Methods, equipment, and media for measuring retinal oxygen saturation based on arteriovenous structure priors.

Abstract

This invention discloses a method, device, and medium for measuring retinal oxygen saturation based on arteriovenous structure priors, belonging to the field of fundus functional imaging technology. The method includes: acquiring two wavelength channel images of the same fundus region; performing vascular segmentation and arteriovenous classification on the acquired images to generate vascular tree masks, arterial masks, and venous masks; performing local background estimation on each wavelength channel image and calculating optical density and optical density ratio; statistically analyzing the optical density ratio distribution intervals within the arterial and venous masks and performing intra-category normalization; setting independent target intervals based on different physiological oxygenation levels of arteries and veins, mapping to obtain oxygen saturation values, and generating an oxygen saturation distribution map. This invention reduces the influence of uneven illumination and central vascular reflection through local background estimation, and utilizes arteriovenous structure priors to achieve intra-category normalization and specific mapping, improving the stability of oxygen saturation measurement and the physiological rationality of arteriovenous layers.

Description

Technical Field

[0001] This invention relates to the field of fundus functional imaging technology, and in particular to a method, device, and medium for measuring retinal blood oxygen saturation based on arteriovenous structure priors. Background Technology

[0002] The retina is the only tissue in the human body whose microcirculatory vascular network can be directly observed in vivo and non-invasively. Its blood oxygen metabolism status can reflect the local tissue oxygen supply and demand balance, and has important clinical value for the early diagnosis, disease monitoring, and efficacy evaluation of fundus diseases such as diabetic retinopathy, retinal vein occlusion, and glaucoma. Therefore, achieving stable and non-invasive measurement of retinal blood oxygen saturation is an important research direction in fundus functional imaging and auxiliary assessment of ophthalmic diseases.

[0003] Currently, the measurement of retinal oxygen saturation is mainly based on dual-wavelength fundus imaging technology. Its basic principle is to utilize the difference in absorption coefficients between oxyhemoglobin and deoxyhemoglobin at different wavelengths. By acquiring images of the same fundus region at two or more wavelengths, the optical density and optical density ratio of the vascular area are calculated, and then the oxygen saturation value is estimated through linear fitting or calibration curves based on Lambert-Beer's law. This method has a certain physical basis and is currently a commonly used technical approach. However, existing technologies still have the following shortcomings in practical applications: 1. Background estimation is easily affected by image quality, leading to unstable optical density calculations. Traditional methods typically rely on manually selecting or automatically detecting several tissue regions near blood vessels as background reference points, estimating background light intensity by the average gray level of the sampled points. When fundus images exhibit uneven illumination, differences in local tissue reflection, central reflection of blood vessels, or decreased signal-to-noise ratio in surrounding areas, the representativeness of discrete background points is limited, making it difficult to accurately reflect the true local background level of each blood vessel pixel. This easily causes deviations in the calculation of optical density and optical density ratio, thus affecting the accuracy of blood oxygen saturation measurement.

[0004] 2. The mapping relationship between optical density ratio and blood oxygen saturation lacks prior guidance from arterial and venous structures. Current techniques often employ a uniform global linear mapping function for all vascular pixels, failing to adequately consider the differences in oxygenation levels, grayscale representation, and optical density ratio distribution between arteries and veins. Physiologically, arterial blood oxygen saturation is at a higher level (typically 90%–100%), while venous oxygen saturation is lower (typically 50%–80%), and the two are naturally distributed in different ranges within the optical density ratio space. Using a globally uniform mapping can easily lead to unclear levels of blood oxygen saturation in arteries and veins, and even non-physiological jumps where arterial oxygen saturation is lower than venous oxygen saturation, reducing the physiological rationality and clinical interpretability of the measurement results.

[0005] 3. Lack of in-depth utilization of vascular structure information. In recent years, deep learning methods have achieved good results in retinal vessel segmentation and arteriovenous classification tasks. However, most existing technologies treat them as independent structural recognition tasks, only outputting vascular masks or arteriovenous classification labels, without further embedding spatial structural information such as vascular trees, arteries, and veins into the blood oxygen saturation calculation process. Especially in ultra-wide-angle scanning laser fundus microscope images, due to the large field of view, complex illumination distribution, and significant signal attenuation in the peripheral areas, it is difficult to obtain a continuous, stable blood oxygen saturation distribution map that conforms to the physiological differences between arteries and veins by relying on traditional global mapping methods.

[0006] Therefore, there is an urgent need for a retinal oxygen saturation measurement method that can combine arterial and venous structure priors, employ robust local background estimation methods, and establish specific mapping relationships for arteries and veins respectively, in order to overcome the above-mentioned shortcomings of existing technologies. Summary of the Invention

[0007] To achieve the above-mentioned objectives and other advantages of the present invention, a first objective of the present invention is to provide a method for measuring retinal oxygen saturation based on arteriovenous structure priors, comprising the following steps: Two wavelength channels were acquired for the same fundus region, with the two wavelengths showing different sensitivities to hemoglobin oxygenation status. The acquired fundus images are segmented into blood vessels and classified into arteries and veins to generate a vascular tree mask, an arterial mask, and a vein mask. Local background estimation is performed for each wavelength channel image. The optical density is calculated based on the relative deviation between the original pixel gray value and the corresponding local background gray value. The optical density ratio is calculated based on the optical density at the two wavelengths, and the optical density ratio calculation result is limited to the effective blood vessel area. For the two blood vessel categories of arteries and veins, the optical density ratio of effective blood vessel pixels is extracted within the corresponding blood vessel category mask, the distribution range of the optical density ratio of the category is statistically analyzed, and the distribution range is used as the normalization reference range for the current blood vessel category. The optical density ratio is then normalized to obtain the normalized optical density ratio of the pixel within the corresponding blood vessel category. Based on the different physiological oxygenation levels of arteries and veins, target ranges for arterial and venous blood oxygen saturation are set. Based on the normalized optical density ratio and the target ranges, the blood oxygen saturation value of the pixel is mapped.

[0008] Furthermore, the specific method for local background estimation is as follows: For any pixel Construct a local neighborhood window centered on that pixel. The pixel grayscale within the local neighborhood window is then low-pass smoothed, and the smoothed result is used as the local background value for that pixel. in, This indicates a low-pass filtering operation applied to pixels within this local neighborhood window. Indicates window size. wavelength The original image grayscale value, wavelength The grayscale value of the lower local background.

[0009] Furthermore, the formula for calculating the optical density is: Calculate the optical density ratio based on the optical density at two wavelengths: The calculated optical density ratio is limited to the effective vascular area: in, and Indicates two different wavelengths. wavelength Next pixel The optical density at that location.

[0010] Furthermore, the specific method for statistically analyzing the distribution range of the optical density ratio is as follows: For blood vessel categories , Indicates artery, Represents veins, within its effective set of vessel pixels. Low percentile of internal statistical optical density ratio and high percentile : in, and This is the preset percentile.

[0011] Furthermore, the formula for the normalization process is: in, Represents pixels Optical density ratio in the relevant blood vessel category The relative level within the range is normalized to the interval [0,1].

[0012] Furthermore, the mapping formula for the blood oxygen saturation value is as follows: in, Blood vessel category The target range for blood oxygen saturation.

[0013] Furthermore, the target range is set or adaptively updated based on prior physiological data of retinal blood oxygenation, statistical results of healthy samples, equipment calibration results, or disease type.

[0014] Furthermore, the step of performing vascular segmentation and arteriovenous classification on the acquired fundus image to generate a vascular tree mask, an arterial mask, and a vein mask also includes obtaining a mask for regions with uncertain classification. ; The effective blood vessel pixel set Defined as: in, This indicates the exclusion of areas with uncertain classification.

[0015] Furthermore, the vessel segmentation and arteriovenous classification are implemented using at least one of deep learning models, image processing algorithms, vessel tracking algorithms, and graph structure analysis algorithms.

[0016] Furthermore, the two wavelength channel images include a first wavelength channel image and a second wavelength channel image; wherein, the first wavelength channel is a reference wavelength channel that is relatively insensitive to changes in the oxygenation state of hemoglobin, and the second wavelength channel is an oxygen-sensitive wavelength channel that is sensitive to the difference in absorption between oxyhemoglobin and deoxyhemoglobin; the fundus images are simultaneously acquired by an ultra-wide-angle scanning laser fundus imaging system and the inter-channel registration is completed.

[0017] Furthermore, the blood oxygen saturation value is limited to the corresponding type of vascular mask to generate an arterial oxygen saturation map, a venous oxygen saturation map, and a fused retinal blood oxygen saturation distribution map.

[0018] A second objective of the present invention is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0019] A third objective of the present invention is to provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method.

[0020] Compared with the prior art, the beneficial effects of the present invention are: This invention provides a method for measuring retinal blood oxygen saturation based on arteriovenous structure priors. This method introduces vascular tree and arteriovenous structure priors into the blood oxygen saturation calculation process, so that blood oxygen saturation analysis no longer depends solely on image grayscale or global ODR distribution, but is constrained by vascular location and arteriovenous category. This effectively reduces the phenomenon of blood oxygen saturation value jumps on blood vessels and improves the structural consistency and physiological continuity of retinal blood oxygen saturation distribution.

[0021] This invention employs a local background estimation method to calculate optical density. Specifically, local background images of the red and green light channels are obtained through low-pass filtering, and then the optical density is calculated based on the deviation of the original pixels relative to the local background. This method can mitigate the impact of uneven illumination, changes in background reflection, and decreased signal-to-noise ratio in the image on ODR calculation, and exhibits better robustness compared to traditional fixed sampling near blood vessels.

[0022] This invention statistically analyzes the ODR distribution in both arterial and venous regions and normalizes it through categorical intervals. This method avoids the mapping bias caused by mixing arteries and veins into a unified ODR scale, allowing arteries and veins to be mapped to blood oxygen saturation according to their own ODR distribution ranges, thus improving the resolution of hyperoxia and hypooxia regions in arteries and veins.

[0023] This invention employs arterial and venous specific oxygen saturation target intervals for mapping. The target oxygen saturation intervals for arteries and veins can be set or adaptively adjusted based on physiological priors, statistical analysis of healthy samples, equipment calibration results, or disease type, enabling the method to adapt to different imaging devices, disease scenarios, and analytical tasks. Even in the absence of external gold standard calibration, it can generate physiologically interpretable semi-quantitative oxygen saturation distribution maps.

[0024] In summary, this invention combines arteriovenous structure classification with dual-wavelength optical density ratio calculation, and solves problems such as instability of traditional linear mapping, unclear arteriovenous layers, and local non-physiological fluctuations in fundus images through local background estimation, intra-category ODR normalization, and arteriovenous-specific mapping. It provides a stable, interpretable, and scalable measurement method for retinal oxygen saturation analysis.

[0025] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Specific embodiments of the present invention are given in detail below with reference to the accompanying drawings. Attached Figure Description

[0026] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 Flowchart of a method for measuring retinal oxygen saturation based on arteriovenous structure priors; Figure 2 This is the optical absorption curve of hemoglobin; Figure 3 Fundus images of the green and red light channels; Figure 4 Diagram of vascular tree mask and arteriovenous mask; Figure 5 This is a retinal oxygen saturation distribution map; Figure 6 This is a schematic diagram of a computer device. Figure 7 This is a schematic diagram of a computer-readable storage medium. Detailed Implementation

[0027] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0028] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0029] The drawing numbers in this application are only used to distinguish the steps in the scheme and are not used to limit the execution order of the steps. The specific execution order is as described in the specification.

[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0031] Example 1 A method for measuring retinal oxygen saturation based on arteriovenous structure priors, such as Figure 1 As shown, it includes the following steps: S100. Acquire two wavelength channel images of the same fundus region. The two wavelength channel images include a first wavelength channel image and a second wavelength channel image that have different sensitivities to hemoglobin oxygenation status. The first wavelength channel is a reference wavelength channel that is relatively insensitive to changes in hemoglobin oxygenation status, and the second wavelength channel is an oxygen-sensitive wavelength channel that is sensitive to the difference in absorption between oxyhemoglobin and deoxyhemoglobin.

[0032] In this embodiment, red, green, and blue light channel images of the same fundus region are first acquired simultaneously using an ultra-wide-angle scanning laser fundus imaging system, and the system software completes the inter-channel registration. The blue light channel is used only for fundus imaging and not for blood oxygen saturation measurement and analysis.

[0033] Figure 2 This is the optical absorption curve of hemoglobin. Oxyhemoglobin and deoxyhemoglobin exhibit significant differences in their absorption characteristics for different wavelengths of light. Specifically, both oxyhemoglobin and deoxyhemoglobin are insensitive to green light at 525 nm, with low and similar absorption coefficients; however, at red light at 660 nm, the absorption coefficient of oxyhemoglobin is much lower than that of deoxyhemoglobin. Based on this physical characteristic, this embodiment selects the green and red light channels as the input images for blood oxygen saturation measurement.

[0034] Registered red channel image and green channel image like Figure 3 As shown. Among them, Two-dimensional coordinates representing image pixels. The two images have been pixel-level registered, meaning they share the same coordinates. The same anatomical location in the fundus is shown in both red and green light images.

[0035] S200. Perform vascular segmentation and arteriovenous classification on the acquired fundus images to generate a vascular tree mask. Arterial mask and vein mask ; like Figure 4 As shown, fundus images are processed using deep learning models (such as U-Net and its variants, Transformer architecture, etc.), outputting three masks: a vascular tree mask. Used to identify all blood vessel pixels; artery mask Used to identify artery pixels; vein mask Used to identify vein pixels. Figure 4 In the diagram, red represents arteries and blue represents veins.

[0036] As another embodiment of the present invention, the vessel segmentation and arteriovenous classification can also be implemented using traditional image processing algorithms (such as matched filtering, morphological operations, region growing, etc.), vessel tracking algorithms, graph structure analysis algorithms (such as graph cut, minimum spanning tree, etc.), or a combination of multiple methods. The present invention does not limit this to any particular method.

[0037] S300. Perform local background estimation for each wavelength channel image, calculate optical density based on the relative deviation between the original pixel gray value and the corresponding local background gray value, calculate the optical density ratio based on the optical density under the two wavelengths, and limit the optical density ratio calculation result to the effective blood vessel area. Due to the characteristics of ultra-wide-angle scanning laser fundus microscope images, such as uneven illumination, significant peripheral signal attenuation, and local reflection differences, traditional methods that estimate background by selecting several background points next to the blood vessel cannot accurately reflect the true local background level of each blood vessel pixel. Therefore, this invention employs a local background estimation method to provide a corresponding local background reference for each pixel.

[0038] For red channel images and green channel image Perform local background estimation separately. For any pixel... Construct a local neighborhood window centered on that pixel. The algorithm performs low-pass smoothing on the pixel grayscale within the local neighborhood window, and uses the smoothed result as the local background value for that pixel. The specific formula is as follows: in, This indicates a low-pass filtering operation applied to pixels within this local neighborhood window. The window size is indicated. In this embodiment, the low-pass filter uses a Gaussian filter; for example, in this embodiment, the local neighborhood window uses a low-pass filter kernel with a kernel size of 61. The local background image is used to characterize the local background reflection level within the neighborhood of each pixel.

[0039] In other embodiments, the low-pass filtering operation can be implemented using algorithms such as median filtering, mean filtering, bilateral filtering, or guided filtering, and is not limited to Gaussian filtering.

[0040] S400. For the two blood vessel categories of arteries and veins, extract the optical density ratio of effective blood vessel pixels in the corresponding blood vessel category mask, count the distribution range of the optical density ratio of the category, use the distribution range as the normalization reference range of the current blood vessel category, normalize the optical density ratio, and obtain the normalized optical density ratio of the pixel in the corresponding blood vessel category. In this embodiment, the optical density at a given wavelength is calculated based on the relative deviation between the original pixel grayscale value and the corresponding local background grayscale value of each wavelength channel image. This deviation simultaneously describes both local darkening caused by vascular absorption and local brightening caused by reflection or localized glare from the center of the blood vessel. The specific calculation formula is as follows: Red light channel optical density: Green channel optical density: in, This represents the deviation of the original pixel grayscale value from the local background grayscale value. It should be noted that the +1 term in the above formula is a constant set to prevent the denominator from being zero. In other embodiments of the present invention, other forms of regularization constants or equivalent optical density calculation formulas may also be used.

[0041] Based on the light density of the red light channel and green channel optical density Calculate the optical density ratio (ODR): Subsequently, the calculated optical density ratio is limited to the effective vascular area: in, This serves as a marker for the vascular tree mask. For non-vascular areas outside the vascular tree mask, the... A value of zero indicates that the blood oxygen saturation is not included in subsequent blood oxygen saturation mapping and statistics.

[0042] Because arteries and veins have significant differences in oxygenation status and optical density ratio distribution, this invention does not mix all vascular pixels in the same scale for processing, but instead statistically analyzes the distribution of optical density ratio in the arterial region and the vein region separately.

[0043] For blood vessel categories , Indicates artery, To represent a vein, the effective set of vessel pixels is defined as follows: in, This indicates the exclusion of regions with classification uncertainty. The mask for these regions with classification uncertainty. It can be obtained simultaneously in the blood vessel segmentation and arteriovenous classification steps to identify areas with low classification confidence, thus avoiding interference from these abnormal areas in the distribution statistics.

[0044] exist Low percentile of internal statistical optical density ratio and high percentile : in, and This is a preset percentile. In this embodiment, we take... and Determining the effective range of optical density ratio by using the percentile method can effectively reduce the impact of local reflections, misclassification in cross-regions, errors at the vessel edges, and outliers in surrounding low signal-to-noise ratio areas on the optical density ratio distribution level.

[0045] In other embodiments of the present invention, the preset percentile may also be adjusted to other values ​​such as 1% and 99%, 2% and 98% depending on the image quality or device characteristics.

[0046] For blood vessel categories The optical density ratio within this category is truncated to The interval is then normalized to the [0,1] interval: in, Represents pixels Optical density ratio in the relevant blood vessel category The relative level within. This normalization process allows the optical density ratios within different vessel categories to be mapped independently within their respective distribution ranges, avoiding mapping bias caused by mixing arteries and veins into a uniform scale.

[0047] S500. Based on the different physiological oxygenation levels of arteries and veins, set target intervals for arterial blood oxygen saturation and venous blood oxygen saturation. Based on the normalized optical density ratio and the target intervals, map the blood oxygen saturation value of the pixel.

[0048] This invention sets different target ranges for blood oxygen saturation for arteries and veins respectively. These target ranges can be set or updated based on prior physiological data on retinal blood oxygenation, statistical results of healthy samples, equipment calibration results, or disease type.

[0049] In this embodiment, a target range for arterial blood oxygen saturation is set based on the physiological prior of retinal blood oxygenation. Target range for venous oxygen saturation .

[0050] For blood vessel categories pixels within The blood oxygen saturation value is mapped as follows: in, This refers to the aforementioned normalized optical density ratio.

[0051] By limiting blood oxygen saturation values ​​to the corresponding category of vascular mask, arterial oxygen saturation maps and venous oxygen saturation maps are obtained: By fusing the two data points, a retinal oxygen saturation distribution map is obtained: Figure 5 This is the final retinal oxygen saturation distribution map. In this map, different colors or gray levels represent different oxygen saturation values, which can intuitively show the distribution of oxygen metabolism status of the entire retinal vascular network.

[0052] In other embodiments, the target interval can be calibrated using statistical results from healthy samples. Specifically, dual-wavelength images of the retina of a group of healthy subjects are acquired, and the optical density ratio of the arterial and venous regions is statistically analyzed within a reference interval. Combined with the device calibration results, the target arterial and venous intervals applicable to the current imaging system are determined. This adaptive calibration method can better adapt to the characteristics of different imaging devices.

[0053] This invention calculates optical density using a local background estimation method. Specifically, local background images of the red and green light channels are obtained through low-pass filtering, and then the optical density is calculated based on the deviation of the original pixels relative to the local background. This method can mitigate the impact of uneven illumination, changes in background reflection, and decreased signal-to-noise ratio in the image on the optical density ratio calculation, and has better robustness compared to traditional fixed sampling near blood vessels.

[0054] This invention statistically analyzes the optical density ratio distribution in both arterial and venous regions and normalizes it through categorized intervals. This method avoids the mapping bias caused by mixing arteries and veins into a uniform optical density ratio scale, allowing arteries and veins to be mapped to blood oxygen saturation according to their own optical density ratio distribution ranges, thus improving the resolution of hyperoxia and hypooxia regions in arteries and veins.

[0055] This invention employs arterial and venous specific oxygen saturation target intervals for mapping. The target oxygen saturation intervals for arteries and veins can be set or adaptively adjusted based on physiological priors, statistical analysis of healthy samples, equipment calibration results, or disease type, enabling the method to adapt to different imaging devices, disease scenarios, and analytical tasks. Even in the absence of external gold standard calibration, it can generate physiologically interpretable semi-quantitative oxygen saturation distribution maps.

[0056] In summary, this invention combines arteriovenous structure classification with dual-wavelength optical density ratio calculation, and solves problems such as instability of traditional linear mapping, unclear arteriovenous layers, and local non-physiological fluctuations in fundus images through local background estimation, intra-category optical density ratio normalization, and arteriovenous-specific mapping. It provides a stable, interpretable, and scalable measurement method for retinal oxygen saturation analysis.

[0057] Example 2 A computer device 600, such as Figure 6 As shown, the system includes a memory 610, a processor 620, and a computer program 630 stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a method for measuring retinal oxygen saturation based on arteriovenous structure priors. For a detailed description of the method, please refer to the corresponding description in the above method embodiments; it will not be repeated here.

[0058] Example 3 A computer-readable storage medium, such as Figure 7 As shown, a computer program is stored thereon. When executed by a processor, the computer program implements the steps of a method for measuring retinal oxygen saturation based on arteriovenous structure priors. For a detailed description of the method, please refer to the corresponding description in the above method embodiments, which will not be repeated here.

[0059] The number of devices and processing scale described herein are for the purpose of simplifying the description of the invention. Applications, modifications, and variations of the invention will be readily apparent to those skilled in the art.

[0060] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

[0061] The apparatus, computer device, and non-volatile computer storage medium and method provided in the embodiments of this specification are corresponding. Therefore, the apparatus, computer device, and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, computer device, and non-volatile computer storage medium will not be repeated here.

[0062] Those skilled in the art will also know that, besides implementing the controller in the form of purely computer-readable program code, the same functions can be achieved by logically programming the method steps, making the controller take the form of logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers (PLCs), and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the devices included within it for implementing various functions can also be considered structures within that hardware component. Alternatively, the devices for implementing various functions can be considered as both software units implementing the method and structures within a hardware component.

[0063] The systems, apparatuses, or units described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. For ease of description, the above apparatuses are described separately as various units based on their functions. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0064] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0065] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0066] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0067] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0068] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0069] This specification may be described in the general context of computer-executable instructions, such as program units, that are executed by a computer. Generally, program units include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification may also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program units may reside in local and remote computer storage media, including storage devices.

[0070] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0071] The above description is merely an embodiment of this specification and is not intended to limit the scope of one or more embodiments of this specification. Various modifications and variations can be made to one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of one or more embodiments of this specification.

Claims

1. A method for measuring retinal oxygen saturation based on arteriovenous structure prior, characterized in that, Includes the following steps: Two wavelength channels were acquired for the same fundus region, with the two wavelengths showing different sensitivities to hemoglobin oxygenation status. The acquired fundus images are segmented into blood vessels and classified into arteries and veins to generate a vascular tree mask, an arterial mask, and a vein mask. Local background estimation is performed for each wavelength channel image. The optical density is calculated based on the relative deviation between the original pixel gray value and the corresponding local background gray value. The optical density ratio is calculated based on the optical density at the two wavelengths, and the optical density ratio calculation result is limited to the effective blood vessel area. For the two blood vessel categories of arteries and veins, the optical density ratio of effective blood vessel pixels is extracted within the corresponding blood vessel category mask, the distribution range of the optical density ratio of the category is statistically analyzed, and the distribution range is used as the normalization reference range for the current blood vessel category. The optical density ratio is then normalized to obtain the normalized optical density ratio of the pixel within the corresponding blood vessel category. Based on the different physiological oxygenation levels of arteries and veins, target ranges for arterial and venous blood oxygen saturation are set. Based on the normalized optical density ratio and the target ranges, the blood oxygen saturation value of the pixel is mapped.

2. The method for measuring retinal oxygen saturation based on arteriovenous structure prior as described in claim 1, characterized in that, The specific method for local background estimation is as follows: For any pixel Construct a local neighborhood window centered on that pixel. The pixel grayscale within the local neighborhood window is then low-pass smoothed, and the smoothed result is used as the local background value for that pixel. in, This indicates a low-pass filtering operation applied to pixels within this local neighborhood window. Indicates window size. wavelength The original image grayscale value, wavelength The grayscale value of the lower local background.

3. The method for measuring retinal oxygen saturation based on arteriovenous structure prior as described in claim 2, characterized in that, The formula for calculating the optical density is: Calculate the optical density ratio based on the optical density at two wavelengths: The calculated optical density ratio is limited to the effective vascular area: in, and Indicates two different wavelengths. wavelength Next pixel The optical density at that location.

4. The method for measuring retinal oxygen saturation based on arteriovenous structure prior as described in claim 1, characterized in that, The specific method for statistically analyzing the distribution range of the optical density ratio is as follows: For blood vessel categories , Indicates artery, Represents veins, within its effective set of vessel pixels. Low percentile of internal statistical optical density ratio and high percentile : in, and This is the preset percentile.

5. The method for measuring retinal oxygen saturation based on arteriovenous structure prior as described in claim 4, characterized in that, The formula for the normalization process is: in, Represents pixels Optical density ratio in the relevant blood vessel category The relative level within the range is normalized to the interval [0,1].

6. The method for measuring retinal oxygen saturation based on arteriovenous structure prior as described in claim 1, characterized in that, The mapping formula for the blood oxygen saturation value is: in, Blood vessel category The target range for blood oxygen saturation.

7. The method for measuring retinal oxygen saturation based on arteriovenous structure prior as described in claim 6, characterized in that, The target range is set or adaptively updated based on prior physiological data of retinal blood oxygenation, statistical results of healthy samples, equipment calibration results, or disease type.

8. The method for measuring retinal oxygen saturation based on arteriovenous structure prior as described in claim 1, characterized in that, The step of performing vascular segmentation and arteriovenous classification on the acquired fundus images to generate vascular tree masks, arterial masks, and vein masks also includes obtaining masks for regions with uncertain classification. ; The effective blood vessel pixel set Defined as: in, This indicates the exclusion of areas with uncertain classification.

9. The method for measuring retinal oxygen saturation based on arteriovenous structure prior as described in claim 1, characterized in that, The vessel segmentation and arteriovenous classification are achieved through at least one of the following: deep learning model, image processing algorithm, vessel tracking algorithm, and graph structure analysis algorithm.

10. The method for measuring retinal oxygen saturation based on arteriovenous structure prior as described in claim 1, characterized in that, The two wavelength channel images include a first wavelength channel image and a second wavelength channel image; wherein, the first wavelength channel is a reference wavelength channel that is relatively insensitive to changes in the oxygenation state of hemoglobin, and the second wavelength channel is an oxygen-sensitive wavelength channel that is sensitive to the difference in absorption between oxyhemoglobin and deoxyhemoglobin; the fundus images are simultaneously acquired by an ultra-wide-angle scanning laser fundus imaging system and the inter-channel registration is completed.

11. The method for measuring retinal oxygen saturation based on arteriovenous structure prior as described in claim 1, characterized in that, The blood oxygen saturation value is limited to the corresponding type of vascular mask to generate an arterial oxygen saturation map, a venous oxygen saturation map, and a fused retinal blood oxygen saturation distribution map.

12. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 11.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 11.

Images