q0 and the initial refractive value G0 to the initial positions Q1 and G1, when Q0>G0, Q1=0, G1=G0-Q0; when Q0≤G0, Q1=Q0-G0, G1=0; collaboratively traversing the optical path position and the refractive position, collecting retinal images in real time, the optical path module and the refractive module move the same distance at each moment, recorded as Δq, and obtaining the corresponding optical path value G respectively through the optical path position sensor, the refractive position sensor, and the image acquisition device at the same frequency. i , refractive value Q i and retinal image I i ; Based on each retinal image I i The signal strength of the retinal image with the largest signal strength is found. i The corresponding optical path value and refractive index are taken as the optimal optical path value and optimal refractive index; The regulator controls the optical path module driving component and the refractive module driving component to drive the optical path module and the refractive module to move to an optimal optical path value and an optimal refractive value.

9. The optical path refraction automatic adjustment device based on a fundus coherence tomography scanner according to claim 8, characterized in that: The optical path refraction automatic adjustment device also includes an optical path position sensor and a refraction position sensor, both of which are connected to the regulator respectively. The optical path position sensor is used to measure the optical path position of the optical path module in real time; the refraction position sensor is used to measure the refraction position of the refraction module in real time.

Images