A method and system for measuring retinal blood flow velocity

Abstract

The present invention discloses a method and system for measuring retinal blood flow velocity, which are applied to the technical field of data processing. The method includes: obtaining a multi-view video set by collecting multi-view videos of the retina; performing localization of the blood vessel area to be measured for the target user to obtain a regional localization result; obtaining a binary vascular image, obtaining the blood vessel wall thickness, and extracting and obtaining a vascular centerline image, and setting a blood flow trace extraction window; obtaining a basic calculation background image, calculating a division image of the background image, constructing an image pyramid of the cell sequence, and tracking from any layer of the image pyramid to obtain the optical flow offset of pixel points; eliminating optical flow abnormal points of the pixel point optical flow offset based on reverse detection of optical flow, obtaining target cells according to the elimination and screening results, and calculating the blood flow velocity according to the target cells. The technical problem that the existing method for measuring retinal blood flow velocity has low measurement accuracy for measuring the blood flow velocity of multi-level blood vessels in the fundus of the eye is solved.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular, to a method and system for measuring retinal blood flow velocity. Background Art

[0002] As an index reflecting the comprehensive effect of vascular morphological changes and other systemic influencing factors, blood flow velocity has very important clinical significance and practical value in ophthalmology, including helping to further clarify the diagnosis and treatment of diseases such as ocular vascular dilation, fundus vascular occlusion tumors, etc. In the early stage of diabetic retinopathy, the blood flow velocities of the central retinal artery and the short posterior ciliary artery are significantly decreased, and as the disease progresses, the blood flow abnormality will worsen. Using color Doppler ultrasound to measure the blood flow velocity of the fundus circulation helps in early diagnosis, etc. Therefore, analyzing the blood flow velocity of the eye not only helps in diagnosing fundus diseases, but is also an effective auxiliary diagnosis and treatment method for systemic diseases. By analyzing the fundus structure, vascular morphology and function, it can provide a reliable basis for the staging diagnosis of diseases, the treatment medication for diseases and the prognosis estimation of diseases, and provides an important basis for the prediction and diagnosis of ophthalmic diseases as well as diseases such as diabetes, hypertension, heart disease, thrombosis, etc., which has very important significance in clinical practice. However, the commonly used flow velocity measurement method in the prior art mainly uses the cross-correlation method. By selecting a section of blood vessel segment as a template in the current frame and finding the blood vessel area with the highest correlation with this blood vessel segment in its two consecutive frame images, the movement distance of the template is obtained, and the final blood flow velocity is obtained by dividing by the time interval between the two frame images. This method is widely used, but it is sensitive to noise, requires high imaging quality of the video, and when the current blood cell contrast is low and the flow velocity is fast, the calculation result error is large.

[0003] Therefore, in the prior art, the method for measuring retinal blood flow velocity has the technical problem of low measurement accuracy for measuring the blood flow velocity of multi-level blood vessels in the fundus. Summary of the Invention

[0004] The present application solves the technical problem of low measurement accuracy for measuring the blood flow velocity of multi-level blood vessels in the fundus by providing a method and system for measuring retinal blood flow velocity.

[0005] The present application provides a method for measuring retinal blood flow velocity. The method includes: performing multi-view video acquisition of the retina of a target user to obtain a multi-view video set; positioning the area of the blood vessel to be measured of the target user based on the multi-view video set to obtain a regional positioning result; extracting the blood vessel contour based on video frame difference according to the regional positioning result and the multi-view video set to obtain a binary image of the blood vessel contour; calculating the blood vessel wall thickness based on the binary image of the blood vessel contour and extracting the blood vessel centerline image; setting a blood flow trace extraction window through the blood vessel centerline image; obtaining a basic calculation background image through the multi-view video set and calculating a division image of the basic calculation background image; constructing an image pyramid of cell sequences based on the basic calculation background image and the division image, and tracking from any layer of the image pyramid, and obtaining the pixel optical flow offset according to the tracking result; eliminating the optical flow abnormal points of the pixel optical flow offset based on reverse detection optical flow, obtaining target cells according to the elimination and screening results, and calculating the blood flow velocity according to the target cells.

[0006] The present application also provides a system for measuring retinal blood flow velocity. The system includes: a video set acquisition module for performing multi-view video acquisition of the retina of a target user to obtain a multi-view video set; a regional positioning result acquisition module for positioning the area of the blood vessel to be measured of the target user based on the multi-view video set to obtain a regional positioning result; a binary image acquisition module for extracting the blood vessel contour based on video frame difference according to the regional positioning result and the multi-view video set to obtain a binary image of the blood vessel contour; a centerline image acquisition module for calculating the blood vessel wall thickness based on the binary image of the blood vessel contour and extracting the blood vessel centerline image; a blood flow trace extraction window acquisition module for setting a blood flow trace extraction window through the blood vessel centerline image; a background image acquisition module for obtaining a basic calculation background image through the multi-view video set and calculating a division image of the basic calculation background image; a pixel optical flow offset acquisition module for constructing an image pyramid of cell sequences based on the basic calculation background image and the division image, and tracking from any layer of the image pyramid, and obtaining the pixel optical flow offset according to the tracking result; a blood flow velocity calculation module for eliminating the optical flow abnormal points of the pixel optical flow offset based on reverse detection optical flow, obtaining target cells according to the elimination and screening results, and calculating the blood flow velocity according to the target cells.

[0007] The present application also provides a fundus camera, which includes the above-mentioned system for measuring retinal blood flow velocity.

[0008] The present application also provides an electronic device, including:

[0009] A memory for storing executable instructions;

[0010] A processor, when executing the executable instructions stored in the memory, implements a method for measuring retinal blood flow velocity provided in an embodiment of the present application.

[0011] It is intended to propose a method and system for measuring retinal blood flow velocity through the present application. By collecting multi-view videos of the retina, a multi-view video set is obtained. And the region to be measured of blood vessels of the target user is located to obtain a region location result. A binary image of blood vessels is obtained, the blood vessel wall thickness is obtained, and a blood vessel centerline image is extracted and obtained, and a blood flow trace extraction window is set. A basic calculation background image is obtained, and a division image of the background image is calculated, an image pyramid of the cell sequence is constructed, and tracking is performed from any layer of the image pyramid to obtain the optical flow offset of pixel points. Based on the reverse detection optical flow, the optical flow abnormal points of the optical flow offset of pixel points are eliminated, the target cells are obtained according to the elimination and screening results, and the blood flow velocity is calculated according to the target cells. The accurate measurement of the blood flow velocity of multi-level blood vessels in the fundus is realized. The technical problem that the existing method for measuring retinal blood flow velocity has low measurement accuracy for measuring the blood flow velocity of multi-level blood vessels in the fundus is solved.

[0012] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the description. And in order to make the above and other purposes, features and advantages of the present application more obvious and understandable, the specific embodiments of the present application are specifically exemplified below. Description of the Drawings

[0013] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings of the embodiments of the present disclosure will be briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present disclosure and do not limit the present disclosure.

[0014] Figure 1 It is a schematic flowchart of a method for measuring retinal blood flow velocity provided in an embodiment of the present application;

[0015] Figure 2 It is a schematic flowchart of obtaining a multi-view video set by a method for measuring retinal blood flow velocity provided in an embodiment of the present application;

[0016] Figure 3 It is a schematic flowchart of obtaining the blood vessel wall thickness by a method for measuring retinal blood flow velocity provided in an embodiment of the present application;

[0017] Figure 4 It is a schematic structural diagram of a system for a method for measuring retinal blood flow velocity provided in an embodiment of the present application;

[0018] Figure 5A schematic structural diagram of an electronic device of a system for a method of measuring retinal blood flow velocity provided by an embodiment of the present invention.

[0019] Explanation of reference numerals: Video set acquisition module 11, Region positioning result acquisition module 12, Binary image acquisition module 13, Centerline image acquisition module 14, Blood flow trace extraction window acquisition module 15, Background image acquisition module 16, Pixel point optical flow offset acquisition module 17, Blood flow velocity calculation module 18. Detailed implementation manners

[0020] Embodiment 1

[0021] In order to make the objectives, technical solutions, and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be construed as limitations on the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present application.

[0022] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0023] In the following description, the terms "first\second\third" are only used to distinguish similar objects and do not represent a specific order for the objects. It can be understood that "first\second\third" can be interchanged with a specific order or sequence when allowed, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application.

[0025] Although various references are made to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on a user terminal and / or a server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0026] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the operations before or below do not necessarily need to be executed precisely in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or several operations can be removed from these processes.

[0027] As Figure 1 shown, an embodiment of the present application provides a method for measuring retinal blood flow velocity. The method includes:

[0028] S10: Perform multi-view video acquisition of the retina of the target user to obtain a multi-view video set;

[0029] S20: Locate the area of the blood vessels to be measured of the target user based on the multi-view video set to obtain a regional location result;

[0030] S30: Extract the blood vessel contour based on the video frame difference according to the regional location result and the multi-view video set to obtain a binary image of the blood vessel contour;

[0031] S40: Calculate the blood vessel wall thickness based on the binary image of the blood vessel contour and extract the blood vessel centerline image;

[0032] Specifically, perform multi-view video acquisition of the retina of the target user to obtain a multi-view video set. Subsequently, locate the area of the blood vessels to be monitored of the target user in the multi-view video set to obtain the regional location result of the blood vessels to be monitored of the target user. Further, according to the obtained blood vessel regional location result and the multi-view video set, extract the blood vessel contour based on the video frame difference. The video frame difference is a method of obtaining the contour of a moving object by performing a difference operation on two adjacent frames in a video image frame. Since the shooting device needs to be moved when obtaining the multi-view video, the method of video frame difference can better achieve the extraction of the blood vessel contour, and then obtain the binary image of the blood vessel contour. Subsequently, calculate the blood vessel wall thickness based on the binary image of the blood vessel contour and extract the blood vessel centerline image.

[0033] As Figure 2 shown, step S10 of the method provided by the embodiment of the present application further includes:

[0034] S11: Collect the retinal image of the target user at the first wide angle and the first angle view, and use the collected image as a positioning image;

[0035] S12: Perform fundus video acquisition of the target user at the second angle view to obtain a measurement video;

[0036] S13: Obtain the multi-view video set through the positioning image and the measurement video.

[0037] Specifically, when acquiring the multi-perspective video set, the retinal image of the target user is collected at the first wide angle and the first angular perspective, and the collected image is used as the positioning image. Subsequently, fundus videos of the target user are collected at the second angular perspective to obtain the measurement videos. Further, the multi-perspective video set is obtained through the positioning image and the measurement videos, thus completing the acquisition of the multi-perspective video set. Preferably, the first wide angle is 100°, the first angular perspective is 60°, and the second angular perspective is 8°.

[0038] The method S10 provided by the embodiment of the present application further includes:

[0039] S14: Construct the fovea centralis feature, match the positioning origin of the positioning image through the fovea centralis feature, and calculate the pixel deviation between the center of the area to be measured and the positioning origin according to the positioning origin matching result;

[0040] S15: Obtain the area positioning result according to the pixel deviation;

[0041] S16: Calculate the phase deviation between the image to be registered and the reference image, and perform reference correction on the image to be registered through the phase deviation;

[0042] S17: Complete the registration of the image to be registered and the reference image through feature point matching.

[0043] Specifically, construct the fovea centralis feature, where the fovea is located at the posterior pole of the innermost layer of the eyeball wall, and there is a small depression in the center of the fovea, and this position is the fovea centralis feature. Match the positioning origin of the positioning image through the fovea centralis feature, calculate the pixel deviation between the center of the area to be measured and the positioning origin according to the positioning origin matching result, and obtain the area positioning result according to the pixel deviation. Calculate the phase deviation between the image to be registered and the reference image, and perform reference correction on the image to be registered through the phase deviation. Complete the registration of the image to be registered and the reference image through feature point matching.

[0044] The method S40 provided by the embodiment of the present application further includes:

[0045] S41: Construct the Hessian matrix, where the Hessian matrix includes eigenvalues λ1 and λ2, and |λ1| < |λ2|;

[0046] S42: Calculate the transfer function of the Frangi filter, and the calculation formula is as follows:

[0047]

[0048] S43: Perform filtering processing on the region localization result and the multi-view video set through the transfer function to obtain a set of response images;

[0049] S44: Perform maximum response point superposition on the set of response images, and generate a blood vessel contour enhancement image according to the superposition result;

[0050] S45: Remove the holes and noises in the blood vessel contour enhancement image, connect the discontinuous parts, and based on the gray level and gradient information of the blood vessel contour enhancement image, segment the complete blood vessel contour through level set, and binarize it to obtain the binary image of the blood vessel contour.

[0051] Specifically, construct a Hessian matrix, which includes eigenvalues λ1 and λ2 in the Hessian matrix, and |λ1| < |λ2|. This Hessian matrix is a well-known solution to those skilled in the art, so the acquisition of eigenvalues is also well-known and will not be elaborated here. Subsequently, calculate the transfer function of the Frangi filter, and the calculation formula is as follows:

[0052] where V is the transfer function of the Frangi filter, e is the natural logarithm, and c is a parameter that is set according to actual calculation requirements. Further, perform filtering processing on the region localization result and the multi-view video set through the transfer function to obtain a set of response images. Perform maximum response point superposition on the obtained set of response images, and generate a blood vessel contour enhancement image according to the superposition result. Use top-hat operation and morphological erosion and dilation operations to remove the holes and noises in the blood vessel contour enhancement image, connect the discontinuous parts, and based on the gray level and gradient information of the blood vessel contour enhancement image, segment the complete blood vessel contour through level set, and binarize it to obtain the binary image of the blood vessel contour.

[0053] As Figure 3 shown, the method S40 provided by the embodiment of the present application further includes:

[0054] S46: Select a one-dimensional Gaussian convolution kernel, and obtain a blood vessel wall image by superposing the maximum response through the one-dimensional Gaussian convolution kernel;

[0055] S47: Filter out artifact information through the blood vessel wall image and the binary image of the blood vessel contour, and improve the contrast of the image through gamma stretching to filter out noises to obtain a blood vessel wall information image;

[0056] S48: Extract pixel lines according to the blood vessel inclination angle in the blood vessel wall information image, and obtain the blood vessel wall thickness according to the pixel line extraction result.

[0057] Specifically, the obtained binary image of the blood vessel contour is further processed. By using the method of matched filtering, a one-dimensional Gaussian convolution kernel is selected and rotated by 30° each time for a total of 12 rotations. The maximum response is superimposed to obtain the blood vessel wall image. Subsequently, artifact information is filtered by using the blood vessel wall image and the binary image of the blood vessel contour. Then, gamma stretching is used to improve the contrast of the image and filter out noise, obtaining the blood vessel wall information map. Finally, pixel lines are extracted according to the blood vessel tilt angle in the blood vessel wall information map, and the blood vessel wall thickness is obtained according to the pixel line extraction result, that is, according to the tilt angle of the blood vessel, pixel lines are extracted in the normal direction, moved pixel by pixel, and the wall pixel values on each normal line are recorded and averaged as the final blood vessel wall thickness information.

[0058] The method S48 provided in the embodiment of the present application further includes:

[0059] S481: Traverse and iterate the characteristic pixel points of the binary image of the blood vessel contour, where the iteration steps are as follows:

[0060] S482: a Mark the boundary pixel points to be deleted;

[0061] S483: b After the marking is completed, delete the boundary pixel points;

[0062] S484: Repeat steps a and b until there are no pixel points that can be deleted and abort, generating the regional center line;

[0063] S485: By using the eight-neighborhood seed growth algorithm, obtain all the branch points P1 of the regional center line and the characteristic pixel points P2 within the 8-neighborhood and record them;

[0064] S486: Set all the pixel points of the branch point P1 and all the pixel points in the 8-neighborhood of each branch point to 0;

[0065] S487: Sort all the line segments in the binary image of the blood vessel contour by length, delete the branch line segments shorter than the threshold length, set all the points within the characteristic pixel points P2 to 1, and set them to 0 if they are isolated points;

[0066] S488: Set all the points within the branch point P1 to 1 and judge whether all the line segments in the image are connected;

[0067] S489: When all the line segments are connected, generate the blood vessel center line image.

[0068] Specifically, traverse and iterate the feature pixel points of the binary image of the blood vessel contour. The specific iteration steps are as follows: Mark the boundary pixel points to be deleted. After the marking is completed, delete the boundary pixel points. Repeat the above steps until there are no pixel points that can be deleted and then abort, generating the regional centerline. That is, remove pixel points along the edge of the blood vessel until there are no pixel points that can be deleted and then abort. At this time, the obtained centerline contains smaller branch points. Through the eight-neighborhood seed growth algorithm, obtain the branch point P1 of the regional centerline and the feature pixel points P2 within the 8-neighborhood and record them. Set all the pixel points of the branch point P1 and the 8-neighborhood of each branch point to 0. Since the pixel point values in the original blood vessel region in the binary image are 1, when processing, set all the pixel point values of the branch point P1 and the 8-neighborhood of each branch point to 0. Subsequently, sort all the line segments in the binary image of the blood vessel contour by length, delete the branch line segments shorter than the threshold length, set all the points within the feature pixel points P2 to 1, and if they are isolated points, set them to 0. Set all the points within the branch point P1 to 1, and determine whether all the line segments in the image are connected. When all the line segments are connected, generate the blood vessel centerline image.

[0069] S50: Set the blood flow trace extraction window through the blood vessel centerline image;

[0070] S60: Obtain the basic calculation background image from the multi-view video set and calculate the division image of the basic calculation background image;

[0071] S70: Construct an image pyramid of the cell sequence based on the basic calculation background image and the division image, and track from any layer of the image pyramid. According to the tracking result, obtain the pixel point optical flow offset;

[0072] S80: Based on the reverse detection optical flow, eliminate the optical flow abnormal points of the pixel point optical flow offset. According to the elimination and screening result, obtain the target cell, and calculate the blood flow velocity according to the target cell.

[0073] Specifically, a blood flow trace extraction window is set based on the blood vessel centerline image, a basic calculation background image is obtained from a multi-view video set, and a division image of the basic calculation background image is calculated. An image pyramid of the cell sequence is constructed based on the basic calculation background image and the division image, and tracking is performed from any layer of the image pyramid. According to the tracking result, the optical flow offset of the pixel point is obtained. Preferably, a 4-layer pyramid is constructed for each cell sequence to obtain information in different scale spaces; since the cell sequence has been extracted, a structure with a window length of 3*11 is used for detection. Tracking starts from the top layer of the image pyramid, and the result is inherited to the upper layer. A finer displacement is searched near the offset at this layer, and finally, it is gradually tracked to the finer details of the bottom layer. Different rates of motion are captured by this method. According to the finally obtained optical flow, the optical flow offset of each pixel point is calculated. Based on the reverse detection optical flow, the optical flow abnormal points of the pixel point optical flow offset are eliminated. According to the elimination and screening results, the target cells are obtained, and the blood flow velocity is calculated based on the target cells. That is, through the reverse detection optical flow, assuming that the optical flow calculation is correct, the forward and reverse optical flows of the two images should exactly correspond, and the reverse optical flow between these two frames is obtained. By tracking and checking the forward and reverse optical flow points, some optical flow abnormal points (usually abnormal points caused by environmental illumination effects) are eliminated. For the finally selected points, the RANSAC operator is used to select the best matching pairs; the blood flow velocity is calculated based on the selected target cells, thereby realizing the accurate measurement of the blood flow velocity, providing more accurate data for the clear diagnosis and treatment of diseases such as dilation of the ocular vascular area and fundus vascular obstructive tumors.

[0074] The method S50 provided in the embodiment of the present application further includes:

[0075] S51: Arrange the line coordinates point by point according to the blood vessel trend and flow direction in the blood vessel centerline image;

[0076] S52: Set the window length as win, and each time take a small line segment line with a length of l of win length;

[0077] S53: Obtain the tangent direction of the small line segment line, and generate a parallel curve set based on the tangent direction and all points recorded on the line;

[0078] S54: Take pixels frame by frame according to the recorded abscissa to form a 7*length(line) matrix, and use the 7*length(line) matrix as the blood flow trace extraction window.

[0079] Specifically, the line coordinates are arranged point by point according to the blood vessel trend and flow direction in the blood vessel center line, that is, the coordinates of the center line are arranged point by point. The window length is set to win, and a small line segment line with a length of l for each win length is taken each time, that is, the window length is set, and a perpendicular line is drawn with each point on the line as the center, and the length of the perpendicular line is 7. The starting point coordinates are recorded in order (with the right and down directions as the positive directions), and the subsequent all points store 7 points in sequence according to the order of the starting point; the line is taken and recorded in sequence to obtain a set of 7 parallel curves extending along the blood vessel curvature direction, and the middle one is the blood vessel center line; the pixels are taken frame by frame according to the recorded horizontal and vertical coordinates and form a matrix of 7*length(line), that is, this matrix is considered as the blood flow trace window of the blood vessel.

[0080] The technical solution provided by the embodiment of the present invention obtains a multi-view video set by performing multi-view video acquisition on the retina of a target user. Based on the multi-view video set, the area to be measured of the blood vessels of the target user is located to obtain a regional location result. Based on the regional location result and the multi-view video set, the blood vessel contour is extracted based on the inter-frame difference of the video to obtain a binary image of the blood vessel contour. Based on the binary image of the blood vessel contour, the blood vessel wall thickness is calculated, and the blood vessel center line image is extracted. A blood flow trace extraction window is set through the blood vessel center line image. A basic calculation background image is obtained through the multi-view video set, and a division image of the basic calculation background image is calculated; an image pyramid of the cell sequence is constructed based on the basic calculation background image and the division image, and tracking is performed from any layer of the image pyramid. According to the tracking result, the pixel point optical flow offset is obtained. Based on the reverse detection optical flow, the optical flow abnormal points of the pixel point optical flow offset are removed, and the target cells are obtained according to the removal and screening results. The blood flow velocity is calculated according to the target cells, realizing accurate measurement of the blood flow velocity of the multi-level blood vessels in the fundus. It solves the technical problem of low measurement accuracy in measuring the blood flow velocity of multi-level blood vessels in the fundus in the existing method for measuring the blood flow velocity of the retina.

[0081] Embodiment 2

[0082] Based on the same inventive concept as a method for measuring the blood flow velocity of the retina in the foregoing embodiment, the present invention also provides a system for a method for measuring the blood flow velocity of the retina. The system can be implemented in a hardware and / or software manner and is generally integrated into an electronic device for executing the method provided by any embodiment of the present invention. As Figure 4 shown, the system includes:

[0083] A video set acquisition module 11, configured to perform multi-view video acquisition on the retina of a target user to obtain a multi-view video set;

[0084] The region location result acquisition module 12 is configured to perform location of the blood vessel region to be measured of the target user based on the multi-view video set, and obtain a region location result;

[0085] The binary image acquisition module 13 is configured to extract the blood vessel contour based on the video frame difference according to the region location result and the multi-view video set, and obtain a binary image of the blood vessel contour;

[0086] The centerline image acquisition module 14 is configured to calculate the blood vessel wall thickness based on the binary image of the blood vessel contour, and extract and obtain a blood vessel centerline image;

[0087] The blood flow trace extraction window acquisition module 15 is configured to set a blood flow trace extraction window through the blood vessel centerline image;

[0088] The background image acquisition module 16 is configured to obtain a basic calculation background image through the multi-view video set, and calculate and obtain a division image of the basic calculation background image;

[0089] The pixel point optical flow offset acquisition module 17 is configured to construct an image pyramid of the cell sequence based on the basic calculation background image and the division image, and perform tracking from any layer of the image pyramid, and obtain the pixel point optical flow offset according to the tracking result;

[0090] The blood flow velocity calculation module 18 is configured to perform elimination of optical flow abnormal points of the pixel point optical flow offset based on reverse detection optical flow, obtain target cells according to the elimination and screening result, and calculate and obtain the blood flow velocity according to the target cells.

[0091] Furthermore, the video set acquisition module 11 is further configured to:

[0092] Collect the retinal image of the target user at the first wide angle and the first angle view, and use the collected image as a location image;

[0093] Collect the fundus video of the target user at the second angle view to obtain a measurement video;

[0094] Obtain the multi-view video set through the location image and the measurement video.

[0095] Furthermore, the video set acquisition module 11 is further configured to:

[0096] Construct a fovea centralis feature, perform location origin matching on the location image through the fovea centralis feature, and calculate and obtain the pixel deviation between the center of the region to be measured and the location origin according to the location origin matching result;

[0097] Obtain the region location result according to the pixel deviation;

[0098] Calculate the phase deviation between the image to be registered and the reference image, and perform reference correction on the image to be registered based on the phase deviation;

[0099] Complete the registration of the image to be registered and the reference image by feature point matching.

[0100] Furthermore, the centerline image acquisition module 14 is further configured to:

[0101] Construct a Hessian matrix, where the Hessian matrix includes eigenvalues λ1 and λ2, and |λ1| < |λ2|;

[0102] Calculate the transfer function of the Frangi filter, and the calculation formula is as follows:

[0103]

[0104] Perform filtering processing on the region positioning result and the multi-view video set through the transfer function to obtain a set of response images;

[0105] Perform maximum response point superposition on the set of response images, and generate a blood vessel contour enhancement image according to the superposition result;

[0106] Remove the holes and noises in the blood vessel contour enhancement image, connect the discontinuous parts, and segment the complete blood vessel contour through level set based on the gray level and gradient information of the blood vessel contour enhancement image, and binarize it to obtain the binary image of the blood vessel contour.

[0107] Furthermore, the centerline image acquisition module 14 is further configured to:

[0108] Select a one-dimensional Gaussian convolution kernel, and superimpose the maximum response through the one-dimensional Gaussian convolution kernel to obtain a blood vessel wall image;

[0109] Perform artifact information filtering through the blood vessel wall image and the binary image of the blood vessel contour, and improve the contrast of the image through gamma stretching to filter out noises, so as to obtain a blood vessel wall information image;

[0110] Extract pixel lines according to the blood vessel inclination angle in the blood vessel wall information image, and obtain the blood vessel wall thickness according to the pixel line extraction result.

[0111] Furthermore, the centerline image acquisition module 14 is further configured to:

[0112] Traverse and iterate the feature pixel points of the binary image of the blood vessel contour, and the iteration steps are as follows:

[0113] a Mark the boundary pixel points to be deleted;

[0114] b After the marking is completed, delete the boundary pixel points;

[0115] Repeat steps a and b until there are no more pixel points that can be deleted and then abort to generate the regional centerline;

[0116] By using the eight-neighborhood seed growth algorithm, obtain all the branch points P1 of the regional centerline and the characteristic pixel points P2 within the 8-neighborhood and record them;

[0117] Set all the pixel points of the branch point P1 and all the pixel points within the 8-neighborhood of each branch point to 0;

[0118] Sort all the line segments in the binary image of the blood vessel contour by length, delete the branch line segments shorter than the threshold length, set all the points within the characteristic pixel points P2 to 1, and if it is an isolated point, set it to 0;

[0119] Set all the points within the branch point P1 to 1 and determine whether all the line segments in the image are connected;

[0120] When all the line segments are connected, generate the blood vessel centerline image.

[0121] Furthermore, the blood flow trace extraction window acquisition module 16 is further configured to:

[0122] Arrange the line coordinates point by point according to the blood vessel trend and the flow direction in the blood vessel centerline image;

[0123] Set the window length to win, and each time take a small line segment line with a length of l of win length;

[0124] Obtain the tangent direction of the small line segment line, and generate a parallel curve set based on the tangent direction and all the points recorded on the line;

[0125] Take pixels frame by frame according to the recorded abscissa to form a 7 * length(line) matrix, and use the 7 * length(line) matrix as the blood flow trace extraction window.

[0126] The retinal blood flow velocity measurement system provided by the embodiments of the present invention can execute the method of the retinal blood flow velocity measurement system provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0127] The various units and modules included are only divided according to the functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of mutual distinction and do not limit the protection scope of the present invention.

[0128] Embodiment III

[0129] Figure 5Schematic diagram of the structure of the electronic device provided in Embodiment 3 of the present invention, showing a block diagram of an exemplary electronic device suitable for implementing the embodiments of the present invention. Figure 5 The displayed electronic device is only an example and should not impose any limitations on the functions and usage scope of the embodiments of the present invention. As Figure 5 shown, the electronic device includes a processor 31, a memory 32, an input device 33, and an output device 34; the number of processors 31 in the electronic device can be one or more. Figure 5 Taking one processor 31 as an example, the processor 31, the memory 32, the input device 33, and the output device 34 in the electronic device can be connected by a bus or other means. Figure 5 Taking the connection by bus as an example.

[0130] The memory 32, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as program instructions / modules corresponding to a method for measuring retinal blood flow velocity in the embodiments of the present invention. The processor 31 executes various functional applications and data processing of the computer device by running the software programs, instructions, and modules stored in the memory 32, that is, implements the above-mentioned method for measuring retinal blood flow velocity.

[0131] Note that the above is only the preferred embodiment of the present invention and the applied technical principle. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein. Various obvious changes, re-adjustments, and substitutions can be made by those skilled in the art without departing from the protection scope of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments. Without departing from the concept of the present invention, it can also include more other equivalent embodiments, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for measuring retinal blood flow velocity, characterized in that, The method includes: Performing multi - perspective video acquisition of the retina of the target user to obtain a multi - perspective video set; Based on the multi - perspective video set, localizing the blood vessel region to be measured of the target user to obtain a region localization result; Based on the multi - perspective video set, extracting blood vessel contours based on inter - frame differences of video frames to obtain a binary blood vessel contour map; Calculating the blood vessel wall thickness based on the binary blood vessel contour map and extracting the blood vessel centerline image; Setting a blood flow trace extraction window through the blood vessel centerline image; Obtaining a basic calculation background image through the multi - perspective video set and calculating a division image of the basic calculation background image; Constructing an image pyramid of the cell sequence based on the basic calculation background image and the division image, and tracking from any layer of the image pyramid, and obtaining pixel - point optical flow offset according to the tracking result; Eliminating optical flow abnormal points of the pixel - point optical flow offset based on reverse - detection optical flow, obtaining target cells according to the elimination and screening result, and calculating the blood flow velocity according to the target cells.

2. The method according to claim 1, wherein The method further includes: Performing retina image acquisition of the target user at a first wide - angle and a first - angle perspective, and using the acquired image as a positioning image; Performing fundus video acquisition of the target user at a second - angle perspective to obtain a measurement video; Obtaining the multi - perspective video set through the positioning image and the measurement video.

3. The method according to claim 2, characterized in that, The method further includes: Constructing a fovea centralis feature, matching the positioning origin of the positioning image through the fovea centralis feature, and calculating the pixel deviation between the center of the region to be measured and the positioning origin according to the positioning origin matching result; Obtaining the region localization result according to the pixel deviation; Calculating the phase deviation between the image to be registered and the reference image, and performing reference correction on the image to be registered through the phase deviation; Completing the registration of the image to be registered and the reference image through feature - point matching.

4. The method according to claim 1, characterized in that The method further includes: Constructing a Hessian matrix, where the Hessian matrix includes eigenvalues λ1 and λ2, and |λ1| < |λ2|; Calculating the transfer function of the Frangi filter, and the calculation formula is as follows: Performing filtering processing on the region localization result and the multi - perspective video set through the transfer function to obtain a set of response images; Performing maximum response point superposition on the set of response images, and generating a blood vessel contour enhancement image according to the superposition result; Removing the holes and noises in the blood vessel contour enhancement image, connecting the discontinuous parts, and segmenting the complete blood vessel contour through level set based on the gray - scale and gradient information of the blood vessel contour enhancement image, and binarizing it to obtain the binary blood vessel contour map.

5. The method according to claim 4, characterized in that The method further includes: Selecting a one - dimensional Gaussian convolution kernel, and obtaining a blood vessel wall image by superposing the maximum response through the one - dimensional Gaussian convolution kernel; Filtering out artifact information through the blood vessel wall image and the binary blood vessel contour map, and filtering out noises by stretching the gamma to improve the contrast of the image to obtain a blood vessel wall information map. Extract pixel lines according to the blood vessel inclination angle in the blood vessel wall information diagram, and obtain the blood vessel wall thickness according to the pixel line extraction result.

6. The method according to claim 5, wherein The method further includes: Traverse and iterate the characteristic pixel points of the binary blood vessel contour diagram, where the iteration steps are as follows: a Mark the boundary pixel points to be deleted; b After the marking is completed, delete the boundary pixel points; Repeat steps a and b until there are no pixel points that can be deleted and abort, generating the regional center line; Obtain all branch points P1 of the regional center line and characteristic pixel points P2 within the 8-neighborhood through the eight-neighborhood seed growth algorithm and record them; Set all pixel points of the branch point P1 and all pixel points in the 8-neighborhood of each branch point to 0; Sort all line segments of the binary blood vessel contour diagram by length, delete the branch line segments shorter than the threshold length, set all points within the characteristic pixel points P2 to 1, and set them to 0 if they are isolated points; Set all points within the branch point P1 to 1 and determine whether all line segments in the image are connected; When all line segments are connected, generate the blood vessel center line image.

7. The method according to claim 1, wherein The method further includes: Arrange the line coordinates point by point according to the blood vessel trend and flow direction in the blood vessel center line image; Set the window length to win, and each time take a small line segment line with a length of l of win length; Obtain the tangent direction of the small line segment line, and generate a parallel curve set according to the tangent direction and all points recorded on the line; Take pixels frame by frame according to the recorded abscissa to form a 7*length(line) matrix, and use the 7*length(line) matrix as the blood flow trace extraction window.

8. A retinal blood flow velocity measurement system, characterized in that, The system includes: A video set acquisition module, used to perform multi-view video acquisition of the retina of the target user to obtain a multi-view video set; A regional location result acquisition module, used to perform regional location of the blood vessel area to be measured of the target user based on the multi-view video set to obtain a regional location result; A binary map acquisition module, used to extract the blood vessel contour based on the video frame difference according to the regional location result and the multi-view video set to obtain a binary blood vessel contour map; A center line image acquisition module, used to calculate the blood vessel wall thickness based on the binary blood vessel contour map and extract the blood vessel center line image; A blood flow trace extraction window acquisition module, used to set a blood flow trace extraction window through the blood vessel center line image; A background image acquisition module, used to obtain a basic calculation background image through the multi-view video set and calculate the division image of the basic calculation background image; A pixel point optical flow offset acquisition module, used to construct an image pyramid of the cell sequence based on the basic calculation background image and the division image, and track from any layer of the image pyramid, and obtain the pixel point optical flow offset according to the tracking result; A blood flow velocity calculation module, used to eliminate the optical flow abnormal points of the pixel point optical flow offset based on the reverse detection optical flow, obtain the target cells according to the elimination and screening results, and calculate the blood flow velocity according to the target cells.

9. An fundus imaging device, characterized in that, The fundus camera includes the system as described in claim 8.

10. An electronic device, characterized in that, The electronic device includes: A memory for storing executable instructions; A processor, when executing the executable instructions stored in the memory, implements a method for measuring retinal blood flow velocity according to any one of claims 1 to 7.

Images